DO
<OU 158478 ?m
OSMANIA UNIVERSITY LIBRARY
Call No. (,S$'$/66t{ C Accession No. 3 ^ 3
Title
This book should be returned on or before the date last marked Wow,
_;... /
CASES AND PROBLEMS IN
MARKETING
RESEARCH
By
DONALD F. BLANKERTZ
PROFESSOR OF MARKETING
WHARTON SCHOOL OF FINANCE AND COMMERCE
UNIVERSITY OF PENNSYLVANIA
ROBERT FERBER
RESEARCH ASSOCIATE PROFESSOR OF ECONOMICS
UNIVERSITY OF ILLINOIS
HUGH G. WALES
PROFESSOR OF MARKETING
UNIVERSITY OF ILLINOIS
THE RONALD PRESS COMPANY NEW YORK
Copyright, 1954, by
THE RONALD PRESS COMPANY
All Rights Reserved
The text of this publication or any part
thereof may not be reproduced in any
manner whatsoever without permission in
writing from the publisher.
Library of Congress Catalog Card Number: 54-7643
PRINTED IN THE UNITED STATES OF AMERICA
PREFACE
The present volume has been prepared to meet the need for
materials which will give practice in working with the problems
encountered in marketing research. It is increasingly recog-
nized that principles and methodology descriptively studied fail
to prepare the student sufficiently to deal effectively with mar-
keting research problems in the field. Too often the student
is lost when he goes out into the business world and meets a
problem necessitating original thinking and some use of re-
search methodology; his background has done little to prepare
him to handle problems under operating conditions. In this
book are presented cases and problems designed to be used in
conjunction with a basic textbook in the introductory course in
marketing research. The student is faced with practical situ-
ations to which he can apply research techniques as soon as he
has mastered them in theory.
The cases and problems have been carefully selected with
this objective in mind. They provide all the necessary data and
fill in the background material, both relevant and irrelevant.
The student is then asked to resolve the problem, choosing for
himself the method by which he will reach a solution. Above
all else, the aim has been to encourage the student to think for
himself to take little for granted, to sort out what is useful
from what is not, to examine all procedures critically, and to
attempt to improvise new methods where existing ones prove
inadequate.
The book is organized into nine divisions, each covering a
major aspect of the research operation. Every division is pref-
aced by a discussion of several pages which surveys the opera-
tion in question and attempts to clarify selected issues which
iii
iv PREFACE
tend to be troublesome to the student. Each of these discus-
sions places the operation in its proper setting and shows how
the cases presented in the section fit into the total process.
These discussions will be valuable for review purposes where
the book is used in a course in applied marketing research which
follows a first course devoted to principles. They will also pro-
vide the basic minimum of text materials for those who use the
book in an introductory course taught by the case method. A
selected, annotated bibliography at the end of each part gives
detailed references to further pertinent information in market-
ing research textbooks and other literature.
The cases included cover a variety of different areas, situ-
ations, and levels of difficulty. Their arrangement within a
particular section has been determined both by considerations
of logical order and by a desire to place the easier cases early
in the section. Although most of the cases refer to separate
situations, occasionally two or more cases have been used to
trace the same problem through successive operations. This
sometimes proves to be a more effective teaching method than
continually taking up new situations.
In writing this book, the authors have had the privilege of
obtaining case materials from a number of people and or-
ganizations. Among them are W. R. Spurlock, Director of
Marketing Research, Eli Lilly and Company; the Bureau of
Economic and Business Research of the University of Illinois;
the Bureau of Business Management of the same University;
Donald E. West, Director of Market Research, McCall Cor-
poration, and Harry Rosten, Research Manager, The New
York Times. There are also many others who, because of the
nature of the material supplied, prefer to remain anonymous.
To all of them, our sincere thanks and appreciation. We
should also like to acknowledge a special debt of gratitude to
Professor Lincoln Clark for his reading of the manuscript and
for the many constructive suggestions he offered.
DONALD F. BLANKERTZ
ROBERT FERBER
HUGH G. WALES
March, 1954
CONTENTS
PART I
PAGE
FORMULATING THE PROBLEM 3
1. What Sales Methods To Use? 9
2. Sales Organizations Plans 10
3. Marketing a New Product 12
4. A Local Market Study 15
5. Is the New Branch Store Profitable? .... 17
6. A Survey Proposal 22
7. A Distribution Cost Study 25
PART II
PRELIMINARY INVESTIGATIONS 29
1. Setting the Stage for a Paint Study . . . . 35
2. Preliminary Investigations for a Small Study . . 38
3. The Evolution of a Hypothesis 40
4. Internal Investigation 46
5. Use of the Consumer Clinic 52
6. Consumer Panel vs. Store Audit Research ... 56
PART III
PLANNING 63
1. A Radio Audience Survey 70
2. Proposed Consumer Market Survey .... 75
3. Use of Internal Data for Studying Magazine Mer-
chandising Effectiveness 79
vi CONTENTS
PAGE
4. Methods for Study of Installment Credit ... 87
5. Pretesting a Questionnaire 91
6. Constructing a Readership Questionnaire ... 99
7. Proposed Questionnaire for an Ice Cream Survey . 102
8. Getting a Luggage Questionnaire into Shape . . 106
9. Following Up on a Department Store Study . . 110
PART IV
SAMPLING 115
1. Random Selection of Sample Members . . . 125
2. Determining the Characteristics of the Foundry In-
dustry 127
3. A Consumer Survey of Philadelphia . . . . 130
4. Designing a Multipurpose Survey . . . . 136
5. Designing a Controlled Experiment . . . . 139
6. A Panel Operation on Consumer Durable Goods
Purchase Plans 143
7. Following Up on a Mail Survey 147
8. Testing the Validity of a Judgment Sample . . 152
PART V
COLLECTION OF DATA 159
1. Obtaining Data on the Operations of a Laundry . 165
2. Designing Forms for Gathering Internal Data . . 167
3. Mail Questionnaire Method of Survey . . . 171
4. Securing Respondent Cooperation .... 173
5. Selecting Interviewers for a Life Insurance Study . 180
6. Instruction for Interviewers 184
7. Depth Interviewing 192
8. How Can the Data Be Obtained? .... 197
CONTENTS vii
PART VI
PAGE
EDITING AND TABULATION 203
1. Editing Responses 208
2. Preparing and Preceding a Questionnaire . . 211
3. Preparation of Forms for Manual Tabulation . . 213
4. Selecting a Method of Tabulation . . . . 217
5. Cross-Classification Tabulations 223
PART VII
ANALYSIS AND INTERPRETATION OF DATA . . . . 233
1. Validation of a Consumer Survey .... 242
2. Extent of Use of Business Life Insurance ... 245
3. Who Listens to Radio Programs? . . . . 251
4. Evaluating the Effectiveness of Advertising . . 263
5. When Should Retail Stores Be Open? ... 271
6. Analysis of Price and Income Data .... 278
PART VIII
PRESENTATION AND FOLLOW-UP 285
1. Preparation of Chart Material 290
2. Preparation of a Retail Trade Area Map ... 293
3. Consumer Reaction to Self-Service Meat Marketing . 296
4. Report on Branch Store Markets and Policies . . 301
5. Follow-Up on Product Performance Studies . . 312
PART IX
ORGANIZATION FOR RESEARCH 317
1. Administering a Life Insurance Study ... 323
viii CONTENTS
PAGE
2. Setting Up a Consumer Panel Operation . . . 326
3. Organizing a Department Store Research Department 328
4. A Research Department for a Pharmaceutical Com-
pany 333
INDEX 337
CASES AND PROBLEMS IN
MARKETING RESEARCH
Part I
FORMULATING THE PROBLEM
The Nature of the Task
Marketing researchers not only gather facts and attempt to
provide solutions to problems, but also must formulate the prob-
lems themselves. Indeed, the research task in general may be
characterized as being (1) to pin point problems, (2) to suggest
possible explanations or solutions to them (these are called
hypotheses), and (3) to test these explanations or solutions (or
hypotheses). Each function may be of equal importance.
But what is a problem? It may be merely to characterize or
measure marketing phenomena, as the number of customers for
a firm, the nature and extent of certain activities of dealers, or
the sales volume by type of outlet. Or a problem may be to
identify the factors that influence the action or event (or phe-
nomenon), as purchases made at trade shows, readership of
trade journals, or the use of credit. Or a problem may be to find
the relationship if any between two or more phenomena, as
family size or income or geographical location and volume of
consumption of a product. Thus, the researcher may be dis-
covering what actually occurs, why it occurs, and what effect
it has.
Sometimes problem formulation is easy. The problems may
be simple and clear; the same, or very similar, problems may
have been studied before; or much preliminary work may have
been done by others. Often, however, the problem is new, or
complex, or scantily described, or deceptive. Then the re-
searcher faces one of his most interesting and challenging re-
sponsibilities: deciding what the problem is. Without a diag-
nosis, no sound course of treatment can be possible.
3
4 MARKETING RESEARCH
Problem formulation includes several separate, though re-
lated, aspects: a true understanding of the business situation;
rephrasing into quantitative or operable terms; specific wording
of possible explanations or solutions; looking ahead to the steps
that will follow immediately; and deciding what, if any, research
is needed.
Working Methods
Understanding the business situation which gives rise to the
problem requires perspective. Much that has been learned in
accounting, economics, marketing, and other fields is often
called into play. For example, almost any question regarding
profitability will involve considerations from the three fields
noted. Attention to the environment of the specific issuethe
policies and operations of the company or industry as well as the
facts of the market placeis needed. Thus the researcher not
only dissects what information is directly at hand when he be-
gins, but also adds all that is relevant which his knowledge, ex-
perience, and imagination can supply.
A common working method to aid problem understanding is
to write out all the possibilities which come to mind. This ex-
plodes the problem into many fields and stimulates thinking.
Each of the possibilities can then be considered as to meaning
and implications. ( Case 4 illustrates this method. )
Rephrasing problems also adds understanding as well as pro-
viding needed precision. Indeed, as soon as one attempts to
write out possibilities, he becomes aware of problems of termi-
nology. "Jobber," for example, may be synonymous with
"wholesaler." In a general discussion, this definition may ap-
pear sufficient, However, concrete expression of specific possi-
bilities may reveal that in this particular case the company's
problem pertains to drop shipping (possession without physical
handling ) . Here, the two terms are clearly not synonymous.
As another example, a client may claim interest in all aspects
of the preferences of hotel guests regarding bathroom linens.
But when this apparently simple question of "What are their
preferences?" is rephrased into operationally meaningful quec-
FORMULATING THE PROBLEM 5
tions, the list grows horrifyingly long. It soon is clear that not
all preferences are important after all.
Consider, even, the implications of so simple a term as "cus-
tomer." For a department store study this might be defined as
any person purchasing at the store, or as any person using or con-
suming goods bought at the store, or as any person buying for
credit, or as any person buying only at the one department store
(in a given period of time), or as any person buying more at
the one store than at any other, or in dozens of other ways. Yet
each term when employed would yield different figures and facts
useful in different ways to examine different aspects of patron-
age.
This is not to imply that the researcher avoids standard ter-
minology. The use of such terminology is invariably an integral
part of a study. It must be recognized, however, that in a real
sense the researcher creates facts (whether numbers, percent-
ages, or otherwise) by the way he defines his terms. Only care-
ful definition will reveal the heart of the specific problem.
Again the common working method is to write out possibilities,
to discuss them with others, and to test their usefulness.
As already suggested, writing out possible solutions or expla-
nations to problems gives scope and direction to the research.
The problem may be one of high, and rising, sales costs. The
possible solutions may include: rerouting salesmen, broadening
the product line, or improving dealer relations. Which are sup-
ported by the information available? (And, of course, there are
such related questions as: what types of rerouting, what aspects
of dealer relationships, and so forth.) More necessary than
they may seem, more difficult than they appear, these hypoth-
eses establish the nature and extent of the research that will fol-
low.
As the doctor detects some symptoms readily but must seek
for others by questioning the patient or examining him, so, too,
the researcher may diagnose some aspects at once but needs to
go further. What persons will he want to see and what ques-
tions should be asked? Where can he find clarifications for
some issues, in what records or books, and how can various pos-
sibilities be pursued? Certain formulations of the problem
6 MARKETING RESEARCH
immediately suggest company data; others obviously call for
interviews. Thus problem formulation leads to the steps which
follow. Already some plans begin to take form because they
are inherent in the situations being considered in the search for
the problem.
This preliminary formulation may reveal that there really is
no problem, or that it is far different from what was anticipated
at first. Formulation, therefore, is no mere logical or literary
job; it is an exploration of business possibilities and of the data
available.
Work procedures at this stage are governed largely by the
specific manner in which the problem has arisen and the nature
of the business situation involved. Among more or less
mechanical methods that are commonly used in this stage of
research are, as noted above, the written expression of ideas and
possible explanations, and the full discussion of them with other
members of the research staff. An essential step is to break
problems into the smallest possible pieces for clearer examina-
tion. Related ideas can then be added and different ideas can
be cross-related. Out of this expression, division, reassembly,
and selection comes the concrete (but still preliminary) formu-
lation.
Special Difficulties
The lack of well-developed processes or techniques for prob-
lem formulation, as is common to other stages of marketing
research, is a grave difficulty, especially for the beginner. For-
tunately, an active, curious mind can do much to overcome this
deficiency.
The greatest difficulty, perhaps, is that this stage of research
has no specific boundaries. Without some knowledge and
data, only crude and probably faulty formula "ions are possible.
The researcher already has talked to some members of his com-
pany or read their statements (or met with the client), so that
to some extent he is already engaged in the internal analysis or
situation analysis. He probably has moved ahead to some con-
sideration of methods of gathering data. In fact, to a consider-
able extent he must already have envisioned the end results at
FORMULATING THE PROBLEM 7
the very beginning. For these reasons not all researchers recog-
nize problem formulation as a separate and distinct phase of
marketing research, and not all of those who do place it as the
first step. However, at least for learning purposes, it seems im-
portant to segregate these functions as sharply, as clearly, and
as early as possible.
The Cases in This Part
Some of the cases which follow are intended to represent
situations as actually faced when available information is scanty,
the problem not clear, and the need for research not certain.
Consideration of these may lead mostly to determination of
logical next steps, or of sources and types of information needed.
The challenge is to analyze and use what facts are available and
to carry problem formulation as far as possible. Although the
recurring cry for more data will be sounded, it is well to learn
early that masses of data alone seldom solve problems. The
information needed is not just any information, but specific
kinds based on the investigation of that already at hand. The
cases should arouse curiosity, reveal the benefits of general
knowledge, and indicate the first order of research problems
faced.
Cases 1 through 3 are of this type. Case 1, "What Sales
Methods to Use?" reveals a common marketing problem in the
somewhat unusual setting of seed corn distribution. Case 2,
"Sales Organization Plans/' presents a similar problem of a
manufacturer of fluorescent lighting fixtures. Case 3, "Market-
ing a New Product," represents a borderline situation regarding
the need for research and the usefulness of available informa-
tion.
Other cases relate to situations in which more information
was available to the researchers and in which problem formula-
tion had been partially completed. These cases provide oppor-
tunities for careful formulation of hypotheses and rephrasings
of original statements. Case 4, "A Local Market Study," de-
scribes the origins of a broad study in which problems have been
specified in some detail by members of the luggage manufac-
turers' trade association. Case 5, "Is the New Branch Store
8 MARKETING RESEARCH
Profitable?" presents tentative formulations made by the re-
searchers and thus, as a first reading, may provide vicarious
experience in problem formulation.
The other two cases in this part represent special situations.
Case 6, "A Survey Proposal," resembles both Case 4 in breadth
and the earlier cases in scarcity of data, but raises the new prob-
lem of tailoring a dehydrated foods study to meet different
interests. Case 7, "A Distribution Cost Study/' serves to intro-
duce the problem of applying these special techniques to the
situation of a service wholesaler.
BIBLIOGRAPHY
AMOS TUCK SCHOOL. Manual on Research and Reports. New York: McGraw-
Hill Book Co., Inc., 1937. Pages 1-19 contain a discussion of library re-
search beginnings. Pages 2-4 pin point definition problems and suggest the
usefulness of preliminary outlines.
BLANKENSHIP, A. B. Consumer and Opinion Research. New York: Harper &
Bros., 1943. Pages 31-37 provide a brief discussion of research stages, with
special application to opinion research.
BRADFORD, E. S. Marketing Research. New York: McGraw-Hill Book Co.,
Inc., 1951. The first research step is characterized (pages 115-24) as a
rough-and-ready view and a feeling of the problem secured through an in-
formal investigation; and the divergent views of L. O. Brown are noted.
Problem formulation is covered in the succeeding chapter (8).
BROWN, L. O. Marketing and Distribution Research. New York: The Ronald
Press Co., 1949. Pages 325-49 summarize all the stages of marketing re-
search. On pages 350-66 is presented a formalized view of the early steps, a
view slanted particularly to broad manufacturing problems.
HEIDINGSFIELD, M. S , and BLANKENSHIP, A. B. Market and Marketing Analy-
sis. New York: Henry Holt & Co., 1947. Pages 18-24 include a discussion of
problem definition.
HOBART, D. M. Marketing Research Practice. New York: The Ronald Press
Co., 1950. Pages 3-8 contain an account of research origins and the genesis
of early projects by Parlin at the Curtis Publishing Company. Defining the
problem is noted on page 37.
LORIE, J. H., and ROBERTS, H. V. Basic Methods of Marketing Research. New
York: McGraw-Hill Book Co., Inc., 1951. Pages 67-80 provide a discussion
of the managerial and policy aspects of research origins. Pages 30-39 relate
to formulating and testing hypotheses, and provide a useful background to
problem formulation.
LUCK, D. J., and WALES, H. G. Marketing Research. New York: Prentice-Hall,
Inc., 1952. Pages 127-39 discuss the first stage in marketing research.
Part I Cases
CASE I WHAT SALES METHODS TO USE?
BACKGROUND
By the summer of 1952, the Central-Illinois Seed Company
was selling a substantial volume of seed corn to farmers of Illi-
nois, Iowa, and Indiana. Although not considered one of the
big seed corn companies of Illinois, it had been growing contin-
uously since it was founded about ten years before. The use
of modern methods of growing and processing made it possible
for the company to compete on a very favorable basis with other
concerns in the business.
The hybrid seed corn companies employed an unusual dis-
tribution plan. Farmers who were looked upon as leaders in
their communities were usually given dealer contracts. These
farmers, who generally turned out to be financially independent
of any income received from outside activities, served as dealers
and sold to other farmers during their off -hours.
The general sales manager of the Central-Illinois Seed Com-
pany, Mr. Lupton, had divided the states of Iowa, Illinois, and
Indiana into 12 districts with a district sales manager over each
territory. Each district sales manager supervised from seventy
to one hundred twenty local farmer-dealers. The district sales
manager sought to obtain productive farmer-dealers and to
stimulate them to sell as much seed corn as possible.
The selling season starts in the spring between March 1 5 and
April 1 5, when the seed corn, ordered the previous year, is de-
livered. When the corn is delivered, the farmer-dealer can fre-
quently get the farmer to commit himself for the coming year
for the largest part of his requirements. About 70 per cent of
seed corn is sold before January 1 for delivery three to four
9
10 MARKETING RESEARCH
months later. The other 30 per cent is sold between January 1
and delivery time.
In a few cases, district sales managers had obtained good sales
results from feed and seed stores. The major hybrid seed corn
companies, nonetheless, continued to use farmer-dealers. Since
the practice was firmly entrenched, there seemed little reason to
depart from it.
THE PROBLEM
The general sales manager put the problem in this form:
"What can be done to help farmer-dealers to get more product
knowledge, to make more calls, and to sell prospects properly?"
His main goal was to build up sales volume; but he was uncer-
tain as to whether to attempt this through regular channels or
through a new method of distribution.
DATA AVAILABLE
Little published information being available, information
would have to be assembled from primary sources.
QUESTIONS TO BE ANSWERED
1 . Is the sales manager's formulation of the problem susceptible
to research?
2. Is the problem to find out the relationship of the farmer-
dealers' product knowledge and their sales?
3. Is the problem to uncover the number and nature of the
calls made by farmer-dealers?
4. Before final formulation of the problem, will it be necessary
to take into consideration the reactions of farmers to sales
arguments for the use of hybrid seed corn?
5. Is the attitude of farmers toward different seeds, alternative
sources of supply, or other matters important to the problem?
CASE 2 SALES ORGANIZATION PLANS
BACKGROUND
The Great Lakes Manufacturing Company makes fluores-
cent lighting fixtures for residential, industrial, commercial, and
CASES IN FORMULATING THE PROBLEM 11
institutional installations. Because of the nature of the indus-
try, it contains essentially three types of firms, which operate at
different competitive levels. The differences are based mostly
on the marketing functions performed and distribution meth-
ods used by individual manufacturers. The first group of firms
can be described as selling to exclusive electrical wholesalers in
each area; their sales generally include the highest priced light-
ing fixtures. The second group sells to any electrical whole-
salers who maintain fixture stocks, and the third group sells not
only to any wholesaler in any quantity, but also directly to con-
tractors and consumers. The Great Lakes Manufacturing
Company has a distribution plan that puts it into the second
group.
The Great Lakes Company does not seek contracts on large
industrial installations, but secures some business from large
remodeling jobs. The major portion of its business comes from
smaller installations, including over-the-counter sales by whole-
salers.
The present sales staff includes four men who sell exclusively
for Great Lakes. In addition, fifteen manufacturers' represent-
atives are utilized. Each group attempts to reach architects and
construction engineers who must be persuaded to include Great
Lakes fixtures in their specifications. Electrical contractors are
also called upon with the same purpose in mind that of recom-
mending the installation of Great Lakes fixtures to builders. In
addition to these calls upon technical personnel, representatives
also reach electrical wholesalers who sell to each of these groups.
Still another important group, composed of lighting con-
sultants of public utility companies, is reached by sales person-
nel. These consultants are frequently asked to make recom-
mendations to customers who are planning to construct new
buildings, remodel old ones, or make repairs.
In order to reach its customers, the company has divided the
United States into eighteen sales territories. On the average,
the salesmen make about sixty calls a week in these territories.
In 1 9 5 1 they accounted for a total sales volume of approximately
$1,500,000.
12 MARKETING RESEARCH
THE PROBLEM
The company feels that it would be better for the sales organ-
ization if the manufacturers' agents were replaced by full-time
company salesmen.
In exploring the possibility, the personnel manager has put
classified ads in newspapers and trade journals. Employment
agencies have also been utilized. In spite of all this effort, the
results, in so far as obtaining qualified sales personnel is con-
cerned, have not been too satisfactory. The personnel man
reported that one agency had sent a number of men who were
below his standards and another had sent men who were above
his standards. Out of seventy-five interviews, he had found
only two men who he felt were desirable enough to be offered
a contract for employment.
A very limited amount of sales training is given the new sales-
men. The sales manager is of the opinion that the average
salesman can learn all he needs to know in one week. There-
fore, he concludes that the main job of a salesman in his com-
pany is that of knowing the products and customers of the
company. In keeping with this idea, he likes to have salesmen
make friends with their customers and prospects from the start.
Compensation of the salesmen is a combination of a com-
mission with a guarantee. For good salesmen, earnings can be
as much as $30,000 a year.
QUESTIONS TO BE ANSWERED
1. What is the key problem and how can it be formulated?
2. Is research needed?
CASE 3 MARKETING A NEW PRODUCT
BACKGROUND
In April, 1950, a concern that had been making devices used
by doctors in the treatment of certain physical handicaps de-
cided to branch out its manufacturing operations. The decision
grew out of a desire not only to increase the profitability of the
CASES IN FORMULATING THE PROBLEM 13
business, but also to utilize one of the wasted by-products of its
principal manufacturing operation. <The waste material was
scrap leather which could be cut into small strips about % inch
in width and 3 to 4 inches in length.
The company used heavy packing cartons, made of straw-
board, for shipping the product to customers. By chance, it
occurred to the owner of the business that these packing cartons
could be impregnated with wax, thereby producing a waterproof
container. After some experiment, he developed an impreg-
nated container that was insulated with fiber glass and that
could be used to keep foods either hot or cold over long periods
of time. Handles for the container were made from the scrap
leather. Detailed sketches of this insulated box, called "Pic-
Pac," are given below.
INSIDE
LID-
PRODI
NAft
r-'-^ 1
I f '"" '"'" '" 1UJ 1 '
1
1
1
I S
u. J
I FIBRE GLASS
-v INSULATION y /
^ V ' J
T r^
1 PIC-PAC
i /
i 4 C J
J \
JCT-' INSIDE-*'
/IE CARTON
HANDLE HOLES
-FOR LIFTING
FIBRE-GLASS
INSULATION
OUTSIDE
CARTON
The major product line of the company, the medical de-
vices, was distributed directly to drugstores and medical supply
houses, and by mail order directly to users. It therefore seemed
that some of these outlets might be employed in the distribu-
tion of the Pic-Pac to consumers. The reasoning of the owner
of the company was that this insulated carton could be used by
consumers to pack picnic lunches (the name of the product sug-
14 MARKETING RESEARCH
gests this) and as a container for cooling bottled or canned
beverages. Hence, drugstores might be used as middlemen for
distributing this product. In addition, since the carton could
be shipped in knockdown form and assembled easily by the con-
sumer, it seemed that a profitable mail-order business might be
developed.
The retail price of the item was set arbitrarily at $1 .98. This
price was justified on the basis of the carton's durability and con-
venience in cooling and preserving picnic foods, beverages, fruit,
and game.
THE PROBLEM
The owner of the business presented his problem to an adver-
tising agency for consideration. He wanted to know what chan-
nels and, particularly, what advertising media would be most
suitable for marketing his Pic-Pac. It was his opinion that mail
order would offer the most economical means of putting his
product on the market, and he, therefore, thought that the ad-
vertising agency might help him market Pic-Pac through this
channel of distribution.
AVAILABLE DATA
No published data were available on the sale of this or similar
products. At that time, several insulated containers of a similar
nature were on the market, manufactured by metal products
companies, selling from $6.95 to $1 5.00. All of them made use
of light-weight metals instead of strawboard as the container
material. They were marketed primarily through retail outlets
department stores, drugstores, and sporting-goods stores, with
some supplementary distribution through mail-order sales.
ACTION NEEDED
1 . How can available information be used?
2. For what specific problems is research needed and how can
these problems be formulated?
CASES IN FORMULATING THE PROBLEM 15
CASE 4 A LOCAL MARKET STUDY
BACKGROUND
The Luggage Manufacturers' Association is a trade associ-
ation whose primary efforts in marketing research have been
directed toward gathering production and sales data from mem-
bers and seeking standardization in the field. A slump in in-
dustry sales plus the urging of the executive secretary has led
to the consideration of a market survey.
THE PROBLEM
No study of consumer purchase or use of luggage has ever
been made. The suggestion that a local market study would
be most economical for the Association has met with favor.
Finally, a research group in a large eastern city has been ap-
proached. The Association has requested that the study be a
broad one, including such areas as consumer use and purchase
habits, consumer preferences for prices and styles, and con-
sumer buying intentions. The study is expected to prove useful
to individual manufacturers in designing, promoting, distrib-
uting, and selling their lines of luggage.
QUESTIONS RECOMMENDED
The researchers, during negotiations, had requested that the
directors of the Association, or some representative group of
members, be asked to state specifically the information they
desired. The executive secretary thereafter canvassed members
and obtained this list of questions:
1. What does the consumer look for when purchasing lug-
gage?
2. Does the consumer prefer leather luggage to striped, canvas-
covered luggage?
3. Does the consumer look for styling in luggage?
4. Do female consumers prefer bright colors within and on the
outside of luggage?
5. What kind of luggage does the male consumer prefer?
16 MARKETING RESEARCH
6. How often does the consumer use his luggage?
7. How old is his present luggage?
8. How often does the consumer travel?
9. What care does he give his luggage when not in use?
10. Where does the consumer store his luggage?
1 1 . Did he buy his luggage or receive it as a gift?
12. Does the consumer purchase luggage for gifts?
13. Does the consumer know whether the luggage he now pos-
sesses differs in style and convenience from that which is at
present being made and sold?
14. Is the consumer in the market for new luggage? If not,
how soon will he be in the market?
15. If the consumer is not considering the purchase of new
luggage now, would he buy if he were aware that the lug-
gage made today has better style and is more serviceable?
16. When purchasing luggage, does the consumer have a def-
inite price range in mind?
17. Does the price determine the luggage which is eventually
purchased, regardless of its type or serviceability?
INFORMATION AVAILABLE
Luggage is a highly variable product. Many materials includ-
ing various leathers, plastics, aluminum, various cloths (such as
canvas, duck, cotton, and linen), and paper are used alone or to
cover plywood boxes. Luggage may be lined, as with linen, cot-
ton, or silk, in various designs and colors, or it may be unlined.
Colored luggage has been increasing in popularity. Prices vary
from $1 to well over $100. Types range all the way from hat
and shoe boxes to large pullman cases with trays. Men's lug-
gage includes overnight bags, fortnighters, one-suiters, two-suit-
ers, folding suit bags, valises, suitcases, sports-style bags, foot
lockers, and even brief cases. Women's luggage includes over-
night cases, train cases, bus cases, hat and shoe boxes, wardrobe
hanger cases, and vanity boxes. Moreover, these names are not
always used in the same way, and nearly every type comes in
varying sizes. Many trade names are used. Such terms as
"Gladstone" and "airplane luggage" have various meanings to
CASES IN FORMULATING THE PROBLEM 17
the industry, to the retailers, and to the public. Matched sets
have been given considerable promotion in recent years.
Distribution is mainly through wholesalers, although many
manufacturers sell directly to retailers. Outlets include luggage
and leather goods stores, jewelry stores, department stores,
men's wear specialty shops, variety stores, sporting-goods stores,
and pawn shops. During the war a large volume went through
military outlets.
ACTION NEEDED
1. What is the problem here? How can the questions fur-
nished and the information on hand be utilized to formulate
the problem?
2. What really is meant by each of the questions supplied?
How important is each?
3. In what ways would answers to the questions be useful?
4. Have any important questions of consumer use and prefer-
ence or purchasing behavior been omitted?
5. How can the questions be grouped or classified?
CASE 5 IS THE NEW BRANCH STORE
PROFITABLE?
BACKGROUND
The Boris Department Store is a large department store in
an Eastern metropolitan area. Located in the central shopping
district, the store does about $50 million annual volume. Al-
most half of this business comes from residents of the city. The
balance is spread over fifteen surrounding counties, except for
a few hundred thousand dollars of sales widely scattered over
the rest of the country and the world.
The store has been approached by the landlord of a property
in a large secondary shopping center. This shopping center,
known as the Ransom shopping district, is located in a heavily
populated suburban area which has been growing steadily.
The landlord was dissatisfied with the low volume of business
done on his property by the branch store of another, but con-
18 MARKETING RESEARCH
siderably smaller, department store in the city; the lease called
for rental payments consisting of a guaranteed minimum plus
an additional sum determined by gross sales. The landlord
offered the lease to the Boris Department Store.
Before purchasing the lease the store had its research staff
make an estimate of probable volume of sales. A research
agency also had been hired to make a consumer survey in the
Ransom district trading area. This study analyzed the attitudes
of consumers toward Ransom district and city stores, their pref-
erences for places to shop, their use of transportation facilities,
their ages, and many other aspects of their buying habits. The
study was primarily intended to aid in determining what type of
operations would be best suited to the market.
The estimate made was that over $4 million in sales could be
expected in the first year and more than $6 million per year
after five years. The previous tenant was doing approximately
$1,500,000 annually.
The lease was purchased. After modernizing and air con-
ditioning the sales floors, the store opened in October.
THE PROBLEM
At the end of the first year of branch operation, net sales were
$4,500,000, or three times the volume previously done on the
site. Although encouraged by results, the management of the
Boris Department Store felt uncertain of the success of the new
branch. Was it really profitable? An answer to this question
would not only affect plans for the branch, but would also help
to determine policy on the possible opening of branches in other
suburbs or nearby cities. A research group was asked to find the
answer.
In early conversations with store officials, several issues were
explored. The most obvious problem seemed to be the ques-
tion of allocating expenses. The branch was not organized as
a completely separate unit. Buyers in the parent store also did
the buying for the branch. Merchandise was received, checked,
and marked in the downtown, parent store, being taken by truck
to the branch store when needed there for forward stock. All
credit and accounting operations were centralized in the parent
CASES IN FORMULATING THE PROBLEM 19
store, as were many other operations. Allocation of indirect
and joint costs, therefore, was a problem. Store officials, how-
ever, seemed little concerned with profitability in the strict
accounting sense. Despite considerable use of sales volume as
an allocation base and the possible use of better bases, it was felt
that any reasonable division of expenses would demonstrate un-
questionably that the branch was profitable.
Neither did store management seem concerned with the effi-
ciency of branch operations. There appeared to be some
divergence of views about the degree of adaptation of the
branch to its particular market. Except for a few different
brands in a few departments and differences in price lines, the
branch carried the same merchandise as similar departments
in the parent store. Some officials believed there should be
more differences and that buyers should be forced to distinguish
between the markets. In general, however, efficiency of opera-
tions was not considered a major problem. Routine data from
stock control records and accounting and statistical data were
believed sufficient to detect important weaknesses in operations.
The principal worry about branch profitability apparently
stemmed from the same fear which had deterred branch expan-
sion by many stores in the years following World War II, when
construction again was possible and when the movement of
population to the suburbs was well established. This was the
fear that branch store business was stolen from parent store
business.
The smaller trading area of the branch store, of course, was
encompassed by the trading area of the parent store. This
meant that Boris customers now could buy a wide range of al-
most identical merchandise at two different locations. Had
the convenience of shopping outside of the congested city-
center area caused customers to shift trade to the new location?
It was felt that if most branch sales were the result of a mere
shifting of business from one site to another, then the branch
was unprofitable. The new expenses incurred in the branch,
it was believed, had to be justified by new business, an addition
to the total volume done by the Boris Company.
Much suspicion existed that a transfer or switching of busi-
20 MARKETING RESEARCH
ness actually had occurred. Parent store sales had matched, in
fact slightly exceeded, the increase in department store business
regionally as shown by the Federal Reserve Board Index of
Department Store Sales. This increase, however, was much less
than anticipated and far below the rate of growth enjoyed by
the parent store in recent years. Perhaps this was due to the
siphoning off of downtown trade by the new branch store.
Insofar as switching of trade was the problem, it seemed im-
mediately to lead to the problems of buying habits, customer
preferences, customer characteristics, and the competitive situ-
ation. The origins of branch business, in short, seemed to in-
volve many market problems. Had the new branch changed
buying habits in relation to purchases downtown or in nearby
cities or secondary shopping centers other than the Ransom
area? How about travel patterns, types of customers affected,
competitive changes, and so on? An investigation of such ques-
tions could illuminate the problems of why and how the branch
was profitable or unprofitable, especially since many forces were
operating outside of parent and branch store operations and
without direct reflection in store records.
A special problem was the effect of having taken over a site
on which a branch store had operated. Had this been a major
factor in developing sales? Few, if any, opportunities to secure
such sites would exist in the future should more branch open-
ings be considered.
QUESTIONS TO BE ANSWERED
These broad, preliminary considerations suggest several pos-
sible problems and numerous specific questions for which re-
search might seek answers. These specific questions include:
1 . How much new business did the branch develop?
2. How many new customers were attracted?
3. How much business was "stolen" from the parent store?
4. How much new business, if any, would branch operations
have to add to make the new branch profitable?
5. Were inactive accounts of the parent store reactivated by
the branch?
CASES IN FORMULATING THE PROBLEM 21
6. Has the branch been "feeding back" business to the parent?
7. How many Ransom area customers deal only with the
branch? Only with the parent store?
8. How many customers deal with both stores, and how do
they divide their purchases?
9. What kinds of customers has the branch attracted?
10. Have some customers of the branch stopped buying from
the parent store or from all downtown department stores?
11. How many shopping trips do customers make? Where?
By what means of transportation?
12. How have specific competitors in the Ransom shopping
district, downtown, and in other shopping centers fared?
DATA AVAILABLE
No published results of any similar study were available;
neither could any noncompetitive store be found which had
made such a study.
Two research reports on the Ransom trading area were avail-
able. As already noted, one had been made for the Boris De-
partment Store prior to purchasing the branch lease. It showed
ratings regarding women's favorite downtown department
stores and their reasons for their preferences; their last purchases
of fourteen categories of goods; how much buying was done in
specified areas; purchases in the Ransom shopping district,
time, type, preferences, stores, means of transportation, and
so on. The other was a basic trading area study delimiting the
area for wartime purchasing and peacetime purchasing, reveal-
ing shopping habits, and estimating present and future sales
potentials.
Store records on credit purchases in the form of microfilm
copies of monthly statements provided basic data for measuring
the place (branch or parent) of purchases, type, and amount.
Sales slips data were so filed as to render their use for enumerat-
ing a sample of customers very difficult. The dates when ac-
counts were opened and all changes in residence were available
from store records. It was estimated that about one sixth of the
total parent store sales volume came from Ransom area custom-
ers. Nearly two thirds of all business was on a charge basis, but,
22 MARKETING RESEARCH
when the cash purchases of credit customers were considered,
more than 90 per cent of all business came from persons with
credit accounts at the Boris Company.
An analysis of 11,555 sales slips for purchases made during
one week in May showed the geographical origin of branch
customers broken down into nine major areas of the Ransom
trading area and twenty subareas.
All regular stock control data, sales data, and other routine
statistical and accounting data were available.
A cursory survey of published sources showed the following:
GOLOFF, J. M. "What Customers Think about Branch Stores/' Journal of Re-
tailing, October, 1948.
"Department Stores Hurrying to Suburbs/' Business Week, October 4, 1947, pp.
24-26.
CONVERSE, P. D. "Factors Determining Retail Shopping Preferences/' Dun's
Review, August, 1947.
A series of articles by E. B. WEISS appearing in the Department Store Economist,
March through July, 1949; other articles in Printers' Ink, Architectural Rec-
ord, and Architectural Forum, most of them about individual stores.
ACTION NEEDED
At this preliminary stage of the study, the researchers wish to
formulate the problem, or problems, tentatively but specifically.
1. Which aspect (or aspects) of the question of branch store
profitability is central to the problem?
2. How can the basic problem, as it now appears, be defined?
3. What are the specific questions involved?
4. How should the study proceed from this point?
CASE 6 A SURVEY PROPOSAL
BACKGROUND
National Publishers, Inc., an organization of several mag-
azine publishers, following the lead of other publishers and
publishing groups, was considering a series of surveys which
would interest its readers, industry and trade members, and ad-
vertisers. One survey suggested pertained to dehydrated food
products. During World War II there had been much public
CASES IN FORMULATING THE PROBLEM 23
interest in the military and foreign-aid uses of dehydrated foods
and in the potential civilian uses.
THE PROPOSAL
Late in 1946 a letter was sent by National Publishers to the
new Dehydrated Foods Council, an organization which at that
time represented processors and distributors of dehydrated
foods. This letter outlined plans for the projected series of
surveys and tentative ideas for a project on dehydrated foods.
Suggestions were made that a survey of housewives might be
of greatest general interest and benefit. The most recent of the
annual consumer surveys and pantry inventories by newspapers
had covered only dry soup mixes. The percentage of house-
wives stocking dry soup mixes was 33.3 in the Chicago Sun-
Times Pantry Poll, and 41.8 in the Omaha World Herald poll.
Affirmative answers to the question, "Do you buy dry mixes for
making soup?" were, in the Milwaukee Journal, 56.7 per cent,
and in the Indianapolis Star, 23.6 per cent.
The suggestion was made that the present survey cover the
full range of dehydrated products, the types of families using
each, frequency of use, methods of purchase, and attitudes to-
ward these products. Of interest in the latter connection were
the likes and dislikes that family members who had served in
the armed forces had acquired through experience with dehy-
drated foods and how these attitudes might affect market devel-
opment. Personal interviewing seemed best, though pantry
counts and magazine surveys might also be used.
Also suggested was the possibility of a related survey of insti-
tutional users and of retail outlets handling dehydrated foods.
In addition, of course, any study would include a discussion of
industry growth, processes, plans, and other matters. The views
and cooperation of the Council were solicited.
THE REPLY
The reply received from the managing director of the Council
expressed much interest in a survey such as was proposed and
offered assistance. Certain qualifications were given, however,
particularly to the effect that the Council felt it would not be
24 MARKETING RESEARCH
good industry policy to have the results published at this time
if they indicated a lack of popularity for dehydrated foods.
During the war years, it was stated, the great volume of dehy-
drated products consisted of potatoes, carrots, sweet potatoes,
cabbages, beets, and onions. Only a small portion of these
products found their way to the consumer's pantry, the produc-
tion being primarily for the armed forces and for foreign aid.
However, in addition to the soup mixes and other products men-
tioned, it was certain that any survey which covered all dehy-
drated foods would disclose the use by consumers of dehydrated
flakes made from celery, parsley, mint, green peppers, onions,
and mixed vegetables.
The industry, it was said, increased from 21 manufacturers
before World War II to 141 at the close of the war. However,
many of these producers were not familiar with the technique
of dehydration, with the result that some of the products made
were below the standards which had been established by the
older producers.
It was to be expected, the Council believed, that the great
tradition among soldiers to complain about food would extend
to a whole line of products, including, unfortunately, dehy-
drated foods.
Regardless of the outcome of any survey, the Council said
that its members would be vitally interested in the results. The
hope was expressed that the problems raised might be discussed
together before any final plans were made.
ACTION NEEDED
1. Should preparations for a survey of dehydrated foods be
abandoned or should they be continued, in view of the reply
of the Dehydrated Foods Council?
2. If continued, how could problems be formulated which
would serve the interests of all groups involved?
3. What appropriate hypotheses concerning consumer uses
and opinions could be set up at this preliminary stage to serve
as a framework for the survey?
CASES IN FORMULATING THE PROBLEM 25
CASE 7 A DISTRIBUTION COST STUDY
4
BACKGROUND
A. R. Merrows, a service wholesaler of grocery lines, operated
a number of branch houses and sales offices throughout the
country. Although the company operated its own cannery, the
bulk of its supplies was purchased from independent canneries,
company-owned labels being attached to these products. Of
nine company brands, seven were applied to a small number of
items, and two, "Made-rite" and "Marvo," were used on wide
lines of canned goods and dried fruits. Approximately 60 per
cent of the dollar sales of the company as a whole were made to
local chains, 1 5 per cent to schools, hospitals, and other institu-
tions, 1 5 per cent to independent grocers and other retailers,
and 10 per cent to jobbers.
THE PROBLEM
One of the branches was located in the wholesale district of
an Eastern metropolis. In this Eastern branch, during the early
'forties, the question arose as to the desirability of making a dis-
tribution cost analysis. The initial stimulus had come from a
speech the company's general manager had made during a
branch office sales convention, in which he said:
In a survey carried out at our main office, we found that 75 per cent
of our sales were being contracted with 25 per cent of our accounts. The
survey revealed further that unless a customer's order averaged $24 or
more, the firm lost money on it. We do not begin to make a substantial
profit on an order until the order averages $40 or more. It is possible
that you people are faced with a similar situation in this market.
As a result of this sales convention, the management of the
branch office adopted two remedial measures, one aimed at the
sales force and the other at customers. The branch sales man-
ager conferred at length with each salesman to convince him of
the unprofitability of selling to customers buying less than $200
annually. Most salesmen proved difficult to convince, assert-
ing that many of their small accounts actually were profitable
26 MARKETING RESEARCH
now or could be built into profitable accounts if they were al-
lowed to continue selling them. (Although evidence was not
complete, it seemed later that the branch had been unsuccessful
in raising any large number of small accounts into the profitable
volume zone.)
In those cases in which salesmen had been convinced by the
arguments presented, they had proceeded to drop a large num-
ber of accounts which were, as defined by the branch, unprofit-
able. Obvious, however, was the tendency of these salesmen
to acquire a large number of new customers in the following
months, nearly all of these customers exhibiting characteristics
similar to the ones dropped.
The remedial measure directed at the retail trade was an at-
tempt to set minimum limits to the size of the order. The sales
force was informed that orders weighing less than 250 pounds
would not be delivered. An average case of goods weighed 42
pounds, so this meant that orders had to consist of at least six
cases. The price to the retailer of a case averaged $4. In effect,
therefore, the branch was endorsing the general manager's re-
port that the company lost money on orders under $24.
The difficulty of convincing the salesmen, the uncertainty
regarding the actual cost situation for the branch, and the less
than phenomenal results achieved by the rigorous methods
used, all served to raise the question of the desirability of a dis-
tribution cost study, and, if needed, the type of study that would
be most useful.
NATURE OF BRANCH OPERATIONS
The branch operated from a multifloor warehouse, part of
one floor being devoted to office space. On the ground floor,
21,400 square feet were used for storage and 2,150 square feet
for shipping and receiving platforms. Incoming goods were
hauled from nearby city docks to the branch house by inde-
pendent trucking agencies. More than half of all merchandise
purchased was transported by steamship. The branch had a
private rail siding and was within half a mile of a large freight
station. Deliveries to customers were made for the branch by
an independent trucking agency.
CASES IN FORMULATING THE PROBLEM 27
On the second floor 5,520 square feet were devoted to office
space, 41 per cent for the sales department, 25 per cent for the
credit department, 24 per cent for an'order-routing section, and
10 per cent for administrative offices. Goods were moved in-
ternally by freight elevators and a metal chute.
Branch sales of $1,500,000 annually were distributed much as
were company sales except that institutional sales were slightly
higher and sales to jobbers slightly lower. The sales depart-
ment consisted of a sales manager, three sales supervisors, and
thirty-five salesmen. Salesmen received a basic commission of
4 per cent and fixed expense accounts, which approximated 1
per cent of sales. Each owned an automobile to cover the broad
territories assigned. Within the limits of their territories, the
salesmen might sell any customer they could obtain.
Salesmen received mimeographed bulletins from the branch
offices each day covering all pertinent market data and develop-
ments. Sales meetings were held twice a month and sales con-
ventions twice a year. Executives from the main offices were
principal speakers at these conventions.
Advertising media used included newspapers, radio, and bill-
boards. Free consumer deals of the premium type were prac-
ticed, the services of outside agencies being used. Cooperative
retailer advertising, advertising allowances, and demonstrators'
services also were provided, subject to the restrictions of the
Robinson-Patman Act.
Credit terms were the same as those generally offered in the
trade: 1 per cent, 10 days; 30 days, net. These time limits were
strictly observed by the branch.
ACTION NEEDED
1. What type of distribution cost analysis might the branch
consider making? Why?
2. What functional expense classifications might be antic-
ipated?
Part II
PRELIMINARY INVESTIGATIONS
The Nature of the Task
Following the original formulation of the problem, but pre-
ceding final plans for the research study, preliminary investiga-
tions are made of the company, the market, and secondary data.
These preliminary investigations are, partly, explorations for
new hypotheses or new ideas and, partly, tests of the concep-
tions of the problem already formed. They also inform the
researcher as to the help that may be given, or the warnings pro-
vided, by previous research and by facts already available.
The extent and the exact nature of this preliminary research
varies. The cautious researcher, however, will not proceed with
the final plans until he has tested his understanding of the prob-
lem both in the market place and among those who will utilize
his findings.
Working Methods
Depending on the knowledge he has gained, the way in which
the project has developed, the contacts he has established, and
his personal preferences, the researcher usually will select one
of the following three types of investigation with which to
begin. He may utilize his staff to conduct all three more or
less concurrently.
1. Interned Investigation, or Situation Analysis. Having
necessarily consulted with company officials, the researcher may
already have progressed from problem formulation to situation
analysis. Often, indeed, it takes much conversation with exec-
utives and other company personnel, plus considerable inspec-
29
30 MARKETING RESEARCH
tion of company records, merely to arrive at a clear expression
of company policies.
The internal investigation is insurance against missing im-
portant possible solutions other than those first formulated.
Conversations with the persons concerned, probing for their
ideas and assumptions, and inspection of company documents
and records are the techniques. These methods also help fur-
ther to sharpen the issues so that a more accurate study can be
made and a more practical application of results obtained.
The situation analysis also should uncover the data which
exist within the company and the form in which they exist.
Many times the most important and most accurate data are
buried in accounting, credit, delivery, inventory, or other rec-
ords. Proper segregation and analysis of these data may provide
the quickest, cheapest, and most accurate findings.
In this work the researcher must be on guard against the bias
of special interests and of preconceptions.
2. Investigation of Secondary Data. Finding useful informa-
tion or guidance in published materials is increasingly possible.
Moreover, research which builds on previous studies usually is
more productive than research utilizing noncomparable tech-
niques or producing noncomparable data. Past studies may
provide usable facts, helpful theories, or special techniques
adaptable to the problem at hand.
The information may be secured from various libraries and
government agencies, from trade associations, and from special
agencies or private companies. Published material will include
books, articles, news columns, and reports of various kinds, all
of these usually located through card catalog files, prepared in-
dexes, and references uncovered as the search progresses. Al-
most every marketing research textbook covers the sources of
information used and the methods employed or gives adequate
reference to them.
At this stage of research, it should be emphasized, extensive
rather than intensive coverage of sources is important. Data
gathering, as such, whether of primary or secondary data, comes
later. Although the researcher will not ignore figures and find-
PRELIMINARY INVESTIGATIONS 31
ings that solve the problem or contribute to its solution, nor
even postpone gathering what he finds, locating facts to solve
the problem is not his real objective at this stage. Rather, he
seeks to guard against useless repetition, to explore new possi-
bilities, and to sharpen his tools.
3. External, Exploratory, or Informal Investigations. Here,
for the first time, research moves physically out of the office into
the market place. The prospective respondents, those who will
be called upon to furnish the basic data of the final, formal
study, usually are interviewed at this time as well as users and
distributors of the products.
One purpose of such interviews is to test assumptions about
the market, or problem, and the hypotheses developed by those
doing the study or for whom the study is being done. Another
purpose is exploration, the uncovering of ideas which cannot be
visualized just by studying the company or published materials.
Still another purpose is realism, the development of a "feel" of
the market, the actual behavior of customers or others, and the
specific terms in which consumers, dealers, and others think
about situations. (Realism is particularly useful if a question-
naire is later formulated.) External investigations may also be
helpful in the selection of specific techniques of data gathering.
These techniques include individual personal interviews of
various kinds, group interviews (or clinics), mail questionnaire
surveys, telephone calls, and observation (as of purchasers or
salesmen in stores). Choice depends upon the number and
geographical distribution of the respondents desired, the nature
of the problem, and time, cost, and personal preference or ex-
perience.
The most common technique is interviewing which to some
extent utilizes probing, or so-called "depth interviewing" meth-
ods. Seldom is a fixed set of questions employed. In general,
the interviewer seeks to get the person talking freely on the
broad subject, being content to steer the monologue back to
relevant areas whenever the respondent wanders too far afield.
Interviews which result in simple yes or no answers rarely are of
any use. Reports are detailed, including verbatim records of
32 MARKETING RESEARCH
statements of particular interest, and may contain not only
identification but a digest of results and an evaluation.
The researcher himself, whenever possible, should conduct
some of these interviews. In sifting the reports of others he
looks both for the sparkle of a new, unique idea and for general
trends. Statistical summaries seldom are valid, however, be-
cause of sampling methods used. Time and again in this proc-
ess the suggestion of one woman has provided a basic hypothesis
or even, eventually, a new product.
Special Difficulties
So much might be done and so much information usually
exists that it is possible to become engulfed in data and lose
sight of central objectives among a mass of details. Contrarily,
there is the tendency to skim over data and overlook good mate-
rials. One protection against such dangers is not to begin until
the problem is sufficiently formulated that a working outline
or a working method, scope and direction are provided. An-
other protection is the use of so-called "standard outlines/' *
The keynote of such an outline is the use of an orderly process
for covering, assembling, and summarizing what information
is available and what is missing.
A related difficulty is knowing when to end the process. For
the external investigation, the end usually occurs when further
interviews fail to develop new ideas or important nuances. For
secondary data the investigation should stop short of true data
gathering and after the risk of missing anything important has
passed a point to be chosen only by judgment. The internal
investigation also should stop short of data gathering before the
researcher makes a nuisance of himself, and after a probing of
all important areas.
1 L. O. Brown, op. cit., p. 354, suggests an outline based on "six major mar-
keting factors which control the sales and profits of a company, i.e., the prod-
uct; the company, industry, and competition; the market; channels of
distribution; sales organization; and advertising." E. S. Bradford, op. cit., pp.
130-31, presents a detailed list with many subtopics and individual questions.
The Check List for the Introduction of New Consumer Products, Economic
Series No. 41, Department of Commerce (Washington, D. C., 1946), exem-
plifies a standard outline for a specific type of research problem.
PRELIMINARY INVESTIGATIONS 33
The Case Problems Presented
Case I, "Setting the Stage for a Paint Study/' is a transitional
case. Ample material is provided for an assignment and session
on problem formulation, although the intent is to raise doubts
regarding continuation of the study and, if it is carried on, how
best to utilize the findings.
Case 2, "Preliminary Investigations for a Small Study," pre-
sents a study at the inception of the preliminary investigation
level and without full problem formulation. Plans for internal
and external investigations are needed, attention being given
to the limited funds available. A special challenge may be the
possibility of securing final answers direct from consulting sec-
ondary sources.
Case 3, "The Evolution of a Hypothesis/' traces the develop-
ment of a promotional study with regard to the influence of
women on sales of a men's-wear store. The case focuses on
the reports resulting from the informal investigation, interviews
with consumers, and their bearing on the study. An oppor-
tunity is provided for students to hold their own interviews on
this problem for comparison with the reports provided and for
evaluation of informal interviewing methods and problems.
Case 4, "Internal Investigation/' shows how a consultant
went about an internal investigation for a publisher and how
the findings affected the scope and nature of his study.
Case 5, "Use of the Consumer Clinic," discusses a technique
of preliminary investigation which is not, as yet, widely used.
The study itself was experimental, and imagination is doubly
needed.
Case 6, "Consumer Panel vs. Store Audit Research/' again is
transitional. Other cases also have pushed the research into
the area of final planning. This one discusses in some detail
the techniques employed in data gathering. The case is in-
cluded here, however, because it represents both a research pro-
posal and a history of how a company attempted to reach a
decision, based on its own informal investigation of the situ-
ation.
34 MARKETING RESEARCH
BIBLIOGRAPHY
AMERICAN MARKETING ASSOCIATION. The Technique of Marketing Research.
New York: McGraw-Hill Book Co., Inc., 1937. Chapter 1 indicates some
of the ways in which a preliminary survey may be conducted.
BLANKENSHIP, A. B. Consumer and Opinion Research. New York: Harper &
Bros., 1943. Chapter 3 contains a discussion of early research steps, includ-
ing (pp. 31-34) the situation analysis and (pp. 34-36) the informal investi-
gation.
BRADFORD, E. S. Marketing Research. New York: McGraw-Hill Book Co.,
Inc , 1951. The author does not clearly separate either problem definition
or planning from preliminary investigations. Chapter 7 and pp. 126-33 of
Chapter 8 include useful examples of schedules for the preliminary investiga-
tion of specific problems.
BROWN, L. O. Marketing and Distribution Research. New York: The Ronald
Press Co., 1949. Pages 350-54 provide a fine summary statement of early
research steps and make a clear distinction between internal and external
analysis; pp. 354-66 give an extended treatment of broad problems of manu-
facturing encountered in internal analysis. Chapter 18 provides both an ex-
cellent summary of purposes and thirteen examples of interview reports.
HEIDINGSFIELD, M. S., and BLANKENSHIP, A. B. Market and Marketing
Analysis. New York: Henry Holt & Co., 1947. Chapters 4 and 5 provide
a highly satisfactory summary of internal investigations useful in conducting
the preliminary step of a broad study or for data gathering.
LORIE, J. H., and ROBERTS, H. V. Basic Methods of Marketing Research. New
York: McGraw-Hill Book Co., Inc., 1951. Chapters 5 and 19 offer good
material on explorations and preliminary problems encountered in coordi-
nating research with the needs of management.
LUCK, D. J., and WALES, H. G. Marketing Research. New York: Prentice-
Hall, Inc., 1952. Chapter 7 provides good treatment of external preliminary
investigations together with much illustrative material.
Part II Cases
CASE I SETTING THE STAGE FOR A
PAINT STUDY
BACKGROUND
As is not unusual, the major part of a course in marketing
research in a certain college consisted of carrying out a full-scale
research project. Numerous considerations entered into the
selection of the project from among several proposed.
The project, it was felt, should be practical. The results
should be useful and profitable, not merely interesting. Such
a project usually depends upon a cooperative client. It was also
desirable that opportunity be provided to prepare a published
report, if warranted, and that there be permission to release all
but the most confidential data. Problems were preferred that
were exploratory and experimental, rather than simply cut-and-
dried. As long as such projects involved relatively little expense,
the students could afford to risk taking new research paths and
thereby provide themselves with much additional experience.
Although full consideration was given to specific proposals
made, freedom of choice had to be left to the class regarding
such matters as the scope of the study, emphasis, and methods
used. Interest in the problem, its feasibility for class study, and
other considerations also entered into the choice. A proposal
made by the secretary of the sales division of the Paint Manu-
facturers' Association seemed to meet these criteria better than
the projects suggested by other associations and firms at that
time.
35
36 MARKETING RESEARCH
THE PROBLEM
The paint project was outlined by the secretary as follows:
So far as marketing problems in our industry are concerned, one of
the greatest problems we have always had is to determine why the
American public is using only about 10 to 25 per cent of the possible
total of the products of our industry for which there is a definite market
and need. In other words, paint for only about 10 to 25 per cent of the
surfaces in this country which should be protected and decorated with
the products of our industry is being sold. In connection with this
problem of selling the public on the advantages of and necessity for the
use of finishing products, we would like to know the answers to the fol-
lowing as examples:
1. The color preferences of the public both for inside and outside
painting.
2. Preferences for types of paint, such as oil paint, water paint,
enamels, and the reasons for such preferences.
3. Average size and extent of paint jobs needed.
4. Who is responsible for the purchase of paint products men or
women?
5. The present attitude of the public toward the quality of paint.
6. The frequency of repainting interior exterior.
7. The relative importance of paint to other things for which the
consumer spends money, such as automobiles, household equip-
ment, vacations, and so on.
8. The attitude of the public toward installment buying of paint-
ing jobs.
9. The reason why the consumer does not paint more often.
10. From what source does the public secure its color ideas for in-
terior painting of homes?
1 1 . How much painting is done by the home owner, by the painting
contractor?
12. What prompts the purchase of paint in most instances: adver-
tisements, solicitation by painting contractors or paint dealers,
and so forth.
No other factual data were given.
TENTATIVE DEFINITION OF PROBLEM
A decision was made to hold other possible projects in abey-
ance and to proceed with the paint study. At first blush the
CASES IN PRELIMINARY INVESTIGATIONS 37
most challenging aspect of the problem and the one most im-
portant potentially seemed to be the emphasis on securing mer-
chandising results which might aid in advertising, promoting,
and selling paint of all kinds for home use. This might require
the collection of data from both consumers and retailers.
Such a merchandising problem, it appeared, would involve
consideration of:
1. Quantities used (in quarts, square feet, or dollars) by types
(and colors?) of paint related to family income, occupation,
size and type of dwelling, home ownership, and so on.
2. Persons doing the painting: family members or others.
3. Influence of various family members in selecting types and
colors of paints and in making the actual purchase.
4. Place and manner of purchase of paints and influence of
advertising, displays, and so forth.
5. Paint type and brand preference (and use) and reasons for
preferences, such as cost, appearance, ease of application,
ease in cleaning, durability, and quality.
PRELIMINARY INVESTIGATION
Among the materials found in a search for secondary data was
a study entitled Paint , Varnish, and Lacquer by the Consumer
Research Division, Curtis Publishing Company, published in
May, 1940. This was a report on 1,741 consumer interviews
made in October, 1939, in 18 cities from coast to coast.
Questions asked included the following: when exterior was
last painted; what interior painting of any kind was done in the
last three years; what painting for both exteriors and interiors;
reason for painting; cost; who did the jobs; who purchased the
paints; who selected the brand and why; who suggested that
painting be done; what rooms have painted walls; what rooms
were changed from paper to paint or vice versa, and why; who
selected color and how; what term "paint styling" means, and
desire for information about it; seeing or use of paint manufac-
turer's book on interior decorating ideas and color selections;
possibility of paying for painting by budget or installment plan;
alterations or additions desired; need to paint exterior, and, if
38 MARKETING RESEARCH
need exists, why it is not being done now; seeing or hearing of
recent paint advertisements; which advertisements most help-
ful, best; and proportion of income spent on various items.
ACTION NEEDED
1. What decisions concerning the project might have been
made in the light of the availability of this previous study?
2. How might the findings of that study be utilized most effec-
tively in planning the present one?
CASE 2 PRELIMINARY INVESTIGATIONS
FOR A SMALL STUDY
BACKGROUND
The Modern Display Company is a small firm located in a
large Midwestern city. The major portion of its business is the
construction of dioramic exhibits or displays, many of them
being animated. These displays have been used most com-
monly in store windows and museums, and at conventions and
fairs. Clients have included large, nationally known corpora-
tions as well as many smaller companies. Company exhibits
were used during the 1930's at the New York World's Fair, the
San Francisco Exposition, and the Chicago World's Fair.
The business was organized in the early thirties. Until the
beginning of World War II it had operated successfully. Then
the business was closed for the duration because the owner-
manager went into service. He had special training and ability
in both art and topographical engineering, and the small busi-
ness he operated required these skills.
Re-establishing the business after the war presented new
problems. Although the owner was able to build up a compe-
tent staff of about twenty workers, he experienced great diffi-
culty in soliciting a sufficient volume of business. Just prior to
the war, sales had run at approximately $12 5,000 a year. In the
first years after the war, sales had never gone above $75,000 and
had been as low as $50,000. Unit sales also were far below pre-
war levels.
CASES IN PRELIMINARY INVESTIGATIONS 39
SELLING METHODS
The company relied for business almost entirely on direct
mail or on the initiative taken by prospective customers. The
major appeal used in direct mail promotions was to the durabil-
ity of displays produced. The mailings often resembled an-
nouncements more than sales letters.
Returns from these mailings often had been disappointing.
For example, the owner had decided the year before to feature
a dioramic display with a Christmas theme suitable for banks
and churches. The display was priced low, and could be pro-
duced profitably only if a hundred or more orders were secured.
A list was purchased that contained names and addresses of over
6,000 organizations. A series of four letters was then sent to
each name on the list. Only ten orders were received. As these
could have been filled only at considerable loss, the orders were
canceled.
At the time that a research consultant was called in, annual
mailings designed to sell exhibits for use at conventions were
being readied. Lists of exhibitors had been secured from the
convention bureau of the local Chamber of Commerce and
from a number of hotels. Letters were to be sent addressed to
the "Director of Convention Exhibits" of the companies on the
list.
THE PROBLEM
The owner had approached a free-lance consultant recom-
mended to him by a business friend. The owner felt that he
could not afford more than a few hundred dollars for research.
Although he was eager for any recommendations on securing
a larger volume of business at a reasonable sales cost, he con-
sidered direct mail to be the only practical means. The prob-
lem, as he saw it, was, "What has been wrong with our mailings
and how can we improve them?" The fee he had in mind was
$200.
ACTION NEEDED
1 . What type of preliminary investigations should the consult-
ant make?
40 MARKETING RESEARCH
2. Should the preliminary investigations actually be merged
with data gathering, and, if so, in what way and with what
probable reliance on secondary data?
3. What possible solutions are there to the company's prob-
lems?
CASE 3 THE EVOLUTION OF A HYPOTHESIS
BACKGROUND
The Rajah Men's Shop was a high-grade store located in the
downtown shopping area of a large city. Although advertising
extensively, it classified itself as a nonpromotional store. Only
a few clearance sales were held annually. The store did not run
price events or stress price or bargains in its advertising. Indeed,
it felt that men's-wear chain stores and most men's-wear depart-
ments of department stores in the city, the promotional stores,
were not really competitive.
Advertising, however, was seldom of the strictly institutional
variety. The bulk of the advertising, of course, was for men's
outerwear, and this advertising stressed brand names, partic-
ularly a line of high-priced clothing with a well-known national
label carried exclusively by this store in that city. Although the
store carried a wide assortment of men's and boys' clothing,
accessories, sportswear, and related items, men's suits provided
the major line.
The store wished to make a survey of men in the area which
would (1) test and expand its knowledge of the characteristics
of its customers, and (2) provide material to serve as a basis
for advertising and promotional efforts being directed toward
cultivating new customers and toward increasing business, or
strengthening store loyalty among present customers.
DEFINING THE PROBLEM
One of the questions raised by a store executive in outlining
the problem was: Does the influence of a woman on a man's
shopping decrease with his increased earning capacity?
Taken alone, this question seemed to be a side issue of rel-
CASES IN PRELIMINARY INVESTIGATIONS 41
atively little importance to the merchandising of men's suits.
Although the dominance of women as buyers was well known,
it was not believed that this was particularly relevant to men's
clothing. In so far as it was relevant, the answer to the specific
question very probably was that woman's role as a purchasing
agent for the family decreased as income increased. Moreover,
recent evidence suggested that, at least in secondary shopping
centers and suburban locations, men were tending to increase
their participation in family shopping. The fact that many
women accompanied men who were buying clothing was recog-
nized, but this was considered a special aspect of personal sales-
manship apart from advertising problems.
As a result, the tentative definition of the problem did not
include any major hypothesis concerning women's influence,
but simply raised the question: What influence have women
on customer behavior?
SITUATION ANALYSIS
Store records revealed nothing about women except that a
minority of accounts were in women's names; presumably these
were single women, widows, or mothers buying boys' clothing.
The advertising manager revealed that advertisements in subur-
ban presses often were used to sell boys' clothing. He assumed
that these advertisements were read mostly by women. Audi-
ence studies also showed that radio and television plugs had
many female listeners and viewers. In both cases some atten-
tion was given to women, but only for gift buying before Christ-
mas and Father's Day did the store attempt to promote men's
clothing to women. Even men's furnishings, extensively
bought by women normally, were advertised only to men.
The internal investigation indicated that about one fourth
of the customers came in with their wives. Salesmen were in-
structed to give at least as much attention in these cases to the
woman's preferences as the man's, although as unobtrusively
as possible. The store manager expressed his opinion that
women unquestionably exerted some indirect influence on the
men's purchases. In his opinion this influence was transmitted
by such comments on the part of women as: "You look good
42 MARKETING RESEARCH
in blue/' or, "Don't come back with one of those things again/'
Information from secondary sources also stressed the influ-
ence of women on men's purchases when accompanying men,
the dominance of women in purchases of furnishings and acces-
sories, and their importance in all lines of men's and boys' wear
sold in department stores. In one instance, the Rajah Shop
noticed an advertisement in an out-of-town paper by another
men's store addressed directly to women. It had not been
known that any men's store used this policy. A letter to the
advertising manager of that store elicited the following state-
ment:
In our stores, especially in the suburban areas, we have noticed that
the majority of men purchasing clothing are accompanied by women.
However, to back up our feelings in the matter I am attaching a bibliog-
raphy which I hope you will find useful. There is no doubt that women
influence a great deal of spending in this country. But I do not have
specific facts at my finger tips to give you at the moment.
The attached bibliography included these items:
Male vs Female Influence in Buying and in Brand Selection, Vol. II. New York:
True E. Fawcett Publications, 1950; also 1948 edition.
The Influence of Women on Buying. New York: Hearst Magazines, Inc , 1948.
Who Buys. New York: Batten, Barton, Durstine & Osborn, Inc., March, 1946.
P. D. CONVERSE and C. M. CRAWFORD. "Buying Habits in the Home," Adver-
tising Agency, XIX (July 26, 1948) .
"Women Dominate Male Spending, Fairchild Says," Advertising Agency, XIX
(July 26, 1948), p. 41.
"Women Still Influence Men in Suit Buying/' Wool Bureau Reports, Tide,
March 2, 1951.
This letter was not received until after the external investiga-
tion was begun. Except by hindsight, however, it is doubtful
that it would have influenced the interviews greatly in any case.
INFORMAL INVESTIGATION
Interviewers were sent into scattered neighborhoods prima-
rily to interview men, although a specified number of interviews
with women, or with men and women together, were allowed.
They were instructed to induce the respondents to talk about
clothing and to report the conversations. Among subjects to
be covered were where those interviewed bought their clothing;
CASES IN PRELIMINARY INVESTIGATIONS 43
what they looked for, liked, and disliked; and the influence on
purchases of advertising, window displays, brands, store name,
salesmen, friends, wives or other women, price, and store serv-
ices. Respondents were to be identified by age, by an estimate
of their income or income group, and by other factors. Selec-
tions and summaries from the more than one hundred reports
filed follow:
1. Man, 50, income about $6,500 per year. When younger bought
at Rajah's and at . . . ; but "they outpriced me/' Not affected by adver-
tising unless a known brand is sold at a markdown. Usually buys alone.
Not influenced by wife. Would like to buy in suburbs if stores had
branches there.
2. Man, 42, about $8,000. "I have been buying . . . (brand) . . .
suits for a long time." "I like to buy where I am known and a smaller
store gets to know you." "I am a stickler for brands." "I have more
confidence in brands." "Stores carrying national brands stand behind
the goods." "I went into a store with my wife to buy a hat, and bought
an overcoat, and then my wife insisted on a blue suit." "My wife insists
on buying more than I would ordinarily."
3. Man, 45, about $6,000. Thinks women influence men's shop-
ping, but "I don't like my wife to help me."
4. Man, 40, about $5,000. "Married men don't care as much about
appearance. We don't have to impress women anymore." "A wife
wants other women to look and think her husband is well dressed."
Buys when he needs a suit. Does not think income affects price paid.
"Credit is an equalizer." Brands do not influence him; prices and styles
do.
5. Man, 63, a fireman, but with other sources of income, $8,000. "I
buy all my clothes at ... ." Dealt there 30 to 40 years. "I know a sales-
man there. He's a good guy. You know, you can kid with him and
everything. I don't have to go down if I don't feel like it he just sends
up what I call up and ask for. He knows what I want." "One day I
said to the wife, 'I want a windbreaker. Want to come along?' She
says, Tine/ She never interferes. I got a good woman. Anyway, we
came back with two suits no windbreaker. She's easy going, a good
judge of clothes, too." "One suit I got I've had quite a while, a good
suit; but it's too young-looking for me now. One day I walked out of
the house with it on and the woman next door said, 'Who do you think
you are, some spring chicken?' I took it off went right back upstairs
and took it off. Haven't worn it since either." "Used to wear single-
44 MARKETING RESEARCH
breasted suits. The salesman says a double-breasted suit looks better on
me. He's full of that stuff. I've been dealing with him for a long time."
6. Man, 50, traveling salesman, about $6,000. Buys suits when what
he has wears out or becomes too small around the waist. "Sometimes
you put it off too late. Then I buy my stuff wherever I happen to be.
Most of the time, though, I buy my clothes in .... That's where the
home office is. I can get my stuff there easier, don't have to kill a Satur-
day when I'm home. Don't care what name is on the label. I go to
chain stores that I know. They make their own stuff. I know it's O.K."
7. Man, 45, shop owner, upper middle income class. Doesn't like
to shop. "I hate to get dressed up." Wife does all his shopping except
suits and shoes.
8. Man and wife, fortyish, middle income class. He didn't have a
chance to talk much. Wife supplied the talk. He doesn't care to shop.
Wife buys most of his clothing, even including suits and overcoats. He
doesn't even go with her, her judgment on purchases apparently being
final.
9. Wife, 60, high income. Husband has all suits tailor-made. Re-
cently he bought a ready-made suit, "and if you ask me, I must say it
looks like the devil."
10. Man, 38, about $5,500. In-laws worked at department store, and
he bought there because of the discount. Now he still goes there be-
cause he has a job nearby and it is convenient for him.
11. Man, 40, $8,000. "I'm not much on clothes." Buys a suit every
two years. "Sometimes my wife goes downtown and buys it. She takes
my size and goes out and buys it." A builder, he wears old clothes most
of the time.
12. Wife, 40, $4,000. "Oh, I never go with him." He buys only one
suit a year and one pair of slacks. "We have seven children, you know."
13. Man, 55, $3,500. "When I need a suit, I just go downtown to
... or ... (several department stores mentioned) and get one." "I
always take my wife. These salesmen try to push a suit onto you no
matter how it fits."
14. Man, 50, upper income class. Buys most suits in ... where he has
one of his offices and the rest made to order here. Wife goes along some-
times. Thinks local selection is poor, and must make a special trip
downtown.
15. Man, 48, insurance broker, upper income class. Usually buys
within walking distance of city-center office or where parking convenient.
Doesn't like department stores. Buys 7 or 8 suits a year.
16. Doctor, 30, $10,000. Wife is dominant in buying clothes. She
CASES IN PRELIMINARY INVESTIGATIONS 45
tells him when to buy. He seems to be proud of the fact that he doesn't
have much time to think about clothes. Wife usually accompanies him.
He bought a rayon suit on his own:"I got stuck. FH never go to another
sale/'
17. Wife, 30 (man, 45, engineer), $11,000. When married "he
was already an old bachelor and knew how to dress." "Once I bought
him some ties and he ranted and raved about them." "Where in h
salesmen ever get the gall to sell these monstrosities to women." He
buys loyally at one store, Rajah; formerly (for 15 years) at store in an-
other city.
18. Reporter, 45, about $6,000. Buys at no particular place. Used
to buy at Rajah's before the war but hasn't for a long time. "They're
high-priced. I don't always have the capital." Used to deal only with
one salesman there. Wife has no particular influence, he says. "I buy
what I want and then catch it when I come home."
19. Man, 35, about $5,000, works for manufacturer of men's apparel
for a large chain. "My wife is like everyone else. She thinks a $45 suit
is not a cheap suit and why buy a suit for $90 when you can get two for
the price." "Nothing like good clothes. I used to buy $3, $5, $8 ties,
and shirts for $8. People used to ask, 'Where did you get that tie?' or
'Where did you get that shirt?' That proves it's good, doesn't it?" "I
still like expensive clothes, but you can't afford those things when you're
married." "I could buy a suit for $17 at my place that wholesales for
$51, but I wouldn't enjoy wearing it so why should I buy it?" "Adver-
tising is the biggest farce in the world."
20. Man, 50, $3,000. Would rather buy two cheap suits than one
expensive one because "they all wear the same anyway." He claimed his
wife didn't influence his buying; but she was there and laughed when he
said this.
21. Man, 55, $9-$l 1,000. "When I buy a pair of shoes, I just walk
in, stick out my foot, and ask if they have any like the ones I'm wearing.
Haven't changed my shoe style in years." "When you have a bay win-
dow and narrow shoulders like I do, you don't dare wear anything but a
conservative suit."
22. Lawyer, 35, $10,000. Has to have all clothes made to measure,
as is about 6' 6" tall, Says he could afford more expensive clothes. Pays
about $60 for a suit; indicates that expensive means around $100.
Started buying at present store when a law student, and has continued
because he finds the clothes completely satisfactory.
46 MARKETING RESEARCH
ACTION NEEDED
The selections listed above from the reported interviews were
chosen largely to reflect findings relevant to the influence of
women on men's clothing purchases. At the same time, how-
ever, these selections suggested to the researcher other possible
hypotheses concerning customer behavior that might be used
as a basis for the full-scale study of the subject. What are these
hypotheses and what justification do these interviews provide
for them?
CASE 4 INTERNAL INVESTIGATION
BACKGROUND
A large university operated its own Press as one of its depart-
ments. The editoralso the directorwas responsible to a
governing board of university officials and faculty members.
The staff included the director, a sales and office manager, an
advertising and promotion manager, a production manager, a
bookkeeper, a shipping clerk, and five secretarial and clerical
assistants.
The Press did not have its own printing facilities. Printing,
binding, and other manufacturing operations connected with its
publication of some twenty to thirty titles each year were con-
tracted for according to Press specifications. All other business
operations, as well as editorial activities, were performed by the
staff. Publications were primarily of scholarly works and of
books of regional or local interest. The Press did not publish
fiction or popular books, and it neither sought nor, because of its
distribution facilities, was able to compete with commercial
publishers for manuscripts or sales. In addition to current pub-
lications, the Press had a back list of several hundred books
previously published and still in print.
In common with nearly all university presses so operated, this
one steadily lost money. No one was disturbed by this fact
until after World War II, when publishing and labor costs
rose, as did the amount of the annual deficit which had to be
CASES IN PRELIMINARY INVESTIGATIONS 47
net from general funds of the university. In 1949 the deficit
exceeded sales. Because the amount seemed excessive, and be-
cause the university was under financial stress at the time, con-
sideration was given to special methods of solution.
A decision was made to hire outside consultants. Two men
from the faculty of the School of Business were hired, one to
work part time for a month on an organizational study and the
other full time for a month on a study of business methods. In
general, these men were to work separately. This case concerns
the activities of the full-time consultant.
THE PROBLEM
Preliminary discussions concerning the troubles of the Press
and the type of study that seemed desirable had centered on
selling and advertising. Lack of volume appeared to be the
chief difficulty, and better promotion the chief remedy. The
hope was that an analysis of the distribution and promotion
policies and practices of other university presses and commer-
cial presses, plus the application of sound business practices in
general, would lead to improvements in operations. Greater
economies in operation were considered possible, but were not
thought to be a major problem, except perhaps as organizational
changes might be involved.
Thus, it was planned that the man concerned with organiza-
tional problems would spend all his time in the offices of the
Press and would make extensive use of its records. The other
man might wish to familiarize himself with current practices
and to examine a few records (or confer with the organization
consultant), but this would be merely a preliminary investiga-
tion on his part, not his main activity. He was assured, none-
theless, that all records would be open to him and that he was
entirely free to develop the study as he saw fit.
This man began his work by talking to various persons in the
offices, discovering all the types of records maintained and the
manner in which they were kept, as well as outside material
available on university presses, and examining and sampling the
records. After four days these methods continued to turn up
new problems and promising leads.
48 MARKETING RESEARCH
A new problem thus arose. Should he devote nearly all his
time to analyzing records contrary to what those hiring him,
and he himself, had planned?
DATA AVAILABLE
Under methods established by the budgeting office of the
university, the Press prepared monthly, semiannual, and annual
profit-and-loss statements. From these could be assembled data
on such natural expense classifications as salaries, postage, and
travel; but functional expenses could be estimated only crudely
from these records.
Duplicate invoices for each individual sales transaction were
filed numerically and showed the names and addresses of pur-
chasers (but not classified as to sales agent, wholesaler, book-
store, library, or individual, except as such classification was
revealed by discounts or customer listings), dates, list prices, dis-
counts given, cash or credit terms, postage costs, and return
allowances.
From the invoices used to assemble orders and ship them, the
shipping clerk kept a ledger by individual book titles showing
the date and number of books received from printers, all ship-
ments to buyers, shipments to those receiving complimentary
copies, returns, and the book inventory balance together with
annual physical inventory counts and adjustments.
From this shipping room ledger, author royalties and pay-
ments to subsidizers of publications were calculated. A sep-
arate record was maintained of amounts paid on each title and
dates of payment.
Contracts with authors and subsidizing organizations, both
university-affiliated organizations and others, were on file.
These yielded methods of calculating payments by subsidizers
to the Press and methods of calculating royalties on sales, au-
thors' copies, other complimentary copies, and other arrange-
ments. The latter included in some cases provisions for sharing
promotional or handling costs, or both.
A production ledger contained detailed quantity and cost
data on each individual title published regarding contracts with
firms, deliveries of sheets, bound books, payments made, total
CASES IN PRELIMINARY INVESTIGATIONS 49
costs, costs per book, storage of type and "killing" of type, and
other details connected with production.
The advertising department had summaries of total monthly
expenses, lists of complimentary copies issued, and mailing lists.
Unsummarized but in the files were details on campaigns, on
placement of display ads and amounts paid, special exhibits,
and all correspondence. Commissions paid to sales agents were
available in summary form.
Outside sources included a recent exhaustive study of thirty-
five university presses, by Chester Kerr. 1 This included detailed
discussion of business methods used together with cost data
(which differed, however, in accounting methods from that used
by the Press and were not strictly comparable in all regards).
Also available were articles on various aspects of university press
operations, contracts used by other presses, and miscellaneous
materials.
The preliminary inspection of these records and sources
hinted at a number of apparently grave internal problems.
Among these were the following:
For more than twenty years the Press had used a rule-of-
thumb method to cover manufacturing costs on subsidized
books, which were over half of all the books it published. To
the out-of-pocket costs paid for paper, printing, binding, and
other production operations was added a flat 10 per cent which
the subsidizing organization, or author, paid to the Press. This
amount was supposed to cover the salary of the production man-
ager, who designed the book and handled all production prob-
lems. It was also supposed to cover the editorial work of the
director and others, incidental expenses, and overhead. A cur-
sory inspection indicated that this sum fell far short of meeting
actual production expenses incurred by the Press. If so, a more
complete analysis could establish the fact and lead, perhaps, to
corrective measures.
In return for their contributing one third, one half, or even
all of out-of-pocket manufacturing costs (plus 10 per cent), the
Press offered its subsidizers widely varying amounts of royalties
1 A Report on American University Presses (Washington, D.C.: American
Association of University Presses, 1949).
50 MARKETING RESEARCH
running up to 65 per cent of the list price of books. These high
royalties were presumed to induce optimistic subsidizers to take
the risk of assuming manufacturing costs. A cursory examina-
tion suggested, however, that these rates were ridiculously high,
and that in many cases the Press was guaranteeing itself a loss
by such contracts. The more books sold, the greater the loss,
because the amounts netted by the Press could not cover hand-
ling costs. Further evidence on this point was supplied by an
examination of the discount schedule of the Press, which was
33*4 per cent on purchases of 1 or 2 copies, 40 per cent on pur-
chases of 3 to 24 copies, 41 per cent on purchases of 25 to 49
copies, 46 per cent on purchases of 50 or more copies, and 50
per cent on purchases made by representatives abroad for for-
eign sale. Over and above manufacturing expenses, the Press
expense as a percentage of the list price of books sold was, appar-
ently, more than 60 per cent. An inspection of invoices showed
few purchases, even by wholesalers, of more than 3 copies at a
time.
Because manufacturing expense was charged as a current
operating expense and was a major item in the budget, there
was a definite pressure to bring out subsidized books. This
pressure, together with the apparent situation on discounts and
operating costs, pointed to a possible policy problem.
Subsidization apparently had another bad aspect. Led to
believe that subsidized books were loss-proof books, staff mem-
bers were extremely liberal in passing out free copies. This
further reduced potential revenue.
An inspection of the size of editions and the amounts sold
suggested that the Press was consistently misjudging the market.
One biographical series of regional interest had 11 titles, 10 of
which had been issued in editions of 1,500 copies and 1 in an
edition of 1,000 copies. Only three of these books, however,
ever had sold more than 556 copies. A hasty inspection of
records showed that the bulk of sales in nearly all cases was made
in the first few months after issue, so that a method of forecast-
ing sales might be feasible. Forecasting could help to deter-
mine the number of sheets originally printed, the number of
copies bound, the advisability of retaining or killing type, the
CASES IN PRELIMINARY INVESTIGATIONS 51
desirability of reprinting. A quick check on reprints showed
that only a small fraction of reprinted books actually were sold,
a particularly disappointing showing on the problem of gauging
the market.
Discussions with staff members and an inspection of methods
of calculating costs revealed another problem. Because the
sales of most books were very small, a calculation of the cost
per book of an edition of 500 or 750 was always discouragingly
high. A frequent solution was to double the size of the edition,
which automatically reduced substantially the cost per book.
The apparent cost reduction was, of course, pure fallacy unless
most of these books actually were sold which the records sug-
gested rarely happened. More realistic costing, therefore, a
definite planned loss as against unrealized hopes for profits,
might be an important economy measure.
As noted, the number of free copies, or "exempts," seemed
abnormally high. These were under the supervision of the ad-
vertising department. In addition to this "no cost" promotion,
expenses of this department appeared to be way out of line with
that of other university presses. This might or might not be
due to the low volume of sales. The general policies of the ad-
vertising department also were suspect. Major expenditures
were for display advertising; that is, advertising in newspapers
and magazines for individual titles, a common practice in com-
mercial presses but not in university presses with their more
specialized markets. The timing also seemed to follow com-
mercial practice and to be badly adapted to the special circum-
stances of the Press.
Monthly payments to sales representatives averaging $5.38
in the South, $10.58 in the West, and $117.98 in the East im-
mediately suggested a possible sales problem, since these pay-
ments seemed ineffective in some places, and superfluous or
overcostly elsewhere.
Pricing policies and practices also seemed fruitful sources for
analysis.
52 MARKETING RESEARCH
ACTION NEEDED
1 . Should the consultant on business problems shift his analysis
primarily to internal records?
2. Was his internal investigation sufficiently conclusive to sup-
port a fixed plan for his study?
3. What problems were most important?
CASE 5 USE OF THE CONSUMER CLINIC
NATURE OF THE CONSUMER CLINIC
The Albion Research Associates, Inc., a general agency, had
devised a method of exploratory group interviewing to be used
during the preliminary and planning stages of consumer survey
research projects. The primary use of this method, called "the
consumer clinic/' was to assist in the more accurate formulation
of hypotheses and plans. Although not designed to supplant
the pretest an advance test under actual conditions of a formal,
often large-scale, surveyclinics had proved helpful in solving
questionnaire construction problems. 1
The clinic consisted of bringing together approximately ten
persons for a discussion of the project with those in charge of it.
The discussions usually lasted about two and a half hours, and
the respondents were paid $1 an hour for their time. House-
wives, for example, were most easily persuaded to attend meet-
ings held on weekdays between 2:00 and 4: 30 P.M., times when
they ordinarily might go shopping. The members of the clinic
were secured by sending out interviewers to solicit cooperation.
A truly representative sample was not considered either prac-
tical or necessary. Efforts were made, however, to secure par-
ticipation from all age, sex, income, or other classifications
relevant to the problems at hand.
Results of the early clinics were difficult to reduce to specific
statements or tabulations. Secretaries and research assistants
1 For a similar idea more recently applied by the Grey Advertising Agency of
New York see G. H. Smith and H. A. Vitriol, "Why We Use Group Explora-
tions/' Printers' Ink, July 18, 1952, pp. 46-47.
CASES IN PRELIMINARY INVESTIGATIONS 53
were used, thereafter, to take full notes on, and impressions of,
the discussions. Prepared questionnaires also were given to par-
ticipants. Indeed, it was found that participants would readily
complete several long questionnaires in addition to discussing
subjects raised. This made it possible to consider a great variety
of possible problems. Meetings also were improved by holding
them in comfortable surroundings and by providing charts,
large advertisements, samples, or other exhibits as stage settings.
Four to six separate meetings with different groups of partici-
pants were held, the number depending on the complexity of
problems faced and the rate of appearance of new ideas from
one meeting to another.
In assessing the consumer clinic as a research device, several
tentative conclusions had been reached. Although clinics prob-
ably were more expensive than usual methods, results also
seemed better. The directors of a project were able to obtain
information at first hand. Seeing and talking to respondents
in groups was far more effective than reading interviewer re-
ports, especially since executives were rarely able to go into the
field interviewing. Group meetings were particularly effective
in probing for opinions, attitudes, and motives. Better prob-
lem formulation, questionnaire construction, and methodology
planning were believed to result from clinics.
On the other side of the ledger were the added costs, if exec-
utive time were counted. In addition, participation in discus-
sions, despite all a chairman might do, depended heavily on the
articulateness and aggressiveness of individuals. No matter
how objective the researcher tried to be, it was easy to be im-
pressed by statements expressed forcefully and well and to mini-
mize ideas put forth hesitantly, though careful reading and
comparing notes later alleviated this problem somewhat.
Moreover, some tendency existed for participants to follow the
leader, though definite splits in opinion also were common.
The researcher directing the meeting might introduce bias
(quite unwittingly) because of gestures, intonations, or other
apparent clues to his views. Some suspicion existed also that
participants tended to become "conditioned" during the clinics,
causing them to react unnaturally. Nevertheless, most of these
54 MARKETING RESEARCH
problems were inherent in interviewing and were not special
afflictions of clinics.
After several years' experience with consumer clinics, includ-
ing their use for projects involving rugs, dish cloths, a beverage,
and the location of a department store branch, the agency was
considering using the consumer clinic for a current project, a
distributive efficiency study.
THE PROBLEM
The exploratory study of distributive efficiency, in which
Albion Research Associates had been asked to participate, was
financed by a grant from one of the large foundations. The
study was to be an attempt to formulate possible measures of
marketing efficiency and a test of the practicality of gathering
the data needed to make actual computations of marketing
efficiency in different situations.
An important segment of the study centered on consumer
marketing activity. One of the many tentative hypotheses, for
example, stated that "the marketing costs of consumers rise or
fall as the number of points to which they must go to do their
share of the marketing task and the distances they must go out
of their routine patterns of day-to-day travel in order to reach
these points increase or decrease." In considering this problem,
it was thought that it might be more significant to consider the
basic location of a consumer not as the usual point of residence
but as being along a line of normal or routine movement an
orbitoccasioned by his occupational, recreational, and social
activities. In some instances, time rather than distance of travel
would be the desirable measuring rod.
Although origin and destination (O.D.) and other traffic
studies, as well as store location projects, provided considerable
data on consumer movements within the metropolitan area to
be studied, none was sufficiently comprehensive or precise to
be useful for present purposes. The primary task of Albion Re-
search Associates, in fact, was to develop and test methods for
obtaining the required data on consumer activities.
The data considered necessary included in part the following
general categories (most of which were subdivided still further) :
CASES IN PRELIMINARY INVESTIGATIONS 55
Time Data:
1. Total time spent in marketing per day by housewives, by
workers outside the home, by school children, and by other
family members.
2. Time spent in getting to and from market areas or individual
stores.
3. Time spent in stores and public places of all kinds, plus pos-
sible breakdowns of activity within stores.
4. Time spent moving from store to store.
5. Time spent in traveling to and from work.
Distance Data:
1. Total distance traveled per day on trips to work or school,
social visits, and buying trips.
2. Distance traveled by types of transportation used, by routes
traveled, and by travel purposes.
Other Data:
1 . Expenditures made by type and amount.
2. Number of stores visited by type per day, per trip.
3. Number of unpaid marketing manhours.
ACTION NEEDED
Among decisions which had to be made were several pertain-
ing to the possible use of consumer clinics. Would consumer
clinics be useful in the preliminary stages of this project? What
types of questions might be discussed profitably? What sorts
of questionnaires could be constructed for use at the clinics?
How many persons should be invited? What types of people?
Because of the nature of the project, it was certain that sur-
veys, observational methods, and experimental methods would
be tested to some degree. Suggestions already had been made
that small panels be set up, with panel participants to keep diary
records of various sorts. If such panels were established, might
clinics composed of panel members be used in order to explore
such problems as those of recording data? Might a clinic be
useful after the panels were disbanded to explore the accuracy
of methods used, the adequacy of compensation, participation
by family members, and so forth? In other words, how might
56 MARKETING RESEARCH
consumer clinics be used in defining and delimiting the prob-
lems to be explored in this project?
CASE 6 CONSUMER PANEL VS. STORE AUDIT
RESEARCH
BACKGROUND
The Plumtree Corporation manufactured "Pomate" (a
special hair preparation) and a line of personal-care products.
Distribution was mostly through wholesalers. Drug stores
.originally had been the only retail outlet of importance. Plum-
tree products now were sold in food stores, department stores,
and variety stores, through mail order houses, and through other
miscellaneous outlets. In fact, sales of the major products in
supermarkets alone rivaled those in drug stores, a welcome
development profitwise but one which was raising vexing dis-
tribution problems.
Since the early 1940's the corporation had subscribed to a
continuous store audit research service. The corporation had
no marketing research department, and important research
decisions were referred to the president, Mr. Ernest Plumtree.
When first approached by representatives of the agency mak-
ing the store audits, Mr. Plumtree had not been conscious of
any pressing need for research. The arguments made, there-
fore, that store audits were a far better technique than were
interviews with either retailers or consumers (because interviews
were subject to error, bias, and memory loss, and often failed to
produce real facts or accurate numerical data), had not seemed,
frankly, of much relevance. Nor had claims that a continuous
study was far superior to static surveys been impressive at the
time. The furor about the Literary Digest poll failure, although
recalling painful memories, had not been entirely convincing
either. In general, he appeared awed by statistical matters and
suspicious about sampling techniques as applied to business
affairs.
What, perhaps, had convinced him most that the corporation
might benefit by contracting for the store audit service was the
CASES IN PRELIMINARY INVESTIGATIONS 57
information it revealed about competitors. It was impressed
upon him that his principal competitor, through this service,
already knew a great deal about his 'company's practices and
accomplishments which he regarded as extremely confidential.
(A private inquiry convinced him at once that the corporation
could not afford its own store audit, and that even if it could,
competitors still could obtain the same information.) The
amount of up-to-date, detailed information he could obtain on
his major competitors, of course, was highly attractive. More-
over, he had seen at once how records on factory sales to whole-
salers might be completely misleading in assessing current pro-
motional campaigns, which were of particular concern to him.
Many other considerations had also entered into his decision
to subscribe to the store audit service.
Once made, the decision had never been regretted. Unques-
tionably, store audit services were a sound business investment.
Although the fees had seemed large at first, they were but a
small fraction of promotion costs alone. The information ob-
tained was extremely helpful in making nearly all major market-
ing decisions. The corporation had applied the services to two
other commodities, one of which, however, was discontinued,
although "Pomate" remained the mainstay of the company.
Moreover, the detection and assessment of competitive moves
had on several occasions led to quick and successful imitation
of the same practices. No doubt, the profit on these ventures
had paid for research costs in full.
These matters were fresh in the president's mind because an
organization offering national consumer panel research services
recently had made a presentation before all top executives of
the corporation. The reactions had been favorable. The ques-
tion facing Mr. Plumtree was: should the corporation contract
for this service also? If it did, should both the store audit and
the consumer panel service be used, or should the corporation
contract only for one?
STORE AUDIT METHODS
Store audits were based on a national sample of retail stores.
Auditors called periodically at these stores to transcribe data
58 MARKETING RESEARCH
from store records on all products for which the agency had
clients. Stores were compensated for their cooperation. Be-
cause convenience goods stores kept no records of unit sales, it
was necessary to estimate unit and dollar sales for each package
size and brand of products studied. This was done by obtain-
ing two sets of figures: (1) purchases made (from manufacturers
or wholesalers as shown on invoices) and the prices paid, and
(2) a physical inventory count. By adding purchases to inven-
tory at the beginning of the period and deducting inventory
holdings at the end of the period, the sales, or disappearance, of
the product could be calculated. (Purchases plus or minus in-
ventory change could also be used to estimate sales.) Estimates
of turnover, stock-sales ratios, margins, and out-of-stocks could
also be computed from these figures. In addition, auditors ob-
served and gathered information on special purchases, on deals
or cut-price sales, on advertising, and on window and interior
displays of products. For purposes of breakdown analysis, all
cooperating stores were classified as food, drug, and so on, stores,
as chain or independent organizations, by size, by service or self-
service, and by city size, county size, and geographical region.
As a result of these audits, the Plumtree Corporation could be
provided with these broad categories of data:
1. Sales to consumers by package size, by brand, by type of
product, and by commodity group in units and in dollars.
2. Retailer purchases in total, by number of reorders, and by
average order size, and by purchases made direct or from
wholesalers.
3. Inventories in units, dollars, and total inventory changes,
days of supply on hand, and out-of-stock items.
4. Prices paid by dealers and by consumers, special deals and
sales, and retailer gross margins.
5. Display and advertising activities by retailers for company
products and those of competitors.
All these data, when cross-classified with each other and with
store classifications, provided a host of breakdown analyses.
Each variety of "Pomate" was analyzed for the company in rela-
tion to each of four major competitive products and in relation
CASES IN PRELIMINARY INVESTIGATIONS 59
to all minor competitive products. Thus, the share of the
market and all trends in the market as a whole or by regions,
city sizes, types of stores, and so on, were periodically available.
In short, store audits constituted a ''waterfront" job of research
providing continuously in one "package" a large amount of
basic market information. This information could be applied
to a whole range of marketing problems, such as problems of
pricing, packaging, display, advertising, and selling. Supple-
menting it with information on competitors obtained from The
Standard Rate and Data Service, from salesmen, and from other
sources, the company could not only assess its share of the
market in total and by many classifications but could also eval-
uate the results achieved by competitors.
CONSUMER PANEL METHODS
The national consumer purchase panel, as operated by the
organization which had made the presentation, utilized diary
records to be kept by panel members. These members were
selected to constitute a cross section of all American families.
As members dropped out, or were rotated, replacements were
added to maintain the representativeness of the panel. All pur-
chases by family members of products being studied were to be
reported on diary forms provided. The information, to be re-
corded as soon as the purchaser returned home, varied some-
what with different types of products. In general, it consisted
of date of purchase, age and sex of purchaser, complete product
description including brand, weight or size, type, number of
units, and price paidand type of store patronized or method
of purchase. Families cooperating were compensated by receiv-
ing coupons redeemable for a wide variety of goods. Each
family unit was identified by such characteristics as number of
members, age and sex of each, occupation, income, education,
and location. Again, cross-classification of the data gathered
made a great number of breakdown analyses possible. The
panel provided a continuous flow of basic market information
about the products of a company and those of its competitors,
and data were useful for a whole range of marketing problems.
60 MARKETING RESEARCH
CONFLICTING CLAIMS
Mr. Plumtree was somewhat disturbed by the conflicting
claims and adverse criticisms made concerning these two sys-
tems. Store audits, it was said, were faulty, or were weakened,
by inability to audit the stores of some large food chains and
syndicate or variety stores, and also by failure to include door-
step and mail buying and miscellaneous retail outlets. These
difficulties, it was asserted, tended to make the sample of stores
unrepresentative of actual purchases. Moreover, poor retail
records, disappearances of merchandise because of shoplifting,
employee thefts, breakage, and other causes contributed to in-
accuracy in data gathering. On the other hand, it was claimed
by the proponents of the store audit method that these prob-
lems could be handled by careful sampling, by proper compensa-
tion of store managers, and by other auditing and statistical
practices.
Panels, it was said, could not be maintained on a probability
basis, because cooperation of some types of families (such as the
very rich, the very poor, and illiterates) was difficult or impossi-
ble to secure; purchases by others than the housewife (or adults)
tended not to be recorded; consumption of articles outside the
home frequently was ignored; families were continually drop-
ping out of the sample; being on a panel made persons act
atypically; merchandise premiums affected purchases made;
and, in general, accurate reporting was difficult to obtain. On
the other hand, it could be claimed that by using great care in
sampling, by providing families with proper report forms and
compensation and incentives, by careful selection and pretrain-
ing of panel members, and by other statistical and control
devices it was possible to secure continued, accurate, normal
purchasing reports.
Mr. Plumtree tended to agree with both sides. He was aware
that risks went with rewards and that when a new policy ob-
tained benefits, it also created new difficulties. He was fully
ready to believe that both systems were practical and reliable,
although each no doubt possessed inherent limitations. Dif-
ferences in cost did not seem important. Although he was con-
cerned with the accuracy of results, particularly as they might
CASES IN PRELIMINARY INVESTIGATIONS 61
affect "Pomate," he believed that the answer for his company
depended primarily on the sorts of information provided.
COMPARISON OF INFORMATION
Mr. Plumtree had asked the vice-president of accounting to
have his statistical department prepare a list of the information
furnished by each method. In preparing this list, consideration
was to be given only to data provided in the basic ''package/ 7
not to data secured by special methods and extra costs. The
resulting list was as follows:
Information Provided Only by Store Audit:
1. Dealer inventories. Stores stocking, amounts on hand, days
of supply, and out-of-stocks, by product types, major brands,
sizes, and in total. (By use of factory sales records, whole-
sale stocks can be estimated.)
2. Dealer purchases. Number and size of orders, special pur-
chases, and purchases from manufacturers or wholesalers.
3. Dealer margins.
4. Dealer display. Shelf position, internal store displays, and
window displays.
5. Dealer advertising. Newspapers, store signs, sales events.
Information Provided Only by Purchase Panel:
1. Users and nonusers. Families by types and size using major
brands, product types, and sizes.
2. Amounts purchased. Frequency of family purchases in
units and sizes, average and total purchases, concentration
of markets among large and small users.
3. Trends in purchases. Gain or loss of customers.
4. Age and sex of purchasers.
5. Brand loyalty. Amount and character of repeat buying,
brand switching, and number of brands used by types of
families.
Information Provided by Both Methods:
1. Purchases at, or sales by, retailers by product types, brands,
and sizes.
2. Types of outlets for sales or purchases made.
62 MARKETING RESEARCH
3. Seasonal variations in sales.
4. Regional and city-size breakdowns.
5. Retail prices.
REFERENCES AVAILABLE
Investigation by Mr. Plumtree and his staff of the literature
on the subject of consumer panels and store audits revealed the
following references:
BIBLIOGRAPHY
BARTON, S. G. "The Movement of Branded Goods to the Consumer/' in A. B.
BLANKENSHIP (ed.), How to Conduct Consumer and Opinion Research.
New York: Harper & Bros., 1946, pp. 58-70.
CAWL, F. R. "The Continuing Panel Technique," Journal of Marketing, July,
1943, pp. 45-50.
CHURCHMAN, C. W. (ed.) et al. Measurement of Consumer Interest. Phila-
delphia: University of Pennsylvania Press, 1947, pp. 155-63 (a speech by
S.G. Barton).
FLEISS, M. "The Panel as an Aid in Measuring Effects of Advertising/' Journal
of Applied Psychology, XXIV (December, 1940), pp. 685-95.
FERBER, R. "Twelve Pointers on Selecting a Consumer Panel Service/' Print-
ers' Ink, CCXXVI (March 4, 1949), pp. 42 ff.
GANLY, R. H., and CRISP, R. D. A series of articles concerning various aspects
of panels beginning with the August 8, 1947, issue of Printers' Ink.
NIELSEN, A. C. "Evolution of Factual Techniques in Marketing Research," in
Marketing Research and Business Management. Urbana: University of
Illinois Bulletin, XLIX, No. 44, 1952.
ROOT, A. R., and WELCH, A. C. "The Continuing Consumer Study," Journal
of Marketing, VII (July, 1942), pp. 3-21.
ROVNER, S. "J. Walter Thompson's Panel Shows How Clients Are Doing,"
Editor and Publisher, LXXXI (May 8, 1948), p. 14.
SELLERS, M. "Pre-Testing of Products by Consumer Juries," Journal of Market-
ing, VI, No. 4, Part 2 (April, 1942), pp. 76-80.
STONBOROUGH, T. H. W. "Fixed Panels in Consumer Research," Journal of
Marketing, VII (October, 1942), pp. 129-38. Reprinted in Marketing Re-
search: Selected Literature, by H. G. WALES and R. FERBER. Dubuque:
Wm. C. Brown Co., 1952, pp. 105-12.
ACTION NEEDED
1. Is there any basis for deciding that one method is inherently
better than the other?
2. Can both methods provide accurate data?
3. What facts should determine the use by a company of either
or both methods?
4. What should Mr. Plumtree's decision be? Why?
Part III
PLANNING
The General Nature of the Task
The cumulative steps of problem formulation and prelim-
inary investigations culminate in specific, formal plans. The
great amount of time often spent in these early stages testifies
to the critical importance of the decision-making which is done
at the planning stage.
Planning consists of establishing specific purposes, choosing
methods for data gathering, preparing all forms needed to gather
data, designing a sample, 1 testing plans, and estimating costs.
*
Specific Tasks and Working Methods
Because planning embraces so many activities, these activities
will be discussed separately and in conjunction with methods
used in carrying out each separate task.
Specific Purposes. The heart of this process is the final, con-
crete, written expression of the hypotheses.
Too often problem formulation stops at the intermediate
stage of characterizing information which it is hoped will be
useful. Correct procedure is to decide purposes in terms of
ultimates the conclusions and decisions which can be based on
specified data.
Although their application is difficult, the methods are sim-
ple: written expression of all important hypotheses that have
developed, selection of a limited number of these hypotheses,
and the clearest possible statement of them.
From such specific purposes it is possible to decide (1)
1 This part of planning is considered separately in Part IV.
63
64 MARKETING RESEARCH
whether or not to continue the study and (2) whether or not
the purposes can be satisfied wholly or partially with second-
ary data.
Methodology. Assuming that the project will require pri-
mary data, a choice of the method, or methods, to be used is now
made.
The three principal methods are survey, observation, and ex-
periment.
The survey methods are by mail, telephone, or personal inter-
view. They also may be classified as to type of data gathered.
Thus, Brown 2 distinguishes between factual, opinion, and inter-
pretive surveys; and Lorie and Roberts 3 characterize the data
(also obtainable, in part, by observation or experiment) as be-
havior, intentions, attitudes, knowledge and habits of thought,
and "reasons why."
Observational methods include store audits, pantry inven-
tories, fashion counts, and traffic counts.
Mechanical methods of recording data may be considered as
.a separate category, as subcategories of the other three methods,
or as a form of observation. They may be used in personal in-
terviewing (wire or tape recorders, or lie detectors), in group
.studies (Lazarsfeld-Stanton Program Analyzer or Hopkins Tele-
vote Machine), to record actions (Purdue Eye-camera, Audi-
meter, Radox), and in many other ways.
No mechanical way of choosing the proper method exists.
For many studies there are a number of alternative methods,
each of which also might be applied to several different sources
of information. A general preference exists for the experimen-
tal and observational (including machine) methods because
they tend to be more scientific. Unfortunately, these methods
often are more difficult to apply (and often impossible) and they
tend to be more time-consuming and costly. In fact, the survey
method, so often inveighed against, remains as the most com-
mon tool of research, and a highly useful one. The researcher
should beware, however, of limiting himself to easy, familiar, or
* L. O. Brown, op. cit., pp. 296-302.
8 J. H. Lorie and H. V. Roberts, op. cit., pp. 211-14.
PLANNING 65
favorite methods. Awareness of the new techniques and the
new applications of older techniques is helpful.
Specific consideration of the methods in relation to the data
sought often leads to a natural choice. Cost and time factors
also may help to determine the appropriate method.
Forms. Forms and schedules used in the collection of data
include questionnaires for surveys, schedules or forms for record-
ing observations or experimental data or for recording secondary
data, instructions and work schedules for the data gatherers,
and letters to respondents.
Questionnaire elements include identification of respond-
ents, explanations, data questions, and request for cooperation.
The various problems encountered, or rules for questionnaire
construction, deal with such matters as length, sequence of
questions, ease of recording answers and making tabulations,
spacing, clarity, specificity, and leading questions. (See the an-
notated bibliography which follows for references on these dif-
ferent aspects.) Most troublesome, perhaps, is the wording of
the questions, which includes a choice of question types: free-
response or open-end, check-list or cafeteria, dichotomous or
yes-no, ranking or intensity-scale, triple-associates, and probing
or follow-up questions.
In designing forms the major problems are clarity, compact-
ness, and ease and simplicity of recording data.
Working methods vary widely with individuals. Of primary
importance, the researcher must clearly grasp the exact char-
acter of the data he is seeking. He also must possess the pa-
tience to attempt many alternative wordings of questions (or
designs of forms) and continually to revise and select. Almost
everything he has learned may prove useful in executing this
most challenging job of communication. For the beginner,
interest in the process and ability to plan research are acid tests
of his prospects for research as a career.
Testing. Testing may relate to the sample, the questionnaire
or schedule including the instructions, the interviewers or field
staff, or the prospective information itself. For example, test-
ing the production rate of interviewers may avoid serious mis-
66 MARKETING RESEARCH
calculations as to time and expense. Additional tests can be
used to ascertain the adequacy and clarity of instructions which
are provided interviewers as well as of the questionnaire itself.
Seldom is it wise to forego a test of the questionnaire or other
data-gathering form. Pretesting almost always will reveal de-
fects, and many times very important ones, or suggest improve-
ments that add to accuracy and reduce time and cost.
Of special importance, and particularly so if the full, written
expression of specific purposes has been skimped, is a test of the
value of the information obtained. This can be done during
questionnaire formulation by anticipating findings. By assum-
ing a range of statistical results, tabulating these data, and then
analyzing them, it can be determined if, or how well, the data
sought will answer the problems.
Costs. The researcher may have to estimate costs in some
detail at the very beginning. With experience and some guess-
ing, this can be done. At other times a budget figure is pro-
vided, which serves as a basis for a formal research proposal. In
still other cases, detailed cost estimates may be the last stage
of planning, carried out as a means of insuring that the project
is feasible within the funds allotted.
In any event, data gathering tends to be the most expensive
single operation. Until methods are chosen, a sample is de-
signed, and some testing is done, costs are difficult to estimate.
Costs may be computed as out-of-pocket expenses with or
without a fee or supervisory salaries, or as out-of-pocket expenses
plus 100 per cent in order to cover the general overhead of the
agency and allow a profit.
Research Proposals
All the steps discussed so far may be telescoped into a prelim-
inary proposal for a research project. If done without design
of forms or any testing, such a proposal may be subject to con-
siderable change. The ability to make a satisfactory proposal
in detail depends on the experience and abilities of the re-
searcher. He may utilize considerable secondary data and
some preliminary investigations.
PLANNING 67
Such proposals are not a substitute for the steps so far dis-
cussed, although good ones should minimize the time and effort
needed later in covering these stages of research.
The Cases in This Part
Case 1, "A Radio Audience Survey/' presents the plans made
for a study as included in a proposal to the manager. Emphasis
is placed upon deciding the specific purposes of the study. The
case includes estimates of costs and some sample problems.
Case 2, "Proposed Consumer Market Survey/' presents the
problems of an independent retailer. Questions of specific pur-
poses are involved, although emphasis is given to methodology,
particularly to the adequacy of the tentative questionnaire form-
ulated.
Case 3, "Use of Internal Data for Studying Magazine Mer-
chandising Effectiveness/' is a full-scale proposal which presents
problems of specific purposes, methods, and sampling. The
emphasis is placed on considerations of using internal data.
Case 4, "Methods for S % tudy of Installment Credit/' is the last
of these cases presenting rather detailed proposals and covering
various planning problems. The emphasis is on the choice of
methods for gathering the data outlined.
Case 5, "Pretesting a Questionnaire," shows the question-
naire designed, the results of a pretest of this questionnaire, and
a revised questionnaire. Designed primarily to provide an ex-
ample of pretesting, the case can be used to cover many aspects
of questionnaire formulation.
Case 6, "Constructing a Readership Questionnaire," provides
practice in constructing a questionnaire and a covering letter.
Case 7, "Proposed Questionnaire for an Ice Cream Survey,"
presents a tentative questionnaire. It can be evaluated in terms
of survey purposes and the "rules" of questionnaire construc-
tion, or a revision can be attempted.
Case 8, "Getting a Luggage Questionnaire into Shape,"
focuses on the variety of specific questions suggested for gather-
ing data of specified types. The many details provided permit
an examination of the important minutiae of questionnaire
formication.
68 MARKETING RESEARCH
Case 9, "Following up on a Department Store Study/' ranges
over many aspects of preliminary investigations and planning.
Thus, it presents an opportunity to review earlier steps in the
light of planning problems.
BIBLIOGRAPHY
AMERICAN MARKETING ASSOCIATION. The Technique of Marketing Research.
New York: McGraw-Hill Book Co., Inc., 1937. An old classic, still useful.
Chapter 1 includes discussions of planning and of earlier research stages;
Chapter 2, "Planning of Procedure"; Chapter 3, "Psychological Aspect of
Questionnaire Development"; and Chapter 4, "Further Considerations of
Psychological Aspects," suggest that most basic principles remain unaltered.
BLANKENSHIP, A. B. Consumer and Opinion Research. New York: Harper &
Bros., 1943. Chapter 3 amplifies statements made in Chapter 2 about the
general nature of planning. In Chapter 4 there is a discussion of the advan-
tages and disadvantages of the mail, telephone, and personal interview meth-
ods and the use of combination methods. Chapter 5 covers the type of ques-
tions which may be used and eight "general rules helpful m proper phrasing/'
Chapter 6 includes a two-page discussion on length and a detailed examina-
tion of problems of question sequence. Chapter 7 includes a discussion of
the pretest (pp. 82-91) for all surveys and of the test-tube study (pp. 91-97)
prior to, or substituting for, large-scale surveys. Chapter 10 discusses and
illustrates forms used in all types of surveys. This book may be read for
problems of question phrasing and for its references to specialized sources.
BRADFORD, E. S. Marketing Research. New York: McGraw-Hill Book Co.,
Inc , 1951. The elements of planning are covered briefly on pages 1 33-38 of
Chapter 8. Chapter 10 outlines the pros and cons of the three types of sur-
veys, includes a discussion of interviews "in depth" and Gallup's "filter"
questions, and ends with an analysis of a television questionnaire and a state-
ment of rules for preparing a questionnaire. Chapter 18 provides a highly
condensed check list for sources of information.
BROWN, L. O. Marketing and Distribution Research. New York: The Ronald
Press Co., 1949. A comprehensive, well-organized discussion of planning
is contained in Chapter 19. This chapter also contains an abbreviated list
of secondary sources (pp. 397-402). Chapter 20 covers communication
problems, the usual pros and cons of the survey methods, a good explanation
and illustration of questionnaire elements (particularly of classification of
data, pp. 427-31), ten "general rules for constructing questionnaires," each
discussed and illustrated, and a brief consideration of forms for observation
and experiments.
EASTWOOD, R. P. Sales Control by Quantitative Methods. New York: Colum-
bia University Press, 1940. Chapter 9 contains much of the usual discussion
of advantages and disadvantages, but is notable for the stand it takes against
business secrecy about data and an analysis of "some deficiencies in the stock
of statistical data." Appendix B, "Mechanics and Principles Governing the
Use of the Mail Questionnaire," is a brief statement of fundamentals.
HEIDINGSFIELD, M. S., and BLANKENSHIP, A. B. Market and Marketing Analy-
sis. New York: Henry Holt & Co., 1947. Chapter 6 is primarily devoted to
PLANNING 69
a consideration of survey methods but includes a statement on panels, notes
the observational method in passing, and discusses the "applications of the
survey technique to business problems." Chapter 7 is an abbreviated discus-
sion of rules, tests, and formats for developing the questionnaire.
HOBART, D. M. (ed.) Marketing Research Practice. New York: The Ronald
Press Co., 1950. Chapter 5 contains a condensed statement of secondary
data, mail and telephone surveys, panels, and controlled experiments. Chap-
ter 6 is a brief consideration of questionnaire development with many illus-
trations.
LORIE, J. H., and ROBERTS, H. V. Basic Methods of Marketing Research. New
York: McGraw-Hill Book Co., Inc., 1951. In contrast to the frontal attack
of other textbooks, the authors attempt an enveloping movement on prob-
lems of communication. These exhaustive statements in Part IV, "Com-
munication and Observation," include Chapter 12, "Introduction," and
five chapters entitled "Securing Information on . . ." the following: Chapter
13, ". . . Behavior"; Chapter 14, ". . . Intentions"; Chapter 15, ". . . Atti-
tudes"; Chapter 16, ". . . Knowledge and Habits of Thought"; and Chapter
17, ". . . 'Reasons Why.' " Although it is difficult to cull useful short
assignments from this original text, perhaps the following will be found un-
usually good in textbook literature : the kinds and sources of response errors
(pp. 203-7), the summary statement on kinds of information (pp. 208-14),
observation and self-observation (panels) (pp. 217-24), usefulness of state-
ments of intention in forecasting (pp. 256-62), and depth interviewing (pp.
312-15).
LUCK, D. J., and WALES, H. G. Marketing Research. New York: Prentice-
Hall, Inc., 1952. Chapter 5, in addition to much of the usual material on
methods, includes discussions on the relationships between the methods (pp.
104-8), comparison of mail and personal interview methods (pp. 113-16),
and features of the panel (pp. 119-24). Chapter 8 contains material with
illustrations of appraising the need for added facts (pp. 155-56), a detailed
plan of data requirements (used by a government body to secure bids) (pp.
157-61), deciding how to secure primary data (pp. 162-64), anticipating the
findings (pp. 164-66), determining the tasks and costs (pp. 166-68), and
pretesting (pp. 170-72). Chapter 10 contains an illustrated treatment of
preparing questionnaires and forms.
PARTEN, M. B. Surveys, PO//S, and Samples. New York: Harper & Bros., 1950.
Although avowedly a specialized book, this volume is notable for its bibliog-
raphy (pp. 537-602) of 1,145 items and for its attention to general principles
as well as specific procedures and facts. Chapter 2 discusses twelve general
questions that need answering before detailed plans for surveys or polls can
be developed. Chapter 3 is a summary of various methods (and their varie-
ties) for securing information from persons. Chapter 6 is a statement of
the construction of the schedule or questionnaire, most generally useful,
perhaps, for discussions of identifying and classifying data (pp. 163-74), forms
of opinion questions (pp. 181-89), and rating or intensity scales (pp. 190-98).
Chapter 11 contains an illustrated discussion of mail questionnaire proce-
dures.
PAYNE, S. L. The Art of Asking Questions. Princeton: Princeton University
Press, 1951. The discussion in Chapters 1 and 2 indicates the searching
questions that researchers can ask themselves in making a preliminary study
of a problem. Specific problems of question formulation are also treated.
Part III Cases
CASE I A RADIO AUDIENCE SURVEY
BACKGROUND
The manager of Station WKGP, Philadelphia, believed that
a study of the complete station listenership might be valuable.
National time buyers, he thought, were not sufficiently con-
vinced of the volume of listening in areas outside the Philadel-
phia metropolitan area. These time buyers tended to use the
wasteful practice, he felt, of buying time from stations in sur-
rounding cities, such as in Baltimore, Wilmington, Reading,
Atlantic City, and Lancaster, to reach areas covered by WKGP's
signal. In his opinion this resulted in much duplication of pro-
grams. Additional time on WKGP, he believed, often would
provide better coverage at lower net cost.
The station, however, lacked up-to-date, accurate data on sta-
tion "circulation" and on duplication of listeners among sta-
tions in the area. In early 1951 no regular radio research service
was available which provided the data desired.
TYPE OF STUDY CONSIDERED
An engineering survey showed that currently WKGP's signal
could be received in more than sixty counties, or parts of coun-
ties. These included seven in Maryland, two in Delaware, all
in New Jersey except a northern tier, and all in the eastern half
of Pennsylvania, except a few adjacent to New York State.
Only the BMB Station Audience Reports * covered this entire
1 The Broadcast Measurement Bureau was organized in 1945 by the Ameri-
can Association of Advertising Agencies, the Association of National Advertisers,
and the National Association of Broadcasters as a nonprofit, cooperative organi-
zation. A mail-ballot study of each county in the United States was published
in 1946. It contained the number of radio families for each county (and for
70
CASES IN PLANNING 71
area for listenership data useful to Station WKGP. The Spring,
1949, report showed that in twenty-thcee counties in Pennsyl-
vania, six in New Jersey, two in Delaware, and one in Maryland,
at least 10 per cent of all radio families listened at least once
each week to a daytime program of Station WKGP. In fifteen
counties listenership was over 30 per cent, and in seven counties
was over 50 per cent.
Nighttime programs were listened to by more than 10 per
cent of radio families in thirty-six counties (one more in New
Jersey and three more in Pennsylvania) . The total weekly audi-
ence in these thirty-six counties for nighttime programs was
approximately 900,000. The weekly nighttime audience in
Philadelphia County included 80 per cent of the 573,000 radio
families. Approximately 40 per cent of the total weekly audi-
ence actually listened six or seven days or nights a week.
Useful as the BMB data was, it lacked much detail desired by
time buyers. Moreover, the manager felt certain that through
technical improvements and new programming the station had
increased its listenership ^greatly in the past two years.
Regular radio research* services, such as those of Pulse, Inc.,
and A. C. Nielsen, Inc., commonly used by time buyers, have
many useful data on program audiences and audience composi-
tion but not for the entire area covered by the station and not
in terms of station listenership. For example, the station had
made a number of experiments with short plugs for give-aways
or specialty items and had secured a greater number of responses
than the total rated audience. These services, therefore, did
not meet the problem the manager had in mind.
The manager was impressed by studies conducted annually
since 1938 for a station in Iowa 2 which, he had been informed,
cities of 25,000 or less, if a station city) and for each station the total weekly
audience (number of radio families listening to a station one or more times a
week) and the daytime and nighttime audiences. Study No. 2, BMB Station
Audience Report, published in the spring of 1949, additionally contained
listenership for six or seven times day or night, three, four, or five times, and one
or two times.
8 See, for example, the 1948 Iowa Radio Audience Survey, conducted by
F. L. Whan, published by the Central Broadcasting Company, Des Moines,
Iowa.
72 MARKETING RESEARCH
were readily accepted by time buyers. In particular, these
studies reported a high degree of "out-of-home" listening.
Such listening was not covered by regular radio research services
and was growing in importance because of television competi-
tion.
The Iowa Survey of 1948, 3 for example, included questions
on (1) station preferencesstations listened to most, stations
heard regularly, station liked best for news at major periods of
the day, and stations preferred for farm programs; (2) radio
ownership; (3) best-liked programs for adult listeners; (4) types
of programs liked best; (5) actual listening by half-hour periods
and by composition of the radio audience; (6) families owning
portable, nonportable, FM, and auto radios; (7) age of sets; (8)
amount of time auto radios are used on long trips and on short
trips; (9) families with some member regularly hearing a set out-
side the home and the location of such sets; (10) location of sets
in the home; (11) early morning listening by quarter-hour peri-
ods; (12) listening from 6 to 7 P.M. by quarter-hour periods; (13)
amount of listening to the "beginning commercial" and mem-
ory of item advertised; (14) frequency of use of two or more sets
at once in the home; and (1 5) criticisms of program heard.
THE STUDY PROPOSED
In May, 1951, the manager of Station WKGP called in a re-
search consultant. The manager, after discussing the station's
problems, requested that the consultant submit proposals for a
study along the lines of the Iowa Reports. The proposal sub-
mitted is reproduced in part below:
The proposed study would be, basically, a study of the "circulation"
of all radio stations giving a satisfactory signal within the service area of
WKGP. The primary purpose would be to show radio service coverage
and the pulling power of stations. In addition, the study would yield
important data on types of programs (but limited, if any, detail on spe-
cific program ratings) and comparative data for television and news-
papers, all data being considered for their value in selling and buying
radio time.
The study would be patterned on the Iowa Radio Audience Surveys,
9 Ibid.
CASES IN PLANNING 73
changed, however, to fit local area interests and problems. Exact
changes would be worked out later in cooperation with WKGP and by
testing in the field. Tentatively, the following changes are suggested for
consideration:
Elimination of data on:
Age of radio sets owned.
Farm program listening (except as a type of program).
Hearing or remembering specific advertising.
Hearing criticisms against radio.
Revision of data on:
Age classifications, to include males and females under twenty-one.
City classifications, to allow city-size summaries.
Newspaper circulation, to include three local papers for daily and
Sunday editions.
Autos, to include number and get more detail on use.
Dual listening and out-of-home listening to get more complete data
(even on nonradio homes).
Addition of data on:
Number of persons in family by age groups and total primary listening
audience.
Television families.
Families using telephone or mail for purchases.
Cities in which specified shopping goods are purchased.
Most of these changes affect so-called identification data which are
economically obtained as a by-product of a survey and which increase
the number and value of the summaries possible.
The proper size of the sample to be used to cover the areas lying
within the service contours of WKGP depends on three considerations:
the statistical accuracy desired, the acceptability of findings to advertisers
and their agencies, and the cost.
The Iowa surveys are based on slightly more than nine thousand
interviews completed out of, presumably, some eleven thousand to
twelve thousand five hundred attempted. This is a far larger number
than is needed to survey Iowa as a state, yet county-by-county data for
the ninety-nine counties of the state are subject to a considerable margin
of error. The study proposed would have the great advantage of cover-
ing a smaller number of geographical units. Serious consideration
should be given, however, to the need for individual county, or part-of-
county, data; for any merging of any two geographical units will increase
74
MARKETING RESEARCH
the accuracy of a given sample size or reduce total sample size and cost
for a specified degree of accuracy.
Tentatively, it is suggested that a primary sample of approximately
eight thousand to ten thousand be considered.
The technique used would be personal interviews supplemented by
some form of roster, aided-recall, or diary technique. Pre-coded ques-
tionnaires and punch-card tabulation methods would be used both for
accuracy and economy.
The elements of cost for a study of the type and scope proposed are
estimated as follows:
COST OF SPECIAL STUDY OF RADIO LISTENERS
Co
st
Activity
Minimum
Maximum
Preparing of questionnaires, pretesting in field, etc. . . .
Printing of interview forms
Personal interviews at $1.25 an hour (or less) plus travel
Hiring, training, and supervising interviewers
Sample design and other statistical work
$ 500
300
12,000
600
500
$ 750
400
16,000
800
500
Editing of returns
Machine tabulations . . ...
Planning, administration, analysis, and report prepara-
tion: the Director's salary .
250
400
2,000
400
500
2,000
Printing the report *
Secretarial help and miscellaneous
350
500
Total
$16,900
$21,850
* Assumed done by WKGP in format and quantity desired.
ACTION NEEDED
Among considerations raised by this proposal were these:
1 . Would the eliminations, revisions, and additions of data sug-
gested make it more usable for WKGP? What other
changes should be considered?
2. In what sense are identification data by-products of the sur-
vey, and how do they "increase the number and value of the
summaries possible"?
3. How might the number of sample sub-units be reduced
without decreasing the value of the findings significantly?
CASES IN PLANNING 75
4. How could "circulation" data be applied to specific time
periods or programs when selling, or buying, time?
5. Would the study proposed solve the problems the station
manager had in mind, and was it likely to be worth the cost?
CASE 2 PROPOSED CONSUMER MARKET
SURVEY
BACKGROUND
The annual sales of the Seldon Department Store were ap-
proximately $400,000, although it was located in an agricultural
valley section in a community of less than 3,000. In a sense,
Seldon, which was on Main Street, combined three stores. On
one side was a former five-and-ten-cent store, or variety store,
and the basement and first floor sales area still was used to sell
general merchandise and limited price lines (indeed, the famil-
iar red banner of the former store, slightly altered, remained on
the store exterior). The middle section of the store was nar-
rower and was used to display and sell women's ready-to-wear.
At the other side was a section containing office space which
also was used to display and sell major appliances and floor cov-
erings. These had been three separate stores at one time, and
had been joined by the expedient of constructing one small
doorway in each separating wall. Furniture was both stored
and displayed upstairs in rooms which once had been apart-
ments and had never been remodeled. Furniture also was
stored, and could be displayed, on the second floor of a building
down the street.
In its transition to a department store status, the store had
become badly crowded for space. Storage space was limited
and poorly located; store walls were jammed with merchandise
right up to the high ceilings; aisles were narrow.
The enterprise was steadily expanding its sales volume and
profits. The proprietor had sufficient capital and owned land
in the rear for a considerable expansion of sales area. Estimates
had been prepared of the increase in volume which might be
obtained in various lilies if these lines were broadened and the
76 MARKETING RESEARCH
EXHIBIT 1
Questionnaire for Consumer Survey, Department Store
Town
Other towns
A
B
C
D
(state which)
1. How far do you live from each of these
towns? Number of miles . .
2. Approximately how many times have you
shopped in each of these towns so far this
year? Number of times
3. In which of these towns during the past 5
years have you purchased the items listed be-
low? (For each item check the town, or
towns, in which you made purchases. )
Furniture for:
Dining room
Bedroom
Living room
Carpets
Women's dresses, priced
at less than $6.00
at $6.00 to $8.99
at $9. 00 or more
Children's clothing
Men's wear
Appliances
Radio or television
Stove or washing machine
Other
dur
ing the past 5
4. Which of the items listed above have you purchased
5. In which town do you prefer to buy your dresses? (Name of town)
Why?
*
CASES IN PLANNING 77
6. In which town do you prefer to shop for furniture?
(Name of town) Why?
7. Perhaps you would prefer to shop in a different town if the merchants
offered better assortments, different store hours, better parking, or some
other conditions. To indicate this, please complete the following state-
ments :
A. I would prefer to buy my clothing in (town) #
B. I would prefer to shop for furniture in (town) ^
if
C. For most shopping other than for food I now prefer to shop in
( town ) because
For statistical purposes only : please tell us the following:
I am married ; single ; other Check which one.
My husband's age is: Under 25 ; 25 to 50 ; over 50
My husband's occupation is (Your occupation, if
not marned.)
The number of persons in our household is:
Males under 2 1 2 1 and over
Females under 2 1 2 1 and over
For sending the gift of to me, please use this address :
Name
Address .
78 MARKETING RESEARCH
sales area increased. Various ideas for the expansion and re-
modeling of the store had been developed and a few of these
had been incorporated into architectural drawings and estimates
of cost.
The proprietor was a man of independent mind who had built
the business from scratch. He liked to hire consultants, but
often failed to follow their advice, being content with the stimu-
lation afforded to his own thinking and the tests provided for
his own developing plans. On this occasion he secured the serv-
ices of consultants from a nearby city. The consultants were
to spend one day in the store discussing plans and records and
inspecting the store and then submit recommendations for
changes.
Among the suggestions offered was that the planned expan-
sion, which seemed desirable, should not be made until after a
consumer survey was conducted.
THE PROBLEM
A specific question the proprietor had to decide was whether
or not to conduct some such survey as was recommended.
DATA AVAILABLE
The Seldon Department Store had no important competition
for its lines from stores in the community. Less than twenty
miles to the south, however, in a community with a population
of 1 3,000, there was a sizable shopping district with many mod-
ern stores including several furniture stores, women's clothing
stores, and chain stores operated by a mail order house and other
large organizations. Some thirty miles eastward was a town of
40,000 population, and some forty miles northward was the
state capital, with a population of well over 100,000. Evidence
suggested that for shopping goods the major competition was
with stores in the nearest of those towns and, especially, from
the large stores in the capital city.
A major share of the floor space to be added, if the proprietor's
basic plan was followed, would be devoted to furniture, cloth-
ing, and appliances. A major share of the increase in volume
of sales planned would also come from sales of these lines. It
CASES IN PLANNING 79
was in regard to this aspect of the contemplated expansion that
the consultants suggested it might be wise to take out some in-
surance in the form of a little consumer research. Appended
to the report were rough plans for such a survey. The survey
itself would be made by store personnel. Plans included the
questionnaire on pp. 76-77, which was to be mailed out to a
sample of women in the area.
Suggestions in some detail were given concerning the method
of obtaining the proper sample and of checking results. It also
was suggested that the company name not be used and that a
box in the capital city be rented to receive returns; that women
answering the questionnaire be sent a small gift or some thirty
or forty cents' worth of stamps, and that if a second mailing was
needed, fifteen or twenty cents' worth of stamps might be en-
closed with a promise of the remainder on reply; and that the
many open-end questions suggested might prove difficult to
tabulate, especially for store personnel, even though they were
useful question types for exploration. The entire project, it was
suggested, might be conducted by the store for an out-of-pocket
cost of from $200 to $400,*a fraction of 1 per cent of the contem-
plated investment in expanded facilities.
ACTION NEEDED
1. Should the proprietor of the Seldon Department Store ac-
cept the proposal to conduct a consumer survey?
2. Would the proposed questionnaire produce the most useful
data or should revisions in plans be made?
CASE 3 USE OF INTERNAL DATA FOR
STUDYING MAGAZINE MERCHANDISING
EFFECTIVENESS
BACKGROUND
The Summit News Company secured a large volume of mag-
azine sales through drugstores, although newsstand sales were
large and other outlets were used to distribute magazines. The
company was convinced that magazines were highly profitable
80 MARKETING RESEARCH
merchandise for these drugstores. For example, a study of a
small group of drugstores in one region of the country had
shown (1) that magazine sales averaged more than 5 per cent
of total store sales, (2) that sales of magazines per square foot
of space used averaged $140 annually, compared to $69 for the
stores' sale of all items, and (3) that gross margin per square foot
of space used was $35 annually for magazines, compared to $20
for all items.
Many drugstore operators, nonetheless, considered magazines
as little better than a necessary evil justified only by customer
convenience.
Poor dealer relationships and a growing interest in means of
increasing magazine volume and profitability in retail outlets
led the company to consider a long-range research program.
THE PROBLEM
As originally visualized by the company, the general objective
of the research program would be to find the most efficient
methods of merchandising magazines in drugstore outlets.
This would be done by retail studies and would include a com-
plete appraisal of dealers' attitudes toward handling magazines.
QUESTIONS TO BE ANSWERED
Among the more specific questions the managers of the com-
pany had in mind were:
1. What kind of display racks should be used in drugstores in
various volume classifications?
2. What sizes of racks should be used in each?
3. What determines the best location for the rack, or racks,
used?
4. How many titles should be carried by each store?
5. How should the titles be selected?
6. What determines the best display or positioning of the
titles carried?
In addition, the company thought that marketing research
should be used (1) to develop standard operating statistics for
CASES IN PLANNING 81
good magazine departments and (2) to study magazine con-
sumer buying habits.
One of the specific problems which arose later concerned the
type and amount of research, if any, to be conducted, based
entirely or primarily on existing records. Indeed, this became
the key problem in the consideration of using research.
DATA AVAILABLE AND THE PROGRAM PROPOSED
Magazines for sale over-the-counter in single copies were dis-
tributed by the Summit News Company under a consignment
system. This permitted each publisher to decide how many
copies of each issue should be put on the stands in each market.
The local distributor allocated the number set by the publisher
among individual dealers. From time to time during the life
of each issue, routemen redistributed the copies among dealers
as they ran short or were found in oversupply.
Dealers, including drugstore owners, accepted this arrange-
ment, under which they had no control over their quota, be-
cause they were permitted to return all unsold copies for full
credit. Each dealer paid the distributor for the copies he actu-
ally sold (as measured by the difference between his "draw" and
his return). Similarly, the local distributor paid the national
distributor, and the national distributor paid the publisher, for
the difference between the number of copies sent out and the
returns.
A private consultant was hired to draft a detailed proposal
for a long-range research program, together with budget esti-
mates and a time schedule.
In brief, the proposals made envisioned three broad types of
research. The first type would rely primarily on available rec-
ords kept by the distributor. The second type would require
the keeping of special records by drugstores or the actual obser-
vation of sales persons and customers at the point of sale. This
might include both direct observation experimental techniques
and the use of the techniques of cost allocation. The third type
of research would involve interviews with consumers and with
dealers.
In more detail, these possible projects were suggested:
82 MARKETING RESEARCH
PROJECT 1
SUGGESTED STATISTICAL ANALYSIS
OF THE SALE OF MAGAZINES IN DRUGSTORES
I. Objective: To test through correlation analysis the effectiveness
of methods used to sell magazines in drugstores.
II. Types of questions to be answered.
1. What effects do differences in methods of displaying, promot-
ing, and selling have on volume, cost, and profit in the market-
ing of magazines by drugstores?
2. How do the various types of stores, when classified by such
factors as size, location, organization, and lines sold, differ in
their effectiveness as distributors of magazines? Why?
3. How widely, and by what means, can all stores imitate the most
effective ones?
4. Why do some individual stores excel other stores of the same
type in selling magazines?
5. What types of racks are best suited to what types of stores?
6. What assortments of magazines are proper for different types
of stores?
7. What types of stores are most profitable to the distributor?
8. What stores serve the consumer most effectively?
III. Methods of study.
1. A detailed study of distributor records for a sample group of
stores.
2. A restricted survey of at least part of the stores in the sample
to obtain classification data.
3. Correlation analysis of the data gathered.
IV. Sample to be studied.
1. A sample representative of all drugstores served in the ...
metropolitan area as a whole.
2. A sample representative of appropriate subdivisions of the area.
3. A sample large enough to permit selection from it of sub-
samples and test groups for subsequent field surveys and ex-
periments.
V. Data to be gathered from distributors' records for each store in the
sample.
1 . Magazine assortments handled.
a) Full assortment.
CASES IN PLANNING 83
b) Minimum assortment.
c) Special assortments.
2. Sales volume, in dollars and units.
a ) Total for all magazines.
b ) Totals for particular assortments or classes.
3. Gross margin, in dollars and as a percentage of sales.
a) Total for all magazines.
b ) Totals for particular assortments or classes.
4. Sales over the life of any selected issue, or issues, for selected
titles.
VI. Other data from distributors' records.
1 . Total sales margin in dollars and as a percentage of sales, total
expenses, and expense breakdowns.
2. Prices of individual magazines and margins on each in dollars
and as a percentage of sales.
3. Draws, sales, and returns for individual magazines and classes
of magazines.
4. Percentage of total draws, sales, and returns represented by
individual magazines and classes of magazines.
VII. Data to be gathered from distributors' records or from maps when
possible, otherwise by a physical inspection of stores and, possibly,
by use of a simple questionnaire.
1 . Classification of each store by location.
a) By route.
b ) By counties, city divisions, or other appropriate areas.
c ) By downtown, shopping centers, neighborhood, and so on.
2. Classification by ownership organization.
a) Chain.
b) Independent.
c) Voluntary chain.
3. Classification by service rendered to the store.
a ) By frequency of calls by routemen.
b ) By provision, acceptance, and use of display aids.
4. Classification by display facilities and methods.
a) Type of racks.
b) Number of racks.
c) Size of racks.
d ) Location of racks .
e ) Positioning of titles on racks.
84 MARKETING RESEARCH
/) Ratio of magazines sold to magazines displayed,
g) Scheduling of displays.
5. Classification by size of stores' sales of all goods.
6. Classification by lines of goods handled other than magazines.
7. Classification by ratio of amount of sales to space used.
a) Total space.
b) Space used for magazines, amount and as percentage of
total.
8. Classification by miscellaneous factors.
a) Hours open.
b ) Type of trade (transient, neighborhood, and so on ) .
c ) Incomes of patrons.
d) Patronage by special groups (occupation, race, religion, na-
tionality, and so forth ) .
e ) Value of magazine stock carried.
/) Methods of handling small-circulation magazines.
VIII. Examples of questions for which answers should be sought and
the breakdowns of data that might be used in the analysis.
1. What variation is there in the annual volume of magazine
sales of these outlets?
a) Number of outlets and percentage of total selling, say, up
to $1,000, $1,001 to $10,000, and so on.
b) Average and median size for all stores and for each size
group.
c) Reasons for the variations.
2. What proportion of total distributor volume comes from each
size group?
3. What is the relation of assortments carried to the volume of
sales?
4. How many and what sorts of stores are not carrying proper
assortments?
5. What is the relation between sales facilities and volume of
sales?
a) Sales volume by number, type, size, and location of racks.
b ) Sales volume by kinds of display policy followed .
6. How many and what sorts of stores are not provided with
proper racks?
7. What types of stores secure the highest (and lowest) sales per
square foot of space devoted to magazines?
8. How does the rate of sale over the life of an issue vary as among
stores?
CASES IN PLANNING 85
9. What significant differences are there in volume and assort-
ment of sale at different types of location?
10. What relationship is there between magazine assortments car-
ried and types of stores?
PROJECT 2
EXPERIMENTAL TESTING OF SALES VARIABLES
Objective: To hold variables constant while changing singly the various
factors to be tested to discover the effect on sales of different racks,
different size racks, different locations of racks in stores, different as-
sortments, different display methods, and different sales methods.
PROJECT 3
DISTRIBUTION COST STUDY OF OUTLETS
Objective: To discover the relative profitability of different types and
sizes of stores and the functional costs of handling accounts and to
establish standards of expense and profitability by assortments and
the unit or volume sales of titles.
PROJECT 4
CONSUMER INTERVIEWS
Objective: To discover the numbers and types of persons who buy maga-
zines at stores, how often, for what reasons, with what store prefer-
ences, and so on.
PROJECT 5
SURVEY OF DEALER ATTITUDES
Objective: To determine dealer attitudes toward handling maga-
zines, distributor organization, methods, personnel, services, display
aids, expansion, and so forth.
PROJECT 6
COMMODITY COST STUDY IN DRUGSTORES
Objective: To discover the profitability to stores of sales of different
types of magazines.
86 MARKETING RESEARCH
ACTION RECOMMENDED
The company had to reach a decision regarding the desirabil-
ity of launching the research program. A good deal of differ-
ence of opinion existed as to how worth while Project 1 in
particular might be to the company.
Operation on consignment is possible only because the dis-
tributor keeps detailed and up-to-date records on each dealer.
As a matter of routine these records reveal such facts as these
for each important magazine individually and for each group
of the less important ones: the number received, sold, and re-
turned for each individual issue; the price paid for each issue;
the total amount paid for all copies of each issue; and the rate
at which each issue has sold during its life.
The officials most closely connected with these operating data
felt they already utilized them to the fullest extent feasible.
They pointed out that declining sales at particular stores or on
particular routes or for particular titles could be detected read-
ily. Since the prices consumers paid for magazines were set by
the publishers, and there was no problem of retailers' selling at
cut prices, gross margins for titles, per issue or over a period of
time, and for dealers were easily obtained.
It was argued, therefore, that although researchers would
have to familiarize themselves with the types of records kept
and the methods of operation, this was simply "education" of
persons unfamiliar with the business. This "education" should
be minimized to save money. What was needed was not to
play statistical tricks with figures already available, but to gather
new facts important to better merchandising of magazines.
Other officials felt that statistical analysis was the proper first
step in the research program. Not only would it be worth while
in itself, but it also would be helpful both in determining the
specific experimental studies to make and also in establishing
them.
ACTION NEEDED
1. What are the merits and defects of the general proposal and
the tentative plan for a statistical analysis of records?
CASES IN PLANNING 87
2. Are the methods proposed well adapted to the specific pur-
poses stated?
CASE 4 METHODS FOR STUDY OF
INSTALLMENT CREDIT
BACKGROUND
A public-spirited, influential businessman had become con-
cerned after World War II with the problems of installment
credit encountered in his field of business and in national affairs.
Installment credit, he believed, had become far more than a
matter of purely monetary policy for both business and govern-
ment. Yet data on the economic and social aspects of install-
ment credit were spotty and few. Being accustomed to using
marketing research, he secured the services of a consultant to
formulate proposals for a broad study, hoping that the new
American Foundation, Inc., or some other foundation or busi-
ness group, might be interested at some later time in conducting
it.
Three alternate techniques for conducting such a study were
proposed.
A. SURVEY METHOD
The first, and relatively most simple, method proposed was
a survey of consumers accompanied by an inventory of their con-
sumer durables. The main purpose of this operation would be
to disclose the extent and nature of durable goods purchased on
the installment plan possessed by families of various types.
Data to be Gathered:
1. Characteristics of respondent families, by age, sex, income, occupa-
tion, education, and nationality of members, home value or rental
and year obtained, year married.
2. An inventory of a specified list of durable goods supplemented by
information on price, make, model, age, month and year purchased;
new or used purchase; make, model, age and value of trade-ins, if
any; how bought; period and payment rate if installment purchase.
88 MARKETING RESEARCH
3. Installment buying, other than for specified durables, as to items
bought, last purchase dates, period and amount of installment pay-
ments.
4. Income(s) for years in which durables purchased and occupation (s),
if different from present.
5. Finances: mortgage, insurance, other assets and obligations.
6. Future plans for purchase of the specified durables.
7. Attitudes toward durable goods ownership and buying on time;
related domestic or housework problems and other needs, entertain-
ment and social use of durables.
Objectives: Problems for Analysis:
1. Influence of individual and family characteristics on the installment
purchase of durable goods (by individual characteristics, multiple
correlation, and identity with established socioeconomic groups) .
2. Number of installment buyers and nonbuyers, total and by types of
durables.
3. Number and characteristics of habitual vs. infrequent installment
buying families.
4. Kinds and total of durables owned.
5. Percentage in units and dollars of financed durables to total dura-
bles, original and depreciated prices.
6. Rate of accumulation of durables owned and relation to financing
used.
7. Used durables purchased, type, total, percentage of all durables, and
differences in financing new and used goods.
8. Types, ages, and values of trade-ins and relation to items bought and
type of financing.
9. The percentage of weekly (or monthly) income "mortgaged"; the
peak, low point, and average weekly obligations during the year, in-
stallment debts, and all debts.
10. The value of owned durables, and financed durables, as percentage
of total assets and ratio to liquid assets.
11. Percentage of total income spent on durables, current and past
years, and by types of durables and financing.
12. Ownership of durables, total and by types, by regions and city sizes.
1 3. The relation of installment durables to standards of living, family
events, attitudes toward installment buying, social and entertain-
ment uses of durables.
14. Future expectations by types of durables and in total, in relation to
present holdings, financing methods, and income.
CASES IN PLANNING 89
Limitations:
Respondent cooperation will be difficult to obtain, especially among
some groups, although payment for cooperation should minimize the
problem. Inaccuracies caused by memory loss and bias, particularly
on dollar amounts, can be expected; but these could be reduced some-
what by various techniques and by using dealer records to check prices
and values. Although the data will be better than those from past
studies, life-cycle data will be incomplete. Probable cost would ex-
ceed $ 50,000. A pretest probably should be made first on a local scale
to iron out interviewing and statistical problems and to arrive at a
better idea of the type of results that could be obtained by using this
method.
B. CONTINUOUS CONSUMER PANEL METHOD
The second method proposed was the use of a continuous
consumer panel, perhaps as a special client of one of the panels
now operating. Although the most costly and long-range type
of study, it would seem to promise the best and fullest results.
Data to be Gathered:
I and 2. Same as for survey method.
3. Weekly mailed records covering changes in family status and in-
come, purchases and commitments for durables in some detail, pay-
ments on durables or loans, family members shopping for durables,
types of stores at which purchased, disposal of durables other than
by trade-ins, and car mileage.
4. Periodic reports on buying expectations.
5. Special data from follow-up inventories or interviews to explore be-
havior patterns revealed by other data.
Objectives: Problems for Analysis:
1 to 14. Same as for survey method.
1 5 . Expectations vs .later actualities .
16. Types of stores at which durables were purchased.
17. Shopping behavior of families buying durables.
18. Variations in installment payments by years, seasons, types of goods,
family-cycle stages, and income changes.
19. Replacement and use cycles for durables and relation to financing
means and accumulated investment in durables.
VU MARKETING RESEARCH
20. Net changes in depreciated value of durables, as per cent of income
earned, expended, and committed, and of savings.
21 . Measurements of consumer holdings other than by dollars.
22. Possession of durables as deterring or facilitating purchases of other
durables, in total and by types.
23. Delinquencies of various kinds, their causes, and action taken.
24. Determination of appropriate time periods for analysis of durable
goods purchases and use patterns.
25. Relation of installment buying to other choices of families and the
determinants of choice.
Limitations:
Such a study would have to be continued for several years to be fruitful,
and, ideally, should cover a full business cycle. Annual cost might ex-
ceed $100,000, in addition to cost of the opening survey. Time and
cost are the principal limitations, although turnover of members and
other problems of panel operations would also be encountered.
C. CASE HISTORIES METHOD
The third method proposed could be used alone or in com-
bination with Method 1 or Method 2. It contemplates the
gathering through long, repeated interviews with husbands and
wives of as complete case histories as possible on the purchase,
use, and financing of automobiles (or any other single durable)
during the lives of family members. The data to be gathered
and the objectives, or problems for analysis, would be similar
to those outlined for the first two methods.
ACTION NEEDED
1 . If the installment credit project is undertaken, which of these
methods should be preferred, and why?
2. To what extent would each method help to answer such
questions as these:
a) How is installment credit employed during the life cycle
of individuals and families?
b) In what ways do the purchases of installment durables
relate to specific needs of families, to the peaks and valleys
of these needs, and to the flow of income as contrasted to
CASES IN PLANNING 91
the expectations, or realities, of these needs and the bur-
den of postponing consumptiou?
c) What is the extent and importance of consumer holdings
of installment durables as savings and investments? How
do families accumulate such holdings? How is their ac-
quisition related to other assets, or savings?
d) What is the function of the depreciated (trade-in) value
of installment durables in dividing total stocks, in sustain-
ing demand over time and during business cycle changes,
and in accumulating personal and national wealth?
e) How does installment credit fit into the management of
family finances, and with what benefits and dangers?
CASE 5 PRETESTING A QUESTIONNAIRE
SMALL ACCOUNTS PROBLEM
The Langdale Department Store, with a volume of sales of
several tens of millions of dollars and with approximately 300,-
000 charge account customers, was disturbed about the large
number of small accounts. A sample taken of accounts by size
had shown that nearly half of them averaged less than $1 a week.
A customer survey was authorized to ascertain the reasons for
the existence of so many small accounts. It was expected that
this information might lead to remedial action.
HYPOTHESES USED
Because the store knew very little about its customers except
for their number, their total cash and credit expenditures as a
group, and their addresses, the basic problem appeared to be
one of characterizing small (and large) customers. Ten possi-
ble reasons for customers spending only limited amounts at the
Langdale Department Store were formulated. These were:
1 . Income. Low income families may make up the bulk of small ac-
counts.
2. Family size, composition, type. Smaller families may buy less. Sex,
age, nationality, and home ownership might influence size of pur-
chases.
92
MARKETING RESEARCH
EXHIBIT 1
Study of Department Store Accounts
About how long have you lived in (city)? Years ; Life ]
Which member of your family, in this household, does the most shopping
in downtown department stores? Verbatim Ans. _____ - .
Also classify the answer:
Self
D.K.
All n; More than half rh Shares
Male d
Other Q Female []
3. What other family members, in this household, also bought in any of these
downtown department stores in the past year? *
Self Q Husband Q; Son Q; Daughter Q Others DDCD D - K - D
4. Have you, or has any other tamily member in this housenola, ordered any
goods from these stores by telephone in the past year? Yes Q No [jj
D.K.Q
If yes, ask: At which store, or stores? Enter store initials - D.K. Q
5. A. At which of the five stores, if any, do you or other members of your
family, in this household, have a charge or credit account?
G. Q; L O Sa. Q Sc. Q T. Q; None Q; D.K. [J
B. About how many years ago were each of these accounts opened?
G. []; L. Q Sa. [j; Sc. PJ; T. Q- ( Enter vears or D -K.)
C. How many members of the family, in this household, make charges on
these accounts? Number _ ; D.K. ]
D. Have any friends, or relatives outside the household, used any of these
accounts in the past year? Yes Qj No Q; D.K. F]
If yes, ask: Which accounts? G. Q L. Q SaT Q; Sc. Q; T. Q
D.K.Q
6. Are there any charge accounts in these 5 stores which the entire family in
this household has definitely stopped using? Yes Q; No Q; D.K. fl
If yes, ask: In which stores? G. Q L. Q Sa. p ; Sc. Q; T. p|; D K. [~|
7. Including both cash and credit purchases, at which of these 5 stores would
you estimate your family had done the most buying in the past year?
Most? (1) _ ; Next most? (2) _ ; Next most? (3) _ ; Next most? (4) _ ;
Least? (5) _ ;D,K.Q
8. In the past year has your family bought more for cash than for credit at:
Greens?
More
Less
About
same
D.K.
Langdale?
Sardon?
Scier?
Taskers?
* Interviewer instructions called for the respondent to be handed a flash
eard with an alphabetical list of five stores before asking question No. 3.
CASES IN PLANNING
93
9. A. Do you feel personally that a charge account offers you any special con-
venience or advantage? Yes Q; No FJ ; D.K. ]
If yes, ask: What conveniences or advantages?
B. Do you feel personally that a charge account causes you any special in-
convenience or disadvantage? Yes Q; No Q]; D.K. Q
If yes, ask: What inconveniences or disadvantages?
10. Do you pay your charge accounts In person n? By check Q? By money
order ["""|? or by some other means |""1? (Describe) D.K. ]
11. A. Do you read any (city) newspapers? Yes ]; No ]; D.K. []]
If yes, ask:
How often: Which ones?
Sunday only?
Weekdays only?
Daily including Sunday?
B. Do you read department store ads before going downtown shopping?
D.K.
I; All
Bugle
News
Star
D.K.
If yes, ask: which store ads? G ]; L Q; Sa. Q; Sc. ]; T.
C. Which store do you think:
Has the most reliable ads
Has the least reliable ads
Offers the most good bargains in its ads
Offers the fewest good bargains in its ads
12. Which of the 5 department stores do you think:
Has the most courteous and helpful sales persons? ,
Has the least courteous and helpful sales persons? .
Has the most pleasing appearance inside?
Has the least pleasing appearance inside?
Gives you the most value for your money?
Gives you the least value for your money?
Offers you the best all-around services?
1 3
Offers you the worst services in general?
D.K.
D.K.
D.K.
D.K.
D.K.
D.K.
D.K.
D.K.
D.K.
D.K.
D.K.
D.K.
A. In which store do you find it easiest in general to locate and inspect the
goods you wish to see?
G. p ; L. n; Sa. n; Sc. n; T. n; D.K. n
B. In which store do you find it hardest in general tolocate and inspect the
goods you wish to see?
G. Q L. Q Sa. [3 Sc. Q T. Q D.K. p
14. Are there any certain items, or types of goods, which you almost always
Greens? Yes
;No
;D.K.
Langdale? Yes
;No
~
;D.K.
Sardon? Yes
_
;No
~"
;D.K.
Scier? Yes
***
;No
B " B
;D.K.
Taskers? Yes
i
;No
-
;D.K.
If yes: Ask: Which?.
If yes: Ask: Which?.
If yes: Ask: Which?.
If yes: Ask: Which?.
If yes: Ask: Which?.
94 MARKETING RESEARCH
15. Are any members of your family employed in one of these 5 department
stores:
Yes Q No Q]; N.A. [] If yes: Ask: Which store or stores?
N.A. [jj
16. Which one of these five department stores is your favorite?
G. Q L. E; Sa. Q; Sc. n; T. H; D.K. Q
Why do you like this one (these) oest?
IDENTITY (Be sure to complete this section)
Name of respondent
Address
Estimated age of respondent: 24 or under Q; 25-34 ]; 35-44 Q; 45-64Q;
over 64 Q]
ASK: How many members in your family in this household under 19 ;
19 or over ; Total :
Would you mind telling me in which of these groups the family income
belongs:
$30 or under per week Q; $31 to $40 Q; $41 to $50 Q; $51 to $70 [];
$71 to $100 Q; over $100 per week O;N.A. D
Interviewer's name
Date of interview
3. Geographical location. Distance (and travel time) might create re-
sistance to purchases in the downtown city area and reduce fre-
quency of shopping.
4. Shopping habits. The greater the number of charge accounts, or
stores patronized, or shopped, the smaller may be expenditures at
this store.
5. Retail expenditures. Customers buying mostly for cash, patronizing
mostly small stores, chains, or secondary shopping centers, and those
who spend relatively little in department stores may have small ac-
counts.
6. Trends in charge buying. Small accounts for any given year might
be small because few if any big-ticket items happened to be pur-
chased that year by many customers.
7. Days and hours for shopping. Some customers may have difficulty
in shopping during regular store hours.
8. Newspaper reading. Low exposure to newspaper advertising by the
store may result in small accounts.
9. Specialized buying. Persons who view the store favorably only for
a few types of goods might be small buyers.
CASES IN PLANNING 95
10. Attitudes. Unfavorable attitudes toward store prices, quality, or
services might restrict purchases.
THE QUESTIONNAIRE DESIGNED
Based on these hypotheses and the results of preliminary in-
vestigations, both internal and external, the questionnaire on
pp. 92-94 was designed.
THE PRETEST RESULTS 1
Interviewers were sent into widely scattered sections of the
city. Names and addresses of store customers in the designated
areas were provided, although some interviews were made with
noncustomers.
For the 198 interviews completed, the interview time aver-
aged 25 minutes, the median time was 25 minutes, and the range
was from 12 to 35 minutes. Only 18 interviews lasted 30 min-
utes or more; two were not completed because of respondent
fatigue.
Question 1 : The opening question worked very well.
Questions 2 and 3 : These were troublesome. Interviewers re-
ported and the questionnaires revealed difficulty in classifying
answers. Respondents seemed to get confused. Forty re-
spondents were reported as answering they did most of the
shopping and fifty-two as saying all, but nineteen of the latter
then reported other family buyers in Question 3. Thirty-three
families were reported as having one buyer, one hundred ten as
having two, and forty-seven as having three or more. How
could classification be improved? Was a cross-check needed?
Question 4: Nearly half of respondents reported telephone
buying and (even more unexpectedly) the great majority of
these reported telephone buying in two, three, four, and even all
of the five stores listed. Numerous persons volunteered reasons
for such buying.
1 For pretesting techniques, see A. B. Blankenship, op. tit., pp. 82-97; L. O.
Brown, op. cit., pp. 483-85; and D. }. Luck and H. G. Wales, op. cit., pp. 170-
72. Only questionnaire aspects are reported for this case.
96 MARKETING RESEARCH
Question 5: A few respondents indicated that more than one
family member had an account at a specific store. One inter-
viewer felt that answers to 5B were vague and another pointed
out that 5C was a generality.
Question 6: Respondent answers included: No, 158; Yes, 20;
DK, 20. An interviewer, several of whose respondents had an-
swered Yes, suggested that a Why question be added.
Question 7: Difficulties in ranking respondents who pur-
chased at all five stores, or at three or four of them, were re-
ported, but few respondents failed to answer.
Question 8: The first two columns ("More" and "Less")
seemed confusing; it was suggested that questions 5A and 7
really covered the same ground; a "None" column appeared
desirable; and it was suggested that the question be restricted
to stores at which accounts had been reported.
Question 9: More than five times as many respondents an-
swered No to 9B as answered Yes. Don't Knows ran heavy.
Nearly all persons answered Yes to 9A and a majority of these
gave one reason.
Question 10: Very few respondents reported other means of
paying accounts; many reported using more than one.
Question 1 1 : The word "reliable" bothered both interviewers
and respondents, "truthful" seeming to be the most common
interpretation by respondents. Don't Know answers ran al-
most to 50 per cent and were twice as high for "least reliable"
and "fewest good bargains" as for alternate questions. Inter-
viewers were troubled at reporting highly infrequent readers of
newspapers or ads under Yes.
Question 12: Most respondents were reluctant to answer in
terms of "worst" or "least." The question was boring and
caused respondents to show irritation. A great many respond-
ents were of the opinion that all stores were equal in terms of
questions asked.
Question 13: No particular problems. The question was
raised as to why this question was separated from Question 12
and why answers were to be recorded in different fashion.
CASES IN PLANNING 97
Question 14: No apparent problems.
Question 15: Numerous respondents reported no members
now employed but that one or more had worked in one of the
stores at some time.
Question 1 6: Answers willingly given. Considerable diversity
in reasons reported. Two interviewers reported poor inter-
viewee reactions because question was repetitious.
Identity Section: Apparently satisfactory. Refusals to an-
swer income question were few but were more numerous among
those interviewers reading the income groups than those hand-
ing the respondent a typed card.
THE REVISED QUESTIONNAIRE
Following the pretest the questionnaire was revised. The
new questionnaire proposed was as given on pp. 98-100.
ACTION NEEDED
A final questionnaire conference was scheduled after revisions
had been made. Among the questions which the research di-
rector wished to raise with staff members (after they had in-
dividually gone over both the pretest results and the revised
questionnaire) were these:
1. Is another pretest necessary because of the number and type
of changes made?
2. Is the new questionnaire too long? If so, how might it be
shortened: by draftsmanship? by rephrasings? by omis-
sions?
3. Can the sequence of questions be improved?
4. Are the changes justified? Are they the best that can be
made?
5. Can interviewers read the questions easily and tabulate an-
swers quickly and accurately? Is phrasing and spacing satis-
factory?
6. Are there any subjects omitted which should be included?
Are present additions justified?
98
MARKETING RESEARCH
EXHIBIT 2
A Study of Department Store Accounts
1. About how long have you lived in (city)? YEARS ["""[; LIFE fl? D - K - F]
2. Which member of your family, in this house, does the most shopping in
downtown department stores? RESPONDENT [""[; OTHER (IDENTIFY) _
DOES ALL ; DOES MOST ); SHARES EQUALLY j D - K -
3. What other family members, in this house, alsobought in any of these 5
downtown department stores in the past year? * IDENTIFY EACH MEMBER:
_ NONE ]; D.K. Q]
4. A Have you, or has any other family member in this house, ordered any
goods by telephone in the past year? YES Q; NO Q]; D.K. Q]
G
L
Sa
Sc
T
None
DK
B If YES, ASK: At which stores?
C Do you buy all , more than half
, or less than half of your de-
partment store purchases by tele-
phone? D.K. Q]
X
X
X
X
X
X
X
5. A At which of the 5 stores, if any, do
you or other family members, in this
house, have a charge or credit account?
X
X
X
X
X
B About how many years ago were each
of these accounts opened? ENTER
YEARS
C How many members of the family, in
this house, have made charges on these
accounts in the past year? TOTAL
NO D.K. ["""]
X
X
X
X
D Have any friends, or relatives outside
the house, used these accounts in the
past year?
YES Q; NO Q D.K. Q]
X
X
X
X
x x
If YES, ASK: Which accounts?
X
6. Are your charge accounts paid in person
Q? by check Q? or by money order [~|?
D.K.[j]
X
X
X
X
X
X
7. Including cash and credit purchases, at
which of these stores has your family done
the most buying in the past year? Next
most? etc. RANK ALL 5.
* Flash card with five store names shown.
CASES IN PLANNING
8. In the past year has your family bought more for cash or credit at:
99
All
for
cash
More
for
cash
4
About
same
Less
for
cash
No
cash
No.
Pur-
chase
D.K.
Greens?
Langdale?
Sardon?
Scier?
Taskers?
10.
11.
9. A Before going shopping do you nearly always |"""|, sometimes | |, or very
rarely ], read department store ads? D.K. Q] NEVER n
B In terms of getting wanted information about goods, feeling sure there
will be a stock left when you get there, and that the goods will really be
"as advertised," which store's ads do you feel are most reliable ?
D.K. ] least reliable ? D.K. Q] ALL SAME Q
C Which store offers the most good bargains in its ads ?
DtK - EH * ne fe wes t good bargains ? D.K. [""]; ALL SAME Q]
May I also have your personal opinion on a few matters concerning the ease,
pleasure, or efficiency with which you can buy at different stores?
How would you rank these 5 stores in relation to :
A The courtesy and helpfulness of sales per-
sons?
B The appearance of the interior of the store?
C The ease of locating and inspecting the goods
you wish to see?
D The services (such as, adjustments in returns,
delivery, lunchrooms, and restrooms) given
you?
E Being able to get the most value for your
money?
G
L
Sa
Sc
T
D.K.
Are there any items, or types of goods, which you almost always would go
to get at:
Greens? YES ; NO ; D.K IF G
Langdale? YES ; NO ; D.K YES, L
Sardon? YES ;NO ;D.K ASK: Sa
Scier? YES ;NO ;D.K WHICH? Sc
Taskers? YES ; NO : D.K T
12. Are there any members of your family now employed, or recently employed,
in any of these 5 department stores? YES [""] NO ["H
YES, Recently r"];NO P1;P.K. [""]
If YES, ASK: Which one (or ones)? INITIALS D.K. Q
13. Which one of these 5 department stores is your favorite? D.K. ]
Why do you like this one (these) best? .
100 MARKETING RESEARCH
IDENTITY (PLEASE BE SURE TO COMPLETE THIS SECTION)
NAME OF RESPONDENT
ADDRESS
ESTIMATED AGE: 24 or less ]; 25-34 Q; 35-44 p; 45-64 ]; over 64 ]
ASK: How many members in your family, in thisnouse, are under 19 years of
age ? D.K. Q; are 19 years old or older ? O.K. p
GIVE INCOME CARD TO RESPONDENT. ASK: Would you please tell me in which
letter group your family income falls? GROUP D.K. [""] N [*""] w f"1
N.A.Q
INTERVIEWER'S NAME
DATE AND HOUR OF INTERVIEW .
REMARKS:
CASE 6 CONSTRUCTING A READERSHIP
QUESTIONNAIRE
BACKGROUND
In 1946 a survey was undertaken for the New York Times by
a private research organization to determine which of a group
of well-known newspapers and magazines are read regularly, and
which are preferred, by persons in a number of selected occupa-
tions. The group included editors of daily newspapers through-
out the country, public officials throughout the nation, college
presidents, and clergymen. The newspapers and magazines
included in the survey were as follows:
Newspapers:
New York Times weekday and Sunday editions separate
New York Herald Tribune " " "
Chicago Tribune " " "
New York Sun (this, as well as the following papers, had a week-
day edition only)
New York World-Telegram
Wall Street Journal
Journal of Commerce (Chicago)
Journal of Commerce (New York)
CASES IN PLANNING 101
Magazines:
Nation's Business
Newsweek
Time
United States News
Business Week
Fortune
Two specific questions were to be answered from the survey,
one on regular readership of these newspapers and magazines,
and one regarding the respondent's first and second order of
preference among those newspapers and magazines that he did
read regularly. It was decided not to ask for any classifying
information on the questionnaire, though breakdowns were to
be compiled by region and by occupation. Data on the former
could be obtained from the postmark, and occupational identi-
fication, it was thought, could be obtained through the use of
some sort of code with the questionnaire.
Mailing of the questionnaire was to be made more or less at
the same time to several hundred persons in each of the occupa-
tions selected for study. Names and addresses were secured
from various trade directories and other lists of people in the
specific occupations. Because of the brevity of the question-
naire, it seemed feasible as well as more convenient to have the
covering letter on one side of a sheet of paper and the question-
naire on the other side. This meant, however, that the respond-
ent's name and address would be known if the questionnaire
were returned. It was thought, nevertheless, that possible non-
response because of this reason would be very small if it were
stressed somewhere that individual identities would not be used
in any way.
THE PROBLEM
It was naturally desired to devise a covering letter and a ques-
tionnaire form that would maximize the response rate. Since
nearly all the occupations covered were of the sort that would
give the sample members very little time to spare, it was im-
portant to make the covering letter and the questionnaire short
102 MARKETING RESEARCH
and to the point. Other devices to increase the rate of response
were also sought, particularly since the schedule for the survey
did not allow for follow-up mailings. Thus, should a separate
covering letter be designed for each occupation, or should one
form of letter be employed throughout?
A coding method was also needed to distinguish between the
various occupations, or should this be dispensed with and identi-
fication made from the names of the respondents on the cover-
ing letter?
A final problem concerned the order of listing the newspapers
and magazines on the questionnaire. According to similar sur-
veys in the past, there was some reason to believe that the news-
paper or magazine given top listing in its respective group might
have some advantage over the others. If that were true, what
order of listing should be used?
ACTION NEEDED
The following things remained to be done before this survey
could be launched:
1 . Write a covering letter, or letters.
2. Prepare the questionnaire form.
3. Suggest any other means to be used for raising response rates.
4. Devise a method of coding occupations, if a code were used.
If not, explain why not.
5. Specify a procedure for avoiding possible bias caused by the
order of listing of the newspapers and magazines.
CASE 7 PROPOSED QUESTIONNAIRE FOR AN
ICE CREAM SURVEY
BACKGROUND
The Lee Dairy Company did not have a marketing research
staff. The sales manager, however, made occasional use of grad-
uate students from schools of business. Through one of the
schools in the area he hired a student, Henry Lodge, on a tern-
CASES IN PLANNING 103
porary, part-time basis to plan and conduct a survey on ice
cream flavors.
THE PROBLEM
The sales manager explained that he wished to survey con-
sumers in the city regarding the selection and use of different
flavors of ice cream. The Lee Dairy Company produced from
sixteen to twenty-seven different flavors, the number depending
on the season. Of these, only vanilla and chocolate were con-
sistently large sellers, although the total volume sold of other
flavors was substantial. Sales by flavors varied considerably
among neighborhoods and outlets.
The sales manager believed that the company could gain
greatly were it able to increase the number of flavors of Lee Ice
Cream which the average retail distributor carried. Equally
important, however, was the need to provide route salesmen
with more information on consumer preferences and habits re-
garding ice cream flavors. Although the problem of flavors
could not be divorced completely from pricing, point-of-sale dis-
plays, and other merchandising considerations, or from com-
petitive factors, the competitive factors were not considered
important at the moment. Basically, it was desired to secure
whatever relevant data were obtainable on consumer purchase
and consumption of ice cream flavors that would aid salesmen
in selling drugstores, food stores, and other retail outlets more
effectively. Some of the information, of course, would be
passed on to retailers to aid them in doing a better sales job.
The sales manager said that he expected about three hundred
to four hundred personal interviews to be made with housewives
in the city. The sample design was up to Mr. Lodge, who also
was expected to hire whatever interviewers he needed to help
him. The manager suggested that Mr. Lodge mull over the
problems, arrange his plans, construct a questionnaire, then talk
with him again before proceeding further.
THE PROPOSED QUESTIONNAIRE
Two weeks later Mr. Lodge returned with the following ques-
tionnaire, reproduced in full:
104 MARKETING RESEARCH
EXHIBIT 1
Questionnaire for an Ice Cream Survey
1. What is your favorite brand of ice cream?
Why?
2. Would you go very far out of your way to get your favorite brand?
Yes No
3. How many brands of ice cream have you used in the last six months?
(Names) , , ,
4. Have you stopped using any brand completely in this time? Yes No
5. Why did you stop using this brand?
6. What was the brand of ice cream you bought LAST, to eat at home?
7. Was it STORE PACKED? Or FACTORY PACKED?
8. Where did you buy it? (Name and address)
9. What flavor(s) did you buy the LAST time? .
10. Was it the flavor(s) you wanted? Yes No Was it your fa-
vorite flavor? Yes No
11. Do you Always, Usually, Seldom, Never, get the flavor(s)>
you want?
12. Was there a large selection of flavors from which to choose? Yes No
Didn't notice How many flavors were there?
13. Did you buy something else at the same time? Yes No Don't
remember If YES, what else did you buy?
14. Do you purchase most of your ice cream at the store where you bought the
LAST time? Yes No If NO, where do you buy most of your ice
cream? What flavors?
15. Why did you buy at that particular store the LAST time?
16. In what OTHER stores do you purchase ice cream? .
17. Are there any things about the store where you USUALLY buy ice cream
which you do not like? Yes No (If the answer is YES, list them. )
18. Do you know of any stores that sell ice cream for less than you pay for it in
the store where you USUALLY buy? Yes No Don't Know (If
YES, will you name the brand or brands?)
19. The LAST time you bought Store Packed ice cream, did you receive a gener-
ous portion? Yes No Don't Know
20. Do you think that it is more than you would get in other stores with whichi
you are familiar? Yes No Don't Know
21. While in any of the stores where you buy ice cream, have you ever noticed 1
any ice cream advertisement or display? Yes No (If YES, please
describe it.)
22. What was the last ice cream advertisement you heard on the radio or tele*
vision about?
23. What was the brand?
24. What was the last ice cream advertisement you saw in the newspaper about?
CASES IN PLANNING 105
25. What was the brand?
26. What type of ice cream do you prefer, Store ^Packed or Factory
Packed?
27. Why do you prefer this type of ice cream?
28. If you buy both Store Packed and Factory Packed ice cream, do you pur-
chase both at the same store? Yes No If NOT, why not?
29. Who USUALLY buys the ice cream for home consumption in your family?
30. Is the selection of the brand and flavor left up to that person? Yes
No If NO, who makes the selection?
31. What time of day is it usually purchased? Morning Noon
Afternoon Before dinner In the evening
32. What day(s) of the week do you USUALLY buy? .
33. Do you buy ice cream for home consumption when you are shopping for
other things? Never Rarely Occasionally Frequently
34. Do you purchase ice cream from the vending peddlers who canvass the
neighborhood? Yes No How many times a week? What
type? Novelties Pints Quarts How much at a time?
What flavors? Who eats it?
35. Does you milkman deliver ice cream along with the milk? Yes No
Don't know Do you ever buy ice cream from him? Yes No
How many times a week? How much at a time?
What flavors?
36. Do you have a deep freezer? Yes No
37. Do you ever buy ice cream and store it in your deep freezer? Yes No
How much do you buy at a time? What flavors?
How long does it last?
38. Do you make your own ice cream? Yes No Why do you make it?
. Which do you prefer? Why?
What flavors? .
ACTION NEEDED
In reviewing this questionnaire, the sales manager raised the
following questions with Mr. Lodge:
1 . Is the problem well and fully covered?
2. Is the questionnaire too long?
3. Are some of the questions, particularly the open-end ones, of
the best type possible for the problem?
4. Can the sequence of questions be improved?
5. How are respondents and their families to be identified or
classified?
6. If this questionnaire were pretested, what responses might
be anticipated? How might such anticipation aid in revis-
ing the questionnaire?
106 MARKETING RESEARCH
CASE 8 GETTING A LUGGAGE
QUESTIONNAIRE INTO SHAPE
BACKGROUND
The associates of Researchers, Inc., were firmly convinced
that a consistently good output of marketing research depended
not on individual brilliance but on effective teamwork. This
principle was incorporated into staff organization and opera-
tions. During the various stages of each project, the project
director and his immediate staff would consult with other senior
and junior staff members at regular meetings concerning the
problems they faced.
That these meetings, as all democratic processes, tended to
be protracted and somewhat confusing was recognized. Yet
they encouraged original suggestions, led to a broad consider-
ation of issues, and provided a searching test of all plans. In
general, it was believed, these methods produced effective re-
search and provided good training for staff members.
On one occasion, a questionnaire concerning luggage was
being drafted for a personal interview survey of families. Drafts
had been prepared by several staff members and many questions
that might be used in the survey had been suggested during the
meetings held on the subject.
HYPOTHESIS ON LUGGAGE USE
A basic problem concerned the factors affecting luggage needs
and ownership. The hypothesis formulated stated: the amount
and kinds of luggage needed and owned by families will be in-
fluenced, among other factors, by the type of business, vacation,
and other travel uses, and by the amount of sharing of luggage
among family members and others.
Questions suggested and considered included these:
1. A. Did you take a vacation this year?
If YES: How many pieces of luggage did you take?
B. Do you plan to take a vacation next year?
If YES: How many pieces of luggage would you probably take?
CASES IN PLANNING 107
If NO: How many pieces of luggage would you probably take if
a trip came up?
2. A. Do you personally own any luggage that is ustd for traveling?
If YES: What types?
B. Is there any other luggage in the family that is used for traveling?
If YES: Who owns this luggage? What types?
C. Are there any members of the family that do not own luggage?
If YES: Who are they?
3. Is any piece of luggage used by more than one person in your family?
4. When traveling away from home, what means of transportation do
you use most often?
5. When did your family last take a trip together?
6. A. When did a member of your family last take a trip alone?
B. What was the purpose, or reason, for that trip?
C. How many pieces of luggage did take along?
7. What luggage do you use the most? For what purposes?
8. Do you ever loan your luggage to other members of the family or to
friends? CHECK WHICH ONE(S).
9. Do the various members of the household use one another's luggage?
10. How many trips away from home have you made during the past
year?
1 1 . Not including trunks, how many pieces of luggage did you take with
you on your last trip?
12. Were any of the pieces of luggage you took with you on your last
trip borrowed from persons outside your family?
13. In your household (A) does each member of the family have his or
her own personal luggage? or (B) is the luggage family property and
used by all members?
14. How many times a year do you (A) use your luggage? (B) loan it to
a member of your family? (C) loan it to a friend or relative?
15. Has the family, as a group, or any member or members of the family
taken a trip, or trips, long enough to involve the use of luggage
within the last six months?
If YES: What mode of travel was used?
Which of the following types of luggage were used?
16. A. When you travel for pleasure, vacation, or week end, what type
of luggage do you use?
B. When you travel on business, or when other members of your
family travel on business, what type of luggage is used?
17. Do relatives or friends or family members ever borrow your luggage?
108 MARKETING RESEARCH
1 8. A. When did you or members of your family last use luggage?
(1) For pleasure?
(2) For business?
B. When do you expect to use your luggage again?
(1) For pleasure?
(2) For business?
19. How many days are you away from home on an average trip?
20. Since January 1 (or: In 19 ) , how many times did you yourself use
any luggage? Which case, or cases, did you use most often? Why
is this case used most often?
2 1 . Who travels most in family?
IF OTHER THAN PERSON BEING INTERVIEWED: Which type, Or typCS,
of luggage did use? Reasons for use of this type most often?
22. A. When was it that you last took a trip? The time before that?
B. How did you travel the last time?
C. How many pieces of luggage did you take with you?
D. What style and material were these pieces?
HYPOTHESIS ON LUGGAGE GIFTS
Another objective of the study was concerned with the
amount and types of luggage bought as gifts, the identity of
purchasers and recipients of gifts, and the occasions for giving.
Among questions suggested for use were these:
1 . Have you ever received a piece of luggage as a gift?
2. If you have received luggage as a gift, on what occasion was it given?
(A) Wedding? (B) Graduation? (C) Birthday? (D) Anniversary?
(E) Christmas? (F) Mother's Day? (G) Father's Day? (H) Retire-
ment? (I) Other? CHECK WHICH ONE(S).
3. Which pieces of your luggage did you receive as gifts? On what oc-
casions?
4. A. When did you last receive a piece of luggage as a gift?
From whom? For what occasion?
B. When did you last give a piece of luggage as a gift?
To whom? For what occasion?
5. Have you or your family ever received luggage as a gift?
Ever given luggage as a gift? What occasion?
How long ago was it?
6. Have you ever given or received luggage as a gift?
7. For what occasions would you consider luggage as an appropriate
gift?
CASES IN PLANNING 109
8. Was the last piece of luggage you bought for your own use or for a
If a GIFT: Would you mind telling me to whom you gave it that
is, husband, daughter, son, etc.?
9. A. Have you or your family ever received luggage as a gift? If YES:
Which member or members?
From whom was it received?
What was the occasion?
When did this happen?
B. Have you or your family ever given luggage as a gift?
If YES: Which member or members?
To whom was it given?
What was the occasion?
When did this happen?
What type luggage was given?
What price was paid?
10. Have you or your family received luggage as a gift in 19 ?
11. How many pieces of the luggage that you or your family now own
were received as gifts?
12. Within the last six months has the family, or any individual mem-
ber of the family, given luggage as a gift?
If YES : Who bought it for whom?
ACTION NEEDED
The questions quoted exemplify the problems faced, al-
though they do not include all the questions suggested. Many
are crudely expressed, and typify questions that might be
drafted on the spur of the moment and culled from the tran-
script of an open meeting. Almost any question ever worded,
of course, can be objected to on some grounds. (Indeed, criti-
cizing the other fellow's questionnaire is a favorite indoor sport
in some circles.) These were, however, the raw materials of a
questionnaire that was being constructed. The problem was to
put this material together into a workable questionnaire. To
do so, however, raises the following questions:
1. If many families are interviewed, will the giving of luggage equal the
receiving, so that data need be gathered only on one or the other?
2. For what time period should data on gifts be asked and by what
method?
110 MARKETING RESEARCH
3. What are the complexities of luggage use that complicate the word-
ing of the questionnaire? Might instructions on whom to interview
simplify the problem?
4. Which questions exemplify common faults or problems of question-
naire construction? How might each such question be improved?
5. Which questions are best for clarity and ease of recording answers?
6. For each of the two problems, what sequence of questions is best cal-
culated to yield the answer?
7. Based on this evidence, would you say an open meeting is a successful
way to construct a questionnaire? Under what conditions might a
meeting for this purpose be most effective?
CASE 9 FOLLOWING UP ON A
DEPARTMENT STORE STUDY
BACKGROUND
The survey made of the customers of the Langdale Depart-
ment Store (Part III, Case 5, pp. 91 to 100) had proved to be
of interest and value to store executives, In particular, the study
had focused attention on the potential value of greater knowl-
edge of customer actions to the merchandising and promotional
operations of the store. A follow-up study, therefore, was auth-
orized to explore some of the many possibilities.
THE PROBLEM
After the problem for this study had been formulated, ques-
tion arose regarding the proper types of follow-up investigations
that should be made.
DATA AVAILABLE
Preliminary problem formulation had resulted in the follow-
ing statement:
Project: The relation of customers to merchandise departments.
Basic problem: How can the store utilize knowledge of customers' be-
havior in relation to purchases of specific types of merchandise in
order to increase sales? In particular, how can an identification of
successful, or key, departments and an analysis of the reasons for their
success (primarily in relation to customer types and customer be-
CASES IN PLANNING 111
havior) lead not only to improved store promotion and increased store
traffic, but also to profitable imitation of Jheir success by other de-
partments?
Hypotheses:
A. The relation of a large department store to its customers is not that
of one retail unit to one market, but rather is that of a hundred or
more small stores to dozens of special markets.
1 . A large number of customers are loyal not to the store but only to
one or a few departments.
2. The store has a few key departments which create most of the
selective buying.
3. The customers for many individual departments are so homo-
geneous by income, geographic location, age, or other character-
istics as to constitute a special market.
4. Many customers almost always go to a given department store to
buy certain merchandise (even shopping goods).
5. The widespread patronage of upper income families may reduce
the number of departments patronized.
B. The volume of store traffic and sales can be increased significantly by
an understanding and use of the relation of customers to specific de-
partments.
1. Some key departments are patronized by a large proportion of all
store customers, but most arc not.
2. Large sales volume in some key departments is due primarily to
reaching a large segment of their potential market.
3. Promotions in key departments are many times as effective in
producing store traffic and the sale of related, impulse, or other
merchandise as are promotions in other departments.
4. There may be some combination of, say, six or ten key depart-
ments which will be 90 per cent as effective in developing store
traffic as are blunderbuss store- wide promotions.
5. Identification of key departments and an analysis of their reasons
for success can lead not only to better store traffic but also to
profitable imitation by other departments.
6. The greater the number of departmental loyalties the larger will
be customer expenditures.
7. Promotions staged by departments with the greatest number of
customers are the best promotions for the store as a whole.
C. The merchandising of specific departments can be improved by
knowledge of their customers.
112 MARKETING RESEARCH
1 . Some departments make relatively large sales to a limited number
of customers, their increase in volume depending mostly on get-
ting new patrons.
2. Some departments make relatively small sales to a great number
of patrons, their increase in volume depending most on increasing
the size or number of purchases by present patrons.
3. Price-line emphasis in some departments may be far out of line
with the relation of store or department customers to price lines.
4. If many customers for some departments are patrons only at times
of special sales events, a promotional weakness or failure of these
departments is indicated.
5. Knowledge of purchase habits can aid in the selection of advertis-
ing media in general or of the merchandise of specific depart-
ments.
Subsidiary Problems:
A. How loyal are customers to specific departments and how loyal are
they to the store?
Possible measures of loyalty are:
To Store:
1 . Patronage only at this department store.
2. Patronage at 3, 4, 5 (or more) other department stores.
3. A given dollar amount of purchases (in relation to income).
4. A given per cent of total family expenditures or total department
store purchases.
5. Patronage of or more departments over a given period of
time.
6. Spending more here than at any other store.
7. Expressed intent (or habit) of always starting a shopping trip by
coming to this store.
8. Expressed general preference for this store.
9. Expressed preference for particular aspects of this store.
To department:
1. A given percentage of total credit expenditures in any one de-
partment.
2. A given dollar amount of purchases (in relation to income) over a
particular period of time.
3. A given number of purchases (or purchase visits) over a particular
period of time.
CASES IN PLANNING 113
4. A department with a loyal following defined as one in which, say,
50 per cent of sales are made to 1 per ce, nt of department patrons.
5 . Regularity of buying or frequency of purchases.
6. Expressed intent almost always to buy given merchandise at a
given department.
7. Expressed intent to visit department upon entering store.
8. Expressed preference for particular departments.
9. Expressed preference for particular aspects of departments.
B. What merchandise is purchased by (a) distant customers, (b) custo-
mers who shop infrequently, and (c) customers who spend a limited
time shopping, and can their purchases be increased by promotion of
telephone and mail services or by other means?
Potential Value to Management:
1. Need for, or desirability of, institutional advertising or activities.
2. Determination of whether or not the successful cultivation of
customers in different income brackets depends on particular de-
partmental appeals, wide assortments or price lines, and so forth,
so as to appeal more effectively to special groups.
3. Recognition of key departments for: (1) promotions aimed at all
income groups, (2) promotions aimed at specific income groups,
and (3) promotions aimed at specific types of persons.
4. Recognition of departments that attract only bargain hunters.
5. Recognition of departments that primarily create store traffic and
thus aid sales in other departments the most.
6. Recognition of weak or strong departments in terms of customer
loyalty, regular patrons, sale patrons, and so forth.
7. Recognition of departments whose customers are so homogeneous
(and loyal) as to constitute a special market, and one, perhaps,
best cultivated by direct mailings.
Data gathered on a previous customer survey were useful
in making these formulations and were expected to provide use-
ful correlations with data gathered on the current study. The
question arose, however, as to the need for gathering new data
directly from customers.
Among the relevant findings of the survey were these: only 4
per cent of store customers made no purchases for credit or cash
at other department stores in the city. Of their customers liv-
ing in the city, however, 22 per cent had credit accounts only at
Langdale's and 36 per cent of their customers outside the city
114 MARKETING RESEARCH
had credit accounts only at this department store. Credit busi-
ness was over 60 per cent of the total, and when the cash ex-
penditures of credit customers were counted, less than 10 per
cent of business came from customers without credit relations
at the store. Approximately 30 per cent of all their customers
patronized at least four out of the five largest department stores
in the city. The higher the income of customers, the greater
was the tendency to patronize more stores, to utilize credit in
buying (primarily the regular 30-day charge account), and to
buy at least part of their goods by telephone. Half of all cus-
tomers had bought some goods from some department store by
telephone during the year, and one third had made telephone
purchases at Langdale's. The same proportions also held for
purchases by mail. The lower the income, however, the more
mail buying, contrary to the pattern for telephone buying.
ACTION NEEDED
1. How could such hypotheses be tested by consumer inter-
views, talks with store personnel, or inspection of company
records?
2. In what ways might the troublesome problems raised by ex-
penditures for cash, on which individual customer records
do not exist, be minimized or solved?
3. Can the study successfully be restricted to the gathering of
data from store records? How can it be coordinated with
the earlier study?
Part IV
SAMPLING
The need for sampling in marketing research arises because
the limitations of time and resources rarely permit study of all
the members of a population. Information with regard to the
total, therefore, has to be obtained by means of a sample, a sub-
set of the total. Nevertheless, data obtained in this manner
may be more accurate than those obtained from a complete
census. The reason for this is that restriction of data gathering
to a sample of the whole permits more attention to be devoted
to the many aspects of a survey operation that are likely to bias
the results. Interviewer training is one of the most important
of these. Errors produced by sampling variation may be negli-
gible relative to those caused by misunderstanding of interview-
ing instructions. Proper attention to the latter in a sampling
operation may, therefore, yield more accurate results for the
population than a complete census which eliminates sampling
errors but leaves room for these other errors.
Properly designed samples can yield population data with
only a negligible loss in accuracy and at a fraction of the cost
of surveying the entire population, and therein lies the great
value of sampling to market research and to many other disci-
plines.
A Review of Some Basic Sampling Concepts
The Objective of Sampling. The basic aim of any sample
is to provide the necessary information with the required accu-
racy at a minimum cost or to maximize the accuracy of the in-
formation obtainable at a given cost.
The Meaning of Accuracy. Accuracy in sampling is com-
115
116 MARKETING RESEARCH
pounded of two distinct factors, precision and bias. The former
derives from the fact that estimates based on samples will neces-
sarily fluctuate about the true, unknown population values
(means, medians, standard deviations, etc.), or parameters, be-
cause of sampling variation, because hardly any sample is likely
to be a perfect miniature of the population from which it is
chosen. If a large number of samples of the same size are
drawn to estimate the same parameter, each sample member
drawn at random from the population, the estimates based on
these samples will be scattered around the true population value
in much the same way that the shots of a rifle aimed at a certain
target are scattered around the target. The probability that
any one estimate will fall within a certain range of the true value
can be worked out by probability theory for different sample
designs. Conversely, on the basis of this theory, the margin of
error, or sampling error, that can be expected with a particular
sample design and sample size is ascertainable at different prob-
ability levels (always, however, under the assumption of random
selection). Therefore, once the allowable margin of error and
the desired probability levelthe precision of the sample are
given, the sample size and funds needed by each of the various
sample designs can be computed.
The other component of accuracy, bias, represents all the in-
fluences that tend to distort the estimate. Arbitrary selection
of sample members, prejudicial question wording, preconceived
interviewer attitudes, faulty editing, and many others can pro-
duce sample values that will not average out to the true popula-
tion values, no matter how large the sample. In other words,
unlike sampling error, bias is largely invariant of sample size,
and for this reason the basic objective of sampling can be re-
written thus: to produce no more than a given sampling error
at minimum cost, or to minimize the sampling error at a given
cost, depending on the situation at hand.
Arbitrary vs. random selection. The manner in which the
individual sample members are selected determines whether
or not the sampling errors of the estimates can be measured.
If the sample members are selected at random from the popula-
tion (which means that each member of the population must
SAMPLING 117
have an equal, or known, chance of being selected in the sam-
ple), probability theory is applicable, the sampling errors are
measurable, and the reliability of the estimates can be gauged.
The availability of so-called random sampling number tables *
and various schemes of systematic selection (such as selecting
every nth name or address, n being the ratio of the size of the
population to that of the sample) often reduce the process of
random selection to a routine matter.
If, however, the individual sample members are chosen in
some arbitrary or haphazard fashion, present probability theory
cannot validly be used to measure sampling errors, and their
estimation becomes largely a matter of individual judgment.
Arbitrary selection is still widely used, partly out of habit,
partly in the mistaken belief of avoiding higher costs associated
with random selection, and partly in a legitimate sense because
random selection may be unnecessary. The last is true for
many pretests, for certain types of taste tests, for audience reac-
tion studies, and in general for any survey whose principal objec-
tive is to secure general reactions rather than specific estimates.
Case 8 in this Part illustrates one such situation. However,
where the purpose of a study is to secure estimates of certain
unknown population characteristics, random selection is emi-
nently desirable, not only as the sole means of measuring
sampling error, but also for avoiding the bias inherent in most
arbitrary selection procedures. Even though the final sample
may not turn out to be perfectly random because of the high
nonresponse rates often obtained on marketing surveys the
use of this means of sample selection, combined with assiduous
callbacks on the nonrespondents, will generally produce more
reliable data than will arbitrary selection.
Probability Samples and Judgment Samples. These terms,
which have gained wide usage, are nothing more than synonyms
for random selection and arbitrary selection, respectively. A
probability sample is one whose sampling error is measurable,
1 A description of the use of such tables will be found in most marketing
research texts. Also see H. N. Broom, New Random Sampling Numbers, Waco,
Texas: Baylor University, Baylor Business Studies, I, No. 1, 1949, pp. 10-15.
118 MARKETING RESEARCH
a condition that exists only, as has been noted above, when ran-
dom selection is used. On the other hand, a judgment sample
is one whose error is not measurable, which holds when arbitrary
selection is used. It cannot be overemphasized that a sampling
design, whether it be area sampling, stratified sampling, or any
other, can be set up either on a probability or on a judgment
basis; that is, it can be set up so that the sampling variations
can or cannot be validly measured by means of the sampling
error formulas.
Some Principal Sampling Designs
A wide variety of sample designs are available for use in mar-
keting research. Perhaps the main ones are the following:
Unrestricted sampling 2 each sample member is selected
from the entire population being studied. If random selection
is used, this means that every member of the population must
have an equal chance of being chosen. Unrestricted sampling
is generally desirable if little is known about a particular popula-
tion, especially if it is relatively small and compact.
Area sampling a modification of unrestricted sampling
based on prior subdivision of the population into areas and
restriction of sample selection to certain of these areas. This
method does not generally reduce the sampling error relative
to that of an unrestricted sample of the same size. However,
it may lead to substantial reductions in cost, particularly where
large geographic areas are involved, because of concentration
of the field work.
Cluster sampling selection of sample members in groups in-
stead of individually. Thus, a sample of three hundred families
might be constructed by selecting sixty groups of five neigh-
boring families instead of making three hundred individual
selections. This method is especially advantageous where an
* This is also widely referred to as random sampling, but the use of the word
"random" in this connection is not recommended, because of the danger of con-
fusing it with random selection, where the word "random" refers to the manner
of sample selection. Thus, one could have a so-called "random" sample with
nonrandom (arbitrary) selection of individual sample members.
SAMPLING 119
appreciable part of the interviewers' time may be taken up with
traveling.
Stratified sampling prior division of the population into
strata based on characteristics considered relevant to the sub-
ject under study, and selection of sample members from each
stratum. In effect, it is a weighted aggregate of unrestricted
samples. For stratified sampling to be useful, the following
three conditions must be met:
1. Certain relevant characteristics are known to influence
strongly the subject under study.
2. Division of the population by the relevant characteristics
is practicable.
3. The relative division of the population by these character-
istics is known with a fairly high degree of accuracy.
This means, for example, that if city size were known to affect
substantially the movie attendance characteristics in a certain
area, the use of stratified sampling could not be considered un-
less unrestricted samples could be selected from each city-size
class, in turn, and the relative size of each city-size group in the
area were known. A stratified sample measuring movie attend-
ance characteristics could then be constructed by segregating
the communities in the area into strata by size, and then select-
ing people at random from each stratum.
If the number selected from each stratum is proportional to
the relative size of the stratum in the population, we have a
proportionate stratified sample; otherwise we have a dispropor-
tionate stratified sample. The latter is preferable if the various
strata become less homogeneous with respect to the character-
istic^) understudy.
A stratified sample can be constructed even if members of
each stratum cannot be identified, as long as the relative sizes
of the strata are known. In other words, a stratified sample of
families by income level is feasible even if the identities of the
families in each income level are not known, as long as the ag-
gregate proportion of families in each income level is known. 3
8 R. Ferber, "The Common Sense of Sampling/' Current Economic Com-
ment, XI (August, 1949), pp. 48-56.
120 MARKETING RESEARCH
The sampling error of a stratified sample may be considerably
below that of an unrestricted sample of the same size. This is
because the stratified design makes use of the additional infor-
mation regarding the breakdown of the population by the rele-
vant characteristic(s) and thereby serves to reduce the margin
for sampling error. The trouble with stratified sampling in
practice is that, though the relevant stratifying characteristics
are usually known, we do not always know with any great accu-
racy the relative sizes of the strata in the population. Because
of this lack of knowledge, the gains made possible by stratifica-
tion may be more than offset by variations introduced by inac-
curate data regarding the sizes of the population strata, a factor
that unfortunately is frequently overlooked. Any evaluation of
the desirability of stratified sampling in a particular case must
take into account the increased sampling error resulting from
inaccurate population weights as well as the inherent gains due
to stratification.
Stratified designs do not preclude the concomitant use of area
sampling or cluster sampling schemes. Thus, the United States
might be subdivided into regional strata, such as North, South,
East, and West, with areas selected within each region and with
sample members selected from each of these areas in clusters.
Similarly, the selection of the members of a stratified sample
could be accomplished either by random or arbitrary pro-
cedures. 4
Double sampling selection of a large initial sample, on the
basis of interviews with which a subsample is chosen for more
intensive interviewing. It is generally most useful when the
subject under study is known to be related closely to certain
characteristics but the distribution of these characteristics is not
known. If, say, automobile purchase habits are known from
* A much-maligned term in recent years is "quota sampling/' which is essen-
tially a synonym for "stratified sampling" arising from the fact that quotas are
generally established for each stratum. The widespread use some years ago of
arbitrary selection in filling these quotas has led many to believe that quota
sampling per se is a biased form of sampling. Unless one defines it specifically
in this manner, however, the fact remains that quota sampling is nothing more
or less than stratified sampling and can be applied with either random or arbi-
trary selection.
SAMPLING 121
past studies to be closely related to income, a study of the sub-
ject in City Z might be conducted by, estimating, from brief
interviews with an unrestricted sample, the city's income distri-
bution, and selecting a subsample of the sample number in each
income bracket for lengthy interviews on their automobile pur-
chases and preferences.
Multistage sampling selection of a sample in successive
stages, most frequently used in constructing area samples.
Thus, of the families in the United States, a sample on a prob-
ability basis might be selected in the following four stages:
1. Combining the 3,000-odd counties in the country into, say,
one or two hundred groups and selecting a few groups for
inclusion in the sample.
2. Within each sample group, selecting one or more areas.
3. Within each sample area, selecting one or more blocks or
smaller areas.
4. Prelisting the inhabitants of these smaller areas and select-
ing the members of the sample.
Sequential analysis a radical departure from conventional
sampling in that the sample size is not predetermined but de-
pends on the sample results themselves. In other words, by the
sequential method a running score is kept of the results obtained
on all interviews made. As soon as this score indicates a con-
clusive answer one way or another, as computed on the basis
of probability theory, sampling is stopped and the survey is fin-
ished.
Sequential sampling can at times prove a great money-saver
in marketing research. It is applicable, however, only when all
the following considerations are met: there must be random
selection; nonresponse must be practically nonexistent or, if
existent, appropriate safety devices must be built into the se-
quential formulas; records must be kept of the sequence in
which the interviews are made either singly or in groups and
it must be possible to cumulate and examine the results of the
interviews periodically; a few key questions amenable to se-
quential treatment must be available.
122 MARKETING RESEARCH
Consumer panels data on most sample surveys are secured
in one interview with the respondent; these are sometimes re-
ferred to as one-spot samples. There are many cases, however,
where repeat interviewing with the same respondents con-
sumer panels, using "consumer" in the broadest sense of the
word is desirable. This is true whenever the purpose of the
study is best served by observing changes in attitudes or behavior
of the same persons over a period of time, a determination
which must be made in each individual case.
The sampling aspects of a consumer panel operation are
slightly more complicated than the corresponding one-spot
samples because the following additional considerations must
be taken into account:
1. The initial sample size must be large enough to counteract
panel mortality.
2. A decision must be made regarding replacement of dropouts.
Should they be replaced, and if so, how? The most common
means are by random selection or by attempting to find
families possessing characteristics as close as possible to those
dropping out.
3. Consideration must be given to methods of maintaining
respondent cooperation and minimizing panel mortality.
Premium deals sometimes prove helpful.
4. The danger of bias is, if anything, greater than with one-spot
samples, as many forms of bias are magnified with time.
Atypical representation of the sample is commonly thought
to be one such form of bias.
Some of the problems involved in setting up a consumer
panel operation are raised in Case 6 of this Part.
The Cases in This Part
The eight cases in this Part illustrate a variety of sampling
problems. The emphasis is on the broader, operational aspects
of sampling, though some of the cases (2 and 3) also go into the
more technical aspects, such as determination of sample size.
The first case presents a simple exercise designed to acquaint
the student with the use of random sampling number tables.
SAMPLING 123
These tables are basic to many surveys, yet most marketing stu-
dents are given little opportunity to use them.
The second case poses some of the problems leading to ques-
tions of method and sampling. The question of the adequacy
of a particular sample design is the subject of the third case. It
presupposes some knowledge of the principal sample designs
and of their relative merits.
Designing a sample from the very beginning is the subject of
the next two cases. The design of a sample to test a particular
hypothesis is the problem in Case 3, whereas the problem in
Case 4 is to set up a sample design to answer more than one
question, a so-called multipurpose survey.
Case 5 indicates the practical ways in which a controlled ex-
periment can be designed. The headaches involved in a con-
sumer panel operation are raised in Case 6, which at the same
time provides further practice in sample design.
The value and desirability of follow-up interviews are the
problems in Case 7. This is a problem that occurs frequently
in practice but to which relatively little attention seems to be
given in the classroom.
Case 8 serves a dual purpose. It shows how a judgment
sample can prove useful and at the same time serves to high-
light its limitations and the difficulties involved in validating
the results.
BIBLIOGRAPHY
AMERICAN MARKETING ASSOCIATION. The Technique of Marketing Research.
New York: McGraw-Hill Book Co., Inc., 1937. While this book contains
limited information on various aspects of sampling theory and the applica-
tions to marketing research, it would be useful to read Chapter 20 and com-
pare it with recent writings in order to note the changes that have taken place
on the subject of sampling.
BRADFORD, E. S. Marketing Research. New York: McGraw-Hill Book Co.,
Inc., 1951. Chapter 9 presents a general introduction to the subject of
sampling, with a more detailed discussion on sample design in Appendix A,
and with practical illustrations of sampling operations in Appendices B to E
andG.
BROWN, L. O. Marketing and Distribution Research. New York: The Ronald
Press Co., 1949. Chapter 21 contains a brief treatment of sampling pro-
cedures as advocated some ten years ago, but which have been revised con-
siderably in more recent books.
COCHRAN, W. G. Sampling Techniques. New York: John Wiley & Sons, Inc.,
124 MARKETING RESEARCH
1953. A good general reference work on sampling techniques with emphasis
on the measurement of sampling errors for samples of different designs.
DEMING, W. E. Some Theory of Sampling. New York: John Wiley & Sons,
Inc., 1950. A technical treatment of sampling theory and applications,
taking up the mathematical bases for the various sampling methods. Chap-
ter 1 contains a good discussion of survey planning, and Chapter 2 presents an
excellent treatment of the various types of errors encountered in sample sur-
veys. The book is replete with illustrative examples and is a valuable refer-
ence volume.
FERBER, R. Statistical Techniques in Market Research. New York: McGraw-
Hill Book Co., Inc., 1949. Chapter 3 discusses the different sampling tech-
niques; Chapter 7 takes up the theory and application of sequential analysis
to marketing research data; Chapter 8 goes into the selection of alternative
sample designs and the determination of sample size; and Chapter 9 is con-
cerned with the various forms of sample bias.
HEIDINGSFIELD, M. S., and BLANKENSHIP, A. B. Market and Marketing Analy-
sis. New York: Henry Holt & Co., 1947. Chapter 8 gives an outline of the
usual procedures in external research used to study a population by means of
a sample.
LORIE, J. H., and ROBERTS, H. V. Basic Methods of Marketing Research.
New York: McGraw-Hill Book Co., Inc., 1951. Part III contains perhaps
the most thorough exposition of sampling techniques of any elementary mar-
keting research text. Chapter 7 contains a simple introduction to sampling
theory and a very clear treatment of the different sampling techniques and
of the relative merits of probability vs. nonprobability sampling.
LUCK, D. J., and WALES, H. G. Marketing Research. New York: Prentice-Hall,
Inc., 1952. Problems of sample design as applied to marketing problems
with a discourse on the relative advantages of judgment and probability se-
lection are stressed in Chapter 9. A short note on sample size is to be found
in Appendix A.
PARTEN, M. Surveys, Polls and Samples. New York: Harper & Bros., 1950. A
treatment of means of selecting sample members is contained in Chapter 8.
A good discussion of sample techniques will be found in Chapter 7 and of the
factors entering into the determination of sample size in Chapter 9.
YATES, F. Sampling Methods for Censuses and Surveys. New York: Hafner
Publishing Co., 1949. A somewhat advanced book on sampling methods and
sample design requiring some knowledge of mathematics, but essential for
the student of sampling.
Part IV Cases
CASE I RANDOM SELECTION OF
SAMPLE MEMBERS
BACKGROUND
A newspaper in a middle-sized community contracted for a
survey to determine the characteristics of its readers in relation
to those of the community as a whole. Information was desired
on such factors as age distribution, income distribution, occupa-
tion, sex, and length of residence in the area.
A personal interview survey was planned to elicit this informa-
tion. About 300 persons were to be included in the sample, all
from the community. Selection of the sample members was to
be made at random in order to obtain numerical estimates of
the proportions of readers and nonreaders having the required
characteristics. To yield a final sample of 300, it was estimated
that an upper limit of 330 names would be required to allow for
nonresponse and for persons who had moved.
THE PROBLEM
For various reasons, sampling was to be restricted to the cor-
porate area of the city. A fairly recent city directory was avail-
able listing all the people residing in the city, and this directory
was brought reasonably up to date with the aid of figures ob-
tained from the local building permit office on new housing
construction in the area. For the purposes of this survey, the
city directory, supplemented in this manner, appeared to be
adequate. The problem that arose was how to select names at
random from it.
The organization of the relevant part of the city directory is
as follows:
125
126
MARKETING RESEARCH
I
q
H <1
W
CASES IN SAMPLING 127
A section of 468 pages lists all adult residents of the city in
alphabetical order. A supplement of six pages was added list-
ing new people in the area, making a total of 474 pages.
Each page contains two columns of names, with 70 lines to
a column. The last page, however, contains only 28 lines (of
two columns).
Some names occupy more than one line sometimes as many
as three lines because data on the person's occupation and
relationship in the family were provided. In addition, the
alphabetical listing was interrupted by a space of five lines when-
ever last names began with a new letter of the alphabet.
ACTION NEEDED
It was decided to employ a table of random sampling num-
bers to secure random selection of the sample members: that
is, to assure each member of the population an equal chance of
being selected in the sample. One page from a random sam-
pling number table is reproduced on page 1 26. What is needed
is a detailed set of step-by-step instructions for selecting the
sample members by random selection with the aid of this
random sampling number table. The table on page 126 is to
be used for illustrative purposes in setting out the procedure.
CASE 2 DETERMINING THE CHARACTERISTICS
OF THE FOUNDRY INDUSTRY
BACKGROUND
In 1947-48 Illinois was the third largest state in terms of num-
ber of foundries, having 436 of the 5,452 reported for the
United States. These 436 foundries were known to be of a
variety of sizes and types, but little else was known regarding
their characteristics. Statistical studies of the foundry industry
had been conducted by government agencies, by associations
within the industry, and by publications serving the industry,
but these studies had not contributed greatly toward providing
statewide information on the characteristics of the industry.
The people in the industry felt that it was time for such a study
128 MARKETING RESEARCH
to be made, and accordingly enlisted the services of a researcher
to direct the study.
THE PROBLEM
The researcher's first task was to lay out a plan for this study
which, when carried out, would yield the required information
at minimum cost. In essence, two types of information were
desired. One type comprised numerical estimates for all foun-
dries in the state relating to such items as amounts of materials
used, distribution of product by industry and by geographic area,
and extent of research operations. In addition, some general
information, more or less on a qualitative basis, was desired on
such subjects as working conditions in the foundries, personnel
policies, degree of mechanization, and community activities of
the foundries.
DATA AVAILABLE
Geographically, the 436 foundries in Illinois in 1947-48 were
scattered over the state, though concentration was evident near
the Chicago and St. Louis areas.
Data on these foundries by type and by number of employees
in Illinois and in the United States in 1947-48 are shown in
Exhibits 1 and 2. A list of these firms was obtainable from
Penton's Foundry List, 1947-48.
ACTION RECOMMENDED
Two questions of sampling arose in planning this survey: how
to obtain the information, and from whom to obtain it. Some
suggested that the information be sought by personal interview,
while others maintained that the data could be obtained just as
easily and much more economically by using mail question-
naires, an important consideration in view of the desire to
restrict expenditures. If mail questionnaires had been em-
ployed, however, something would have had to be done about
the nonrespondents.
Should a sample of the 436 firms be selected, or should an at-
tempt be made to secure information from all the firms?
CASES IN SAMPLING
129
Those favoring the former procedure thought that a carefully
selected sample, carefully interviewed, should yield as reliable
information regarding all the firms in the state as a complete
census. Those taking the other point of view maintained that
interviews with all the firms were not unwarranted here because
of the small size of the population, and that the bother of select-
ing a sample was hardly worth it in such a case. They admitted,
however, that if the data were to be obtained by personal inter-
view, considerable savings could be effected if only part of the
population were interviewed.
ACTION NEEDED
A sampling plan had to be prepared outlining the procedure
to be followed in gathering the data for the study. This plan
had to include consideration of the following questions:
1. Should the data be obtained by mail questionnaire or by
personal interview? If the former, what provision should
- be made for handling the problem of nonresponse?
2. Should data be sought from all the firms in the state or from
a sample of them? If a sample is used, what should its design
be, and how should the sample members be selected?
EXHIBIT 1
DISTRIBUTION OF FOUNDRIES BY TYPE OF METAL CAST, UNITED STATES
AND ILLINOIS, 1947-48
Type of metal
Number of
U. S
foundries
Illinois foundries
Number
Percent'of
total
U S
Gra iron
3,058
368
133
2,676
2,563
134
3,218
209
30
19
207
169
12
246
6 X
8 2
14 3
7
6 6
9
7 6
Malleable
Total nonferrous
5,452
436
8
Source: Foundry Industry Data Book, 1947-1948. Compiled by the market research
department of The Foundry (Cleveland, Ohio). .
NOTE : Totals do not equal the sum of details, since some foundries produce castings ox
more than one kind of metal.
130 MARKETING RESEARCH
EXHIBIT 2
DISTRIBUTION OF FOUNDRIES BY NUMBER OF EMPLOYEES,
UNITED STATES AND ILLINOIS, 1947
Covered in this study
Number of
employees
United States
Illinois
Questionnaire
replies
Visits to plants
Number
Number
Number
Number
of
Percent
of
Percent
of
Percent
of
Percent
found-
of total
found-
of total
found-
of total
found-
of total
ries
ries
ries
ries
1 ,000 or mere . .
47
9
19
2 3
7
3 1
I
2 4
500-1,000
103
1 9
14
3 2
8
3 S
2
4 8
250-500
29-1
5 3
22
S
16
7
6
14 3
100-250 .
763
14
62
14 2
28
12 3
9
21 4
50-100 . . .
944
IT 3
78
17 9
45
10 8
10
23 8
25-50 . .
1,059
19 4
95
21 8
SI
22 5
6
14 3
Less than 25
2,245
41 2
155
35 6
72
31 7
8
19
Total, all foundries
5,452
100
436
100
227
100
42
100
Source: Foundry Industry Data Book, 1947-1948, for the United States and Illinois.
* Foundries with only 8 employees numbered 24 and accounted for 10.6 per cent of the
total reporting.
CASE 3 A CONSUMER SURVEY OF
PHILADELPHIA
BACKGROUND
In the spring of 1947, a sample design was needed for a per-
sonal interview survey of housewives in Philadelphia concerning
the use by families of a class of food products. This class of
food products had been included in pantry inventory studies in
San Francisco (1941) and in Philadelphia (1946) and in a na-
tional study of foods used by urban dwellers (1942). Few de-
tails on either use or users were available, but these studies, as
well as production figures, showed a rapidly expanding market.
Approximately 60 per cent of Philadelphia families were be-
lieved to use the products to some extent. It was estimated that
with the funds and time available and with the breakdowns
desired, a sample of about five hundred would be sufficient.
CHOICE OF SAMPLING METHOD
Data for a highly stratified sample were not available on a
reasonably up-to-date basis. A random, or unrestricted, sample
CASES IN SAMPLING 131
from the population at large was also considered unfeasible.
No complete listing of families, or dwelling units, was currently
available, the last directory for Philadelphia having been issued
in 1935. Thus a pure random, or systematic random, selection
based on a physical listing was impossible. A random sampling
of the population directly would involve walking the more than
12,000 blocks in the city or the use of the patented Alfred Politz
mechanical guide, neither of which was economically feasible.
Stratified sampling with arbitrary selection was deemed pos-
sible but not satisfactory because of the danger of interviewer
bias and because of inability to establish sufficient controls over
respondent characteristics. Besides, the statistician was con-
vinced that area sampling would yield better results.
Area sampling, therefore, was chosen as the basic method,
because it could provide an up-to-date 'listing/' Use of San-
born maps, however, for automatic selection was far too costly
for this survey.
THE SAMPLE DESIGN
The 1940 Census for Philadelphia listed a total of 533,332
dwelling units, of which 506,980 were occupied in that year.
Using one interview to represent 1,000 dwelling units would
result in a primary sample of 533.
As a first step this sample was to be geographically stratified.
All Census subtracts were to be arranged in geographical order
so that any subdivisions later chosen would yield a solid block
of land rather than a collection of scattered locations. By
counting the dwelling units in each subtract during this process
and making each unit contain approximately 1,000 dwelling
units, the city would be broken into a primary sample of 533
areas, each with 1,000 dwelling units. Each area, it was be-
lieved, would be a more or less homogenous segment of popula-
tion with respect to socioeconomic factors.
Proportionality by income groups was then established.
Within each of the 533 areas in the primary sample, one block
was to be selected as the block in which interviews would be
made. This secondary sample of blocks was to be chosen on
the basis of the latest Federal Reserve Board data on incomes
132 MARKETING RESEARCH
through correlation of incomes with rental values. Thus, one
block would represent each area, and the total sample would
yield proportional representation to each rental (income) group.
For example, since Census data indicated that 18.5 per cent of
Philadelphia families were paying from $20 to $24 in weekly
rents, or owned homes with this rental value, 18.5 per cent of all
interviews would be made with families living in blocks in which
average rentals in 1940 were within these limits.
It was realized that both rentals and incomes had increased
since 1940 and that the increases had not been equal. No up-
to-date data were available which could be used to measure
these shifts. It was, therefore, assumed that the increases, al-
though not equal, were roughly proportional. To minimize
possible distortions caused by this assumption, it was planned
to divide the income classification into four groups of equal size.
Basing the selection of families to be interviewed on the char-
acter of the block itself at the time of interviewing would further
reduce distortion, it was believed.
SAMPLE CONSTRUCTION
Owing to the strong plea of the field supervisor not to scatter
interviews widely, plus other considerations, the sample design
was amended. Instead of using 533 areas, each page of the 1 52
pages of block listings in the 1940 Housing Supplement, Phila-
delphia Block Statistics, was used as a primary sample unit.
This accomplished the original objective but with fewer areas.
Directions followed in constructing the sample were:
1. Add all monthly rentals shown in the last column on each page of
Philadelphia Block Statistics, omitting cents. This is Total rentals.
2. Count the number of figures in this same column. This is Number
of rental figures.
3. Divide To tal rentals by Number of rental figures. This equals the
Average rental per block for the page.
4. Select that block on the page with the rental nearest to the Average
rental. Identify this block by average rental, Census tract and subtract
designation, and block number.
5. Add total dwelling units on page (column 4).
CASES IN SAMPLING 133
If the total is less than 1,500 dwelling units, one interview should be
made in the designated block.
If the total is between 1,500 and 2,499 dwelling units, two interviews
should be made in the designated block; and so on.
6. Identify each block with half or more nonwhite populations.
7. Identify each block with apartment houses (an average of four or
more dwelling units per structure).
This process yielded the following sample distribution by
rental groups:
Quota by Unadjusted
Rental Group Census Sample
$14 or less 42
15-19 60 23
20-24 99 129
25-29 93 127
30-39 125 152
40-49 58 62
50-59 25 10
60-74 16 15
75-99 8 6
$100 and over 7 2^
533 526
A check showed that the distortion which had resulted was
caused primarily by the wide geographical dispersion of most of
the low rental groups and some of the high rental groups.
Hence, these groups were lost in the averaging process. Using
533 areas, it was shown by a test, would reduce the distortion
only slightly. Instead, a method of adjustment was devised
by which all high and low rental groups, as shown in Table 2
of Philadelphia Block Statistics, were listed separately. From
these lists, representative rental groups could be keyed easily to
their proper areas, and thus to particular pages. Individual
blocks were selected in the same manner as described in step 4
above. In this way, proportionate interviews for the $14-or-less-
rental group were chosen, as well as interviews for the $1 5 to $19,
$75 to $99, and $100 and over groups.
After the extreme rental group quotas were adjusted, the bal-
ance of the sample was adjusted by minor shiftings to adjacent
groups. Thus, if fifty-eight interviews were needed in a group
and sixty-two had been provided by the sampling method, the
134
MARKETING RESEARCH
nearest sample block to the class limit was rechosen, the average
being taken as, for example, $39.43 rather than $39.60 for the
block. This completed the income stratification adjustment.
The adjusted sample distribution compared favorably with the
Census distribution when rental groups were combined into
quartiles, though distortions were still present when subrental
groups were considered.
The sample method used provided the desired quota of apart-
ment house interviews without adjustment. Nonwhite inter-
views came out to fifty-eight instead of seventy. To secure the
added twelve another Census subtract list was made showing
all subtracts in the city with 50 per cent or more nonwhites. By
shifting the average rental by a few cents in three areas the
necessary adjustment was made.
THE ADJUSTED SAMPLE
Despite these difficulties, the consensus was that sample
accuracy had not been seriously affected by these various adjust-
ments. The sample provided geographical and income (rental)
stratification, as well as controls by apartment dwellers and non-
EXHIBIT 1
Sample interviews
Total
Census
per cent
Apt.
Nonwhite
Rentals
of total
No.
Per cent
interviews
interviews
$14 or less
7.8
42
7.9
12
$14.50419.49
11.2
60
11.3
22
$19.50424.49
18.5
99
18.5
7
13
$24.50429.49
17.5
93
17.4
4
15
$29.50-$39.49
23.4
125
23.4
20
8
$39.50-$49.49
10.9
58
10.9
14
$49.50459.49
4.8
25
4.7
4
$59.50474.49
3.0
16
3.0
$74.50499.49
1.5
8
1.5
$99.50andover
1.4
7
1.3
4
Total
533
53
70
CASES IN SAMPLING 135
whites, which the nature of the food product and the evidence
of the preliminary investigations revealed were desirable.
A comparison of the adjusted sample distribution with that
of the 1940 Census is shown in Exhibit 1, p. 134.
Apartment house interviews planned were 9.9 per cent of the
total compared to 10.1 per cent in the Census. Nonwhite inter-
views totaled 13.1 per cent as against 12.9 per cent for the Cen-
sus, but owing to an oversight not discovered until too late, the
distribution of nonwhite interviews by rental groups (of ascend-
ing magnitude) was:
Census Sample
Rentals % of Total % of Total
$14 or less 178 17.1
$14.50419.49 19.7 31.4
$19 50-$24.49 28.4 18.7
$24.50429.49 18.8 21.4
$2950439.49 12.5 114
$39 50-$49.49 20
$49.50-$59.49 0.8
$59.50474.49
$74.50499.49
$99.50 and over
Interviewer instructions for selecting families were: walk
around each block assigned and select the typical structure
(either apartment house, as assigned, or family type); rotate
interviews north, south, east, and west as far as possible; rotate
position on side of the block with corner locations to be used
in about one interview out of ten; and take occasional upstairs
interviews in the two-to-four family structures.
ACTION NEEDED
Before launching the operation, the sample design outlined
above was reviewed by the researchers. In doing so, a number
of questions were raised concerning the merits of this design.
Specifically:
1. How good is the geographical distribution of interviews in
such a sample?
2. How likely is this design to provide effective representation
in the sample of different family sizes, occupations, religions,
136 MARKETING RESEARCH
and of other population breakdowns that might be relevant
in studying the use of these food products?
3. Does the design provide sufficient safeguards against bias
resulting from shifts in population density since 1940?
4. How well are the instructions for selecting families likely to
offset the possibility of interviewer bias?
5. Is this sample design needlessly complex?
A brief memorandum is needed evaluating the merits and
possible shortcomings of this design, with specific reference to
the five questions raised above.
CASE 4 DESIGNING A MULTIPURPOSE SURVEY
BACKGROUND
Very little has been said or done in the past about setting up
and conducting surveys that have more than one objective or
purpose. The design of such surveys requires a little more effort
and ingenuity than that put into the more common single-pur-
pose survey, but it is generally not an extremely difficult task;
and the return from a well-designed survey of this type, in terms
of the additional information gained as a result of the extra
effort, can hardly be better. For, once a survey is to be con-
ducted, any additional expenditures incurred in adding a few
extra questions to the questionnaire or in altering the sample
design are likely to be negligible in comparison to the original
cost of the operation.
The present case illustrates the sort of problems involved in
designing a multipurpose survey, the objective of which was to
supply information on consumers' attitudes toward a particular
product and to throw light on some questions of sample meth-
odology. As will be seen in the course of working out this case,
the design of such surveys is more a matter of ingenuity than of
technical sampling knowledge.
THE PROBLEM
In the summer of 1950 a research organization was asked to
conduct a survey to determine public attitudes toward prefabri-
CASES IN SAMPLING 137
cated housing and to determine the effect of advertising and
other prefabricated housing literature on these attitudes. The
survey was to be carried out in a Midwestern city of about
60,000 population, characterized by a heavy incidence of pre-
fabricated housing of a variety of brands and forms. The data
were to be collected by personal interview.
Since the chances of sampling bias were to be minimized,
the sample members were to be selected in some random man-
ner. In other words, a probability sample was to serve as the
core of the study. At the same time, however, it was desired,
for future reference, to obtain some information on the possible
bias involved in collecting data on the same subject by means
of a judgment sample. If the use of a judgment sample ap-
peared to produce little or no bias in the data, this method of
sample selection might be used in the future to obtain similar
information with less expense than is possible with a probability
sample. The problem, therefore, became one of laying out a
sample design that would yield accurate information on atti-
tudes toward prefabricated housing on a probability basis and
also provide evidence on the possible bias, if any, that might
result if the same data were collected through arbitrary selection
of sample members instead of through random selection.
ACTION RECOMMENDED
A number of suggestions were received regarding the manner
in which the study might be made. One suggestion involved
selection of the sample members from the population at large,
part by a probability method and part by leaving the selection
entirely to the interviewers. Another suggestion would have
set up a quota system for the judgment sample, giving the inter-
viewers latitude to select the sample members to fill the quotas.
Still another suggestion was that the population be divided into
two areas of more or less equal characteristics, with a probability
sample being selected from one area and a judgment sample
from the other area. An extension of this suggestion was that
the population be subdivided into three areas of more or less
equal composition; a probability sample would be selected from
one area, a judgment sample from another, and both probability
138 MARKETING RESEARCH
and judgment samples from the third. Such a design, it was
contended, would enable some evaluation to be made of the
effect of differences between the areas on the findings regarding
possible bias resulting from the use of a judgment sample.
Thus, there was general agreement that selection of part of
the sample on a probability basis and part on a judgment basis
was the most desirable way of ascertaining any bias that might
result in the latter case. Agreement was also fairly widespread
that some division of the population into areas or strata was
desirable as a means of insuring comparability between the prob-
ability and judgment samples. The differences arise in the
method of obtaining this comparability. In addition to the
suggestions in this respect cited above, two others deserve men-
tion. One would divide the population into about six or eight
areas of more or less equal composition and have every inter-
viewer conduct both probability and judgment interviews in
each area. The other suggestion was to make these areas as dis-
similar to each other as possible; that is, the composition of the
areas would be homogeneous within and heterogeneous with-
outthe opposite of the preceding schemeand every inter-
viewer would conduct both probability and judgment interviews
in each of these areas.
DATA AVAILABLE
With 1950 Census of Population breakdowns not available
at the time the survey was planned, not too much up-to-date
information about this city could be found. Its population
was known to be roughly sixty thousand. The city directory
supplied an enumeration of dwelling units and block listings as
well as occupational data by individuals.
On the basis of the cost and time allowed for the study, it was
estimated that about five hundred interviews could be con-
ducted. As many as sixteen experienced interviewers were avail-
able at the time for use in this study. No previously published
data describing surveys on people's attitudes toward prefabri-
cated housing could be located. However, the following refer-
ences, dealing with the bias in judgment samples relative to
probability samples, were found:
CASES IN SAMPLING 139
BROWN, G. H, "A Comparison of Sampling Methods," Journal of Marketing,
XI (April, 1947), pp. 331-37.
CHURCHMAN, C. W. (ed.). Measurement of Consumer Interest. Philadelphia:
University of Pennsylvania Press, 1947.
HOCHSTIM, J. R., and SMITH, D. M. K. "Area Sampling or Quota Control?
Three Sampling Experiments," Public Opinion Quarterly, XII (Spring,
1948), pp. 75-80.
ACTION NEEDED
The basic question is that of designing a sample that will
yield the desired substantive information, and at the same time
provide as closely comparable data as possible on the effects of
resorting to judgment selection. Specifically, the following
questions must be answered:
1. What sort of general sample design should be used: unre-
stricted, stratified, area breakdown, or what? If area or
strata breakdowns are to be used, on what principles
should these divisions be constructed? How will the prob-
ability and judgment methods of selection be fitted into this
design? Why is this particular design preferable to the
alternatives?
2. How should the total number of interviews be split up be-
tween judgment interviews and probability interviews?
How many interviewers should be employed in this survey?
How many interviews of each type should every interviewer
make?
3. Exactly what procedure should be used in selecting sample
members for the probability sample and for the judgment
sample?
CASE 5 DESIGNING A CONTROLLED
EXPERIMENT
BACKGROUND
The radio listenership ratings reported by a certain rating
service exhibited a declining trend over a number of early post-
World War II years. A user of this rating service was puzzled
140 MARKETING RESEARCH
by this phenomenon and desired to investigate it further. Was
this reported trend a true picture of what was happening in the
population, or could it be the result of some bias in the method
used by the rating service to collect the data? This was the
question that the user of this service asked his researchers to
answer.
THE PROBLEM
On an a priori basis, support could be found for either case.
The listenership surveys of this rating service were conducted in
large cities where television stations were in operation. Special
studies had shown that families purchasing television sets
tended to reduce their listening to radio. The reported declin-
ing trend in radio listenership ratings would seem to be entirely
in accord with these findings. On the other hand, sales of radio
sets had been well maintained during these early post- World
War II years. By 1951, about 42 million homes were estimated
to have one or more radios, the highest such figure ever reported.
Such a state of affairs would hardly seem to be consistent with
a decline in radio listenership.
If the method used by the rating service in collecting the data
was at fault, one of two recent innovations made by the rating
service in its procedure seemed to be the most likely culprit.
Both of these innovations had been made during the period
under study. One innovation followed from the introduction
of television when the rating service began to measure listener-
ship to television as well as to radio. This was done by altering
its former question on radio listenership to include television
as well. The former question was: "Were you listening to your
radio just now?" ( If the answer was "Yes," the respondent was
questioned about the name of the station and of the program.)
As revised, the question read: "Was anyone in your home lis-
tening to the radio or looking at television just now?" (If the
answer was "Yes," further queries were made about the program
and station.)
In this new question, the order of mention of radio and of
television was alternated. Nevertheless, the fact that listener-
CASES IN SAMPLING 141
ship to both media was included in the same question might, it
was felt, lead to underestimates of listenership to one or both
media.
The second innovation consisted of the addition of a ques-
tion asking the respondent about listenership to radio or tele-
vision in the household fifteen minutes before. In other words,
after the original question in its revised form, as shown above,
was read, and the accompanying questions on program and sta-
tion, the respondent was asked: "Was anyone in your home lis-
tening to the radio or looking at television between and
. . ; or about 15 minutes ago?" (If the answer was "Yes,"
there were follow-up queries on program and station.)
This procedure was used by the agency as a means of dou-
bling the size of the sample at a negligible increase in cost, for in
effect it provided twice as much data on listenership at a par-
ticular time, although there was no change in the number of
interviews made. Could it be, however, that this procedure
might bias the listenership ratings in some form or other?
Investigation failed to turn up any research done previously
on this subject, and hence the answer seemed to lie in designing
a survey that would supply the necessary data. Such a survey
would have to consist, in essence, of a series of controlled experi-
ments in which one factor would be altered at a time.
The exact form the controlled experiment might take was in
question. Suggestions ranged all the way from using just one
method in the survey that was, to ask about radio and tele-
vision listenership separately, followed by queries about listen-
ership fifteen minutes prior to the interview, in conjunction
with the listenership ratings provided by the service to using
four distinct methods. In the latter case, listenership data for
the same period would be collected from one group of respond-
ents by the same procedure used by the rating service, from an-
other group by the method proposed above, from a third group
by asking one question only about listenership to both media
fifteen minutes before, and from a fourth group also by asking
only about listenership fifteen minutes before but using separate
questions for radio and for television listenership.
142 MARKETING RESEARCH
ACTION RECOMMENDED
In the opinion of some, the rating service should be hired to
make this survey, so that one could be absolutely sure that the
survey procedures were the same on the controlled experiments
as in actual practice. Others maintained, however, that dupli-
cation of these procedures was no problem anyway inasmuch
as they were simple and widely knownthe interviews were
made by the telephone coincidental method and that the use
of an independent research agency was the only means of assur-
ing complete impartiality.
The design of the survey clearly had to be one where data on
radio listenership would be collected simultaneously in suffi-
ciently different ways to permit an evaluation to be made of
the effect, or lack of effect, of each of these two innovations on
the accuracy of the data. Question arose, however, over the
number of different program periods to be surveyed. These
program periods are generally fifteen minutes long in the day-
time and a half hour at night the frequency with which pro-
grams change. It was estimated that not more than 5,000 tele-
phone interviews could be completed with the resources avail-
able for the survey. Should these interviews be distributed
more or less uniformly over the day (and night) and over the
days of the week, as some suggested, or should they be concen-
trated in particular groups, as others suggested?
In this connection it is pertinent to note that there is some
tendency in analyzing listenership data to distinguish weekday
broadcasts from week-end broadcasts and morning, afternoon,
and evening broadcasts. It also deserves mention that the rat-
ing service conducted about four hundred telephone interviews
for each fifteen-minute broadcast period over an interval of two
weeks. It might further be noted that ratings of many daytime
radio programs are in the vicinity of 1 that is, 1 per cent of
those interviewed during the particular period were listening
to that program and that ratings on the most popular evening
shows are not likely to rise much above 10 or 12.
CASES IN SAMPLING 143
DATA AVAILABLE
The information on the conditions ancl the limitations sur-
rounding the survey has already been provided in the preceding
sections. One additional pertinent fact is that the survey was
to be concentrated in one city, a city for which this rating service
provided continuing listenership data.
No previous studies could be located on the effect of the two
innovations in question except for an oblique reference in the
following publication :
PAYNE, S. L. The Art of Asking Questions. Princeton, N. J Princeton Uni-
versity Press, 1951, pp. 69-70, 102-103.
ACTION NEEDED
In designing this survey, the following questions have to be
considered:
1. Who should be hired to conduct the survey the rating serv-
ice or an independent research agency? Why?
2. How should the controls be set up to provide all the neces-
sary data? Should the regular data provided by the rating
service be incorporated within these controls or should
duplicate data be obtained in the course of the survey by
the same method used by the rating service?
3. How should the sample be distributed by broadcast periods
over the week? Should the distribution be fairly uniform
over the week or should the sample be concentrated in cer-
tain periods? If the latter, on what basis should this con-
centration be established?
CASE 6 A PANEL OPERATION ON CONSUMER
DURABLE GOODS PURCHASE PLANS
BACKGROUND
One of the basic problems of marketing and economic anal-
ysis is that of forecasting consumer purchases. From the point
of view of attempting to maintain the economy on an even level,
144 MARKETING RESEARCH
the forecasting of consumer durable goods purchases is of par-
ticular interest because of the often erratic nature of such pur-
chases and because of their leverage effect on business plans and
business actions. Efforts to make such forecasts in the past
have centered largely on the derivation of statistical relation-
ships for recent years between aggregate expenditures on con-
sumer durables and other relevant variables (aggregate income,
time, and so on) and the extrapolation of these relationships
into the future.
In general, however, these efforts have not proved very suc-
cessful, and attention in recent years has been focused on the
possibility of forecasting such purchases by asking the con-
sumers themselves. This approach is predicated on the assump-
tion that purchases of the major durable goods, which involve
large sums of money, are generally planned well in advance, and
that by finding out about such plans it would be possible to fore-
cast actual expenditures for these durable goods. Even if a
direct correspondence between plans and purchases does not
exist, it is believed that some knowledge of changes in the out-
look of consumers with regard to business conditions should
yield valuable clues as to the future course of consumer durables
purchases.
A researcher obtained a grant of several thousand dollars to
conduct a pilot study on the value of this new approach as ap-
plied principally to major durable goods ("major purchases"
defined as those involving an expenditure of $25 or more). A
continuous consumer panel operation was indicated as the
means of testing the approach, since it permits checks to be
made on the realization of any stated purchase plans. Because
the focus of the study was on major consumer durable goods,
the project was planned so that scheduling interviews one
month apart represented a reasonable compromise between
cost considerations and the desire for accurate information on
the realization of purchase plans. On this basis, it was esti-
mated that approximately one year's observation with the panel
would be needed to check the validity of this new forecasting
approach.
CASES IN SAMPLING 145
THE PROBLEM
A distinct possibility existed that the type and frequency of
consumer purchase plans, as well as the relationship between
plans and realizations, might vary by different population
groups. The sample, therefore, had to be set up so that infor-
mation could be obtained on plans and realizations by popula-
tion groups (income, occupation of head of family, age of head,
family size, and so forth) as well as for the aggregate.
Another problem arose, namely, that of incorporating into
the panel operation a means of detecting and, if possible, elim-
inating the effects of sampling bias on the nature of the results
obtained. Sampling bias is particularly likely in a continuing
panel operation principally because of (a) the repetition of in-
terviews with the same respondents, and (b) panel mortality.
These factors are likely to distort appreciably the substantive
findings, and it was therefore extremely important that the
study be designed so that such effects could be detected and, if
possible, measured.
A third problem arising from the panel operation was that of
replacing families that drop out of the panel. Dropouts in such
a study could be expected to be substantial. On what basis
should such families be replaced? How frequently should re-
placements be made?
DATA AVAILABLE
The panel was to be set up in a city of some 65,000 people,
which in many respects was thought to be a fairly representative
American community. Preliminary investigations, however,
revealed that scant data were available on the characteristics
of the city's inhabitants. The study was designed in late 1950,
a time when the results of the 1950 Census of Population had
not yet been published. The 1940 Census data were judged to
be outdated inasmuch as the city had expanded greatly in the
intervening decade and had acquired many new industries.
A 1950 population estimate for the city was available. Also
available was a city directory for 1950 which, as far as could be
ascertained, was reasonably accurate and listed the city's in-
146 MARKETING RESEARCH
habitants and the marital status, occupation, and place of resi-
dence of each by blocks. A block map of the city was also
obtainable.
The preliminary investigation revealed little published in-
formation regarding sampling bias in continuous panel opera-
tions or regarding panel mortality rates. The only bit of
information even remotely related to this subject was that non-
response rates on the usual one-spot personal interview samples
generally vary between 10 per cent and 20 per cent.
Pretests with the questionnaire to be used in the study indi-
cated that between one hundred and one hundred twenty-five
completed interviews could probably be obtained each month
for twelve months with the funds made available for the project.
ACTION NEEDED
From the sampling viewpoint, the following questions had
to be answered by the researcher in designing the study:
1 . What sort of sampling design should be used? How can the
desired breakdowns be obtained when so little is known
about the population? How should the sample members
be selected?
2. What sort of sampling biases are likely to arise in the course
of the panel operation? How can the sampling design be
adjusted to detect such biases if they exist?
3. How much allowance should be made for panel mortality
on the various waves of interviews? On what basis should
replacement families be selected? How frequently should
replacements be made?
The researcher was asked to prepare a memorandum on the
design of the study and dealing with the above questions. How
would you answer them?
CASES IN SAMPLING 147
CASE 7 FOLLOWING UP ON A MAIL SURVEY
BACKGROUND
In the fall of 1950, the Bureau of Economic and Business
Research of the University of Illinois, in conjunction with the
Division of Marketing of the same University and with the co-
operation of the Associated Credit Bureaus of America, con-
ducted a survey of occupational credit ratings. This was a repe-
tition of an earlier survey on the same subject made by P. D.
Converse in 1940, and to maintain comparability with that
earlier study, the same procedure was followed in the later one.
Mail questionnaires were sent out to some eight hundred credit
managers half of them store credit managers and the other
half managers of credit bureausasking them to rate each of
some forty-odd occupations as good, fair, or poor credit risks.
Since the meaning of these adjectives undoubtedly varies from
one person to another, the respondents were asked to define
each adjective in terms of percentage of defaults. This device
not only served as a means of quantifying the occupational
ratings but also provided a vehicle for evaluating the reliability
and significance of the results.
A copy of one of the two forms of the questionnaire used in
the survey is reproduced on pages 148-49. The other form is
identical with the one shown except for the fact that the order
of listing of the occupations was reversed. This precautionary
measure was taken to forestall the possibility that the credit
ratings or the definitions of the adjectives might be influenced
by the order in which the occupations were listed on the ques-
tionnairea possibility that, unlikely as it may seem, proved to
be an actuality. However, since the forms were alternated
among the respondents, the resulting bias did not affect the
validity of the results.
THE PROBLEM
The questionnaires were mailed out through the office of the
Associated Credit Bureaus of America. This was done partly
to facilitate the mailing of the questionnaires and partly as a
,0 O
-1
I
"3
I
rt
1
w .a
&W
P
ES8
5; -^ Q 5 tc
s'> W (X'K
OH P,
.
-r a -
Q"&
fair, and poor c
pation objectivel
ease define
d rate each
ce is limited.
DO YOU DEFINE GO(
per cent of DEFAULTS:
if part (a) cannot be answ
W
a)
a
b
en
1.
c p o
sjh
< .r -"
a. IS
O
pa
2 43
P
-sis
Hi Illl
o
148
g .5
S3 B ^ '
siis
|D5
150 MARKETING RESEARCH
means of demonstrating to the credit managers the interest of
their association in the survey, an interest which, it was thought,
would bolster the response rate. At the same time, however,
the arrangement about the mailing was such that no identifica-
tion of the individual returns was possible except by type of
credit manager (store or bureau manager) and by city and state
of residence, as ascertained from the postmark.
The returns on the survey revealed that 25 per cent of both
the 400 store credit managers and of the 400 credit bureau
managers circularized sent in questionnaires. Although this
response was not out of line with the sort of response rates
generally encountered on mail surveys, nonrespondents never-
theless comprised most of the sample, and question imme-
diately arose regarding the representativeness of the returns.
Did the returns really reflect the views of all credit managers
in the sample and, more broadly, of those in the population,
with respect to the subjects under study? Was there any way
of determining quickly and economically whether or not the
returns were biased and, if so, in which direction were they
biased?
ACTION RECOMMENDED
Two alternative courses of action appeared to be open in
view of the limitations on further work. One alternative was
to do no further work at all on the matter of representativeness;
the other alternative was to send out follow-up questionnaires.
Following the first approach meant that the returns that had
been received would be analyzed and the results presented as
based on a judgment sample of unknown representativeness.
The rationale underlying this approach was that resort to fol-
low-up questionnaires in such a case was not sound because
there was no knowing what one had even if a substantial num-
ber of additional returns were received. Many of these returns
might duplicate those received on the original mailing, since
those respondents would also necessarily receive follow-up ques-
tionnaires. In addition, the fact that no classifying information
was requested meant that differences between the distribution
of the respondents on the two mail waves by selected character-
CASES IN SAMPLING 151
istics could not be used as a means of indicating the character-
istics of the nonrespondents, as is sometipies done.
In other words, it was contended that the additional returns
received by such a follow-up mailing were not likely to remove
any bias that might be present in the original returns, because
the type of person who did not respond in the first place, and
whose views might differ from those of the respondents, was
not likely to answer now. It was also contended that to send
follow-up queries to the respondents might create some irrita-
tion on their part toward the sponsors of the study.
On the other hand, those advocating the use of follow-up
questionnaires asserted that any additional information was bet-
ter than nothing. Although the new returns might point to
much the same results as the original ones a phenomenon that
admittedly might be due to duplication of returns rather than
to representativeness of the original respondents there was al-
ways the possibility that bias, if existing, might show up in the
form of sharp differences in the occupational credit ratings on
the two sets of returns. Other things being equal, the larger
the rate of response, the less the likelihood of the presence of
bias due to nonresponse. And although there was no way of
avoiding the receipt of follow-up questionnaires by credit man-
agers who had already responded, a properly worded accom-
panying letter, it was claimed, should hold duplicate responses
down to a negligible number and at the same time cause little
irritation among the former respondents.
DATA AVAILABLE
The necessary data pertaining to this survey have already
been given in the previous sections. In addition, a considerable
body of literature exists on the forms of bias likely to be en-
countered in a mail survey. The following references were
found pertinent to the problem at hand:
CLAUSEN, J. A., and FORD, R. N. "Controlling Bias in Mail Questionnaires/'
Journal of the American Statistical Association, XLII (December, 1947), pp.
497-511.
ROBINSON, R., "Five Features Helped This Mail Questionnaire Pull from 60%
to 70%," Printers' Ink, CCXIV (February 22, 1946), pp. 25-26.
152 MARKETING RESEARCH
SEITZ, R. ML, "How Mail Surveys May Be Made to Pay/' Printers' Ink, CCIX
(December 1,1944), pp. 17 if.
ACTION NEEDED
1 . If you were the researcher in charge of this study, would you
recommend the use of follow-up questionnaires or not?
Why? How would you answer the objections of those favor-
ing the other course of action?
2. If you do favor follow-up questionnaires, outline the pro-
cedures you would use to maximize response (nonduplicated)
and avoid irritating the sample members.
3. In either case, how would you interpret the data in your
analysis of the substantive results of this study?
CASE 8 TESTING THE VALIDITY OF A
JUDGMENT SAMPLE
BACKGROUND
In the spring of 1946 a qualitative study of magazine interests
of women was conducted under the sponsorship of the McCall
Corporation. The purpose of the study was to provide infor-
mation regarding some of the qualitative aspects of magazine
reading who reads magazines and why, and to provide com-
parison with similar studies made in 1939 and 1941. The ques-
tionnaire used in this study is reproduced on pages 154-56.
Interviews were scheduled in 153 communities in 32 states.
All in all, well over 5,000 completed interviews with women
were contemplated,
THE PROBLEM
The sample used in past studies of this kind, and scheduled
to be used in this one, was designed on a stratified basis with
judgment selection. To quote from the original report, the
sample was restricted "to contain only women who live, for the
most part, in urban places, who are at home when the inter-
viewer calls, and who show the interviewer any copy of one or
more of the 22 magazines on the list." (Italics included.)
Strata were set up by geographic and city-size divisions, and for
CASES IN SAMPLING 153
each of the 153 communities selected on this basis additional
strata were outlined, using marital status and socioeconomic
status (prosperous, comfortable, getting by, poor) as controls.
The reasons for using this sample design in the present study
were principally to maintain comparability with the earlier
studies and to reduce expenses; to conduct such a study on a
probability basis on a nationwide scale was felt to be beyond the
resources available for the project. In addition, it was reasoned
that the restriction of the study to questions of qualitative im-
portthe who and the why of magazine readershipshould lead
to little or no bias in the results because of the use of the judg-
ment method of selection.
Although this line of reasoning may seem plausible off hand,
no empirical evidence was available at the time the study was
planned to prove or disprove it. Therefore, some sort of test of
the validity of this point was deemed necessary in order to give
the results of the study maximum reliability. After some con-
sideration, it was decided to run a probability sample concur-
rently with a judgment sample in one of the communities
selected for study and evaluate the reliability of the national
judgment sample on the basis of a comparison of the results of
these two samples. In essence, this evaluation study had two
objectives:
1. To compare the findings based on a judgment sample with
those based on a probability sample with regard to both
factual data, such as possession of magazines, and subjective
data, such as reading interest.
2. To measure differences between those homes where a
woman was found at home and interviewed on the first call
and those homes where a woman was found at home and
interviewed only after one or more callbacks.
The problem was to set up a sample plan that would fulfill
these objectives.
ACTION RECOMMENDED
To minimize the cost of this experiment and to permit use
of ABC (Audit Bureau of Circulation) data on magazine circu-
154
MARKETING RESEARCH
IH cj to th to
fc
&
i ^s^i'l4l
Z* 1?llll'i|l!
S. H
z " 1 5^1
O c Es2 ; E 8*^^
r,
H
5 S?
IX I.UOQ JO IW
puiuj
uouduxons p
~ X
! ]1 !i
4 r 5
liin
SSiei
s
E
m
IS
1
e
American Horn* 2
i
1
CASES IN SAMPLING
155
156
MARKETING RESEARCH
tag*
dvtrtUnf M whoU inttrwdnc to
you, r would you ridwr tun no advtrtiMag, t
dom't BMk. M y diffwwKrf
lattiwting 10 1
Pnfw oo MhrMtiriag a
9*StfoR 9. Who dot* tfao principal pUiuung md burins *
food (or yoor f MBJIyr
Srff II I
Other 2
10. A you <*iafIly> *nployd <moi4. your
YM Pact Tin*
Own .
12. Numto In
Anatd Swvk)i
Number of children wukr 2
Number of child 2-12
NmatrofcMldrmlM0
Nuatwr of diildrm 17-20
Nonbroffaulwovr2O
Total in family -
Question 19. Numiwr fa hou**ho!d a
pwttion 14.
16 trnploywl
Number of DMlw m
Source : The McCall Corporation.
lation by counties, it was suggested that the study be made in
Milwaukee County, Wisconsin. Nearly 99 per cent of the
population of this county lives in the city of Milwaukee, which
facilitates the selection of a probability sample. It was also
suggested that ABC circulation figures be used to check on the
accuracy of the circulation estimates obtained by the probability
and judgment samples, even though those figures were between
one and three years old. If the samples were chosen with this
purpose in mind, however, there arose the problem of relating
the results obtained to those for the main study in which, it will
be recalled, sample members were restricted to women at home
when the interviewer called.
A recommendation was also made that in the interest of econ-
omy, the questionnaire used in this experiment be reduced
where possible as compared to the one used in the main study
CASES IN SAMPLING 157
(page 1 54-56). Much the same questions necessarily had to be
asked in this survey as were asked on the* nation wide operation,
but the question was raised whether use of the complete ques-
tionnaire was needed to produce the desired information.
DATA AVAILABLE
Data available for the city, or county, of Milwaukee did not
go much beyond estimates of its aggregate population, of distri-
bution of the population by socioeconomic level, and of mag-
azine circulation. Statistics on dwelling units by local areas and
blocks could also be obtained. In addition, preliminary cost
estimates indicated that roughly a thousand interviews could be
made by each method of selection with the available resources.
ACTION NEEDED
The principal task in this problem is to outline a procedure
for selecting the judgment and probability samples and securing
sufficient information to fulfill the stated objectives of the ex-
periment. In the course of preparing this outline, the following
questions have to be answered:
1. What strata should be used for the judgment sample?
What strata for the probability sample?
2. On what basis should the members of each sample be se-
lectedon the basis used in the nationwide survey, or on the
basis suggested in the "Action Recommended" section? If
the latter, how can the results be carried over to evaluate the
reliability of the data obtained on the major study?
3. To what extent, if any, can the questionnaire used in the
main study be shortened for the purposes of this experiment?
Which questions can be used to study the possible bias in
the judgment method in estimating facts, and which ques-
tions can be used to study bias arising in the judgment
method in obtaining subjective information?
4. How can the second of the two main objectives of this ex-
periment be attained with this sample design?
Part V
COLLECTION OF DATA
A listing of the principal methods used to collect data in
marketing studies was provided in the introduction to Part III
(pages 62-69). In the introduction to this Part, we shall re-
view these methods from the point of view of the problems and
pitfalls involved in the actual collection of data by these meth-
ods.
The three principal methods of collecting data are, it will be
recalled, by surveys, by observation, and by experiment, with
mail questionnaires, telephone calls, and personal interviews as
the main variants of the first method. However, from the point
of view of the physical processes involved in collecting the data,
and from the point of view of assessing field problems, the fol-
lowing fivefold classification would seem to be more relevant:
internal records, mail questionnaire, personal interview, tele-
phone calls, and mechanical recording. 1 The relative merits of
each of these methods are summarized below.
METHODS OF COLLECTING DATA: ADVANTAGES
AND DISADVANTAGES
INTERNAL RECORDS
Advantages
1 . Highly economical in terms of the variety of data and num-
ber of breakdowns obtainable.
2. Accuracy and reliability of data are generally very high.
3. Most easily kept up on a continuing basis.
4. Relatively few problems of data collection.
1 Other methods, such as inventory polls and group participation methods,
are not taken up in this review.
159-
160 MARKETING RESEARCH
Disadvantages
1. Danger of antagonizing officials of the company if demands
for data are excessive or not well justified.
MAIL QUESTIONNAIRES
Advantages
1. Economical, though sometimes nullified by very low rates of
response. 2
2. Wide geographic distribution, if this is desirable.
3. Elimination of interviewer bias.
4. Possible greater frankness of response on questions that
might prove embarrassing in a personal interview.
5. Certain segments of the population, such as higher income
groups, more easily contacted in this manner.
Disadvantages
1. Time consuming.
2. Danger of sample bias: returns may not be representative of
the population.
8 A well-conducted mail survey will usually have at least a 1 5 per cent return.
If we assume that a personal interview survey secures a 100 per cent return, a
personal interview would cost less than each mail questionnaire returned only if
the variable cost of the former was less than roughly seven times (100 divided
by 1 5) the variable cost of each mail questionnaire sent out. Variable cost means
the total unit cost less unit common costs. Common costs in this case represent
all costs common to both methods; that is, those costs that would be incurred
irrespective of the method by which the data are collected.
As an illustration, assume the following: 20 per cent return on mail question-
naires, variable cost per mail questionnaire, 20 cents. What is the break-even
point between personal interviews per return and mail questionnaires per return?
Solution: 100 per cent divided by 20 per cent equals 5 mail questionnaires that
must be sent out to receive one return. Five times 20 cents equals $1, the cost
of each mail questionnaire returned with a 20 per cent return and a 20-cent
variable cost per questionnaire. This means that for a corresponding personal
interview survey to be no more expensive, the cost per interview would have to
be no more than $1, the break-even point. The general principle, then, is that
the break-even point between the two methods occurs when the ratio of the
variable cost per mail questionnaire to that of the personal interview is the same
as the percentage returns for the mail questionnaires.
If callbacks and follow-up letters are used, an increase in unit costs for those
sent out will occur. On the other hand, callbacks and follow-ups will cause
returns to be higher, so that the proportionate increase in expenditure resulting
from follow-ups may be small relative to the increase in the rate of return.
COLLECTION OF DATA 161
3. Questionnaire must be relatively short.
4. Questions must be simple and easily answered. Open-end
questions not very effective.
5. No assurance that the individual addressed is the one who
replies.
6. Certain segments of the population excluded, such as illiter-
ates.
7. Follow-ups necessary to interpret omissions and detect non-
response bias.
PERSONAL INTERVIEWS
Advantages
1. Detailed information obtainable on behavior and attitudes.
2. The identity of the respondent is known.
3. Response rate considerably higher than on a mail survey,
which tends to improve representativeness.
4. All segments of the population are approachable.
Disadvantages
1 . Relatively expensive.
2. Danger of interviewer bias.
3. Problem of cheating and misrepresentation by interviewer.
TELEPHONE CALLS
Advantages
1. Perhaps the least expensive of all, in terms of cost per inter-
view.
2. Data quickly obtainable.
3. Random selection very easy; members of the population
known.
Disadvantages
1 . Method limited to telephone-owning homes.
2. Interview must be brief.
3. Questions must be simple, either of the "yes-no" type or a
type which requires short, factual replies.
4. Danger of sample bias even for population sampled because
of the influence of the time when call is made on those who
162 MARKETING RESEARCH
respond; for example, evening calls produce a larger number
of male respondents.
MECHANICAL RECORDING
Advantages
1 . High degree of accuracy of what is being measured.
Disadvantages
1. Interpretation of data may be perplexing. Thus, an audi-
meter may record when a radio is on, but does not indicate
whether or not anyone is listening.
2. Danger of respondent bias when respondent is conscious
that actions or words are being recorded. The use of tape
recorders in personal interviews is a case in point.
The Cases in This Part
The most commonly employed means of gathering data are
by using internal records, by personal interview, and by a mail
questionnaire. It is on these three methods that the cases in
this Part are focused. (Reference might also be made to Case
6 of Part II, which deals with the collection of data by another
means.)
Gathering data from internal records is the subject of the first
two cases. Case 1 poses the problem of designing forms for
such data, knowing what is wanted, and Case 2 deals with the
question of deciding what data should be gathered.
Gathering data by means of a mail questionnaire is the sub-
ject of Case 3, while Case 4 poses the problem of securing re-
spondent cooperation in a survey conducted partly by personal
interview.
The next three cases deal with the use of personal interviews.
Two of the cases are concerned with the training of interviewers,
perhaps the prime source of bias in this method. Case 5 bears
on the selection of interviewers for a life insurance survey,
whereas Case 6 deals with the associated question of the train-
ing of interviewers. Case 7 raises the question of the value of
a depth interviewing scheme, including some of the so-called
COLLECTION OF DATA 163
projective techniques, in evaluating the factors behind con-
sumer purchases of a drug product.
The last case in this Part requires a fair degree of original
thought, as it raises the question of devising a means of securing
data in a rather unique situation, that of detecting bias in a per-
son's response regarding family purchases and habits.
BIBLIOGRAPHY
AMERICAN MARKETING ASSOCIATION. The Technique of Marketing Research.
New York: McGraw-Hill Book Co., Inc., 1937. Chapters 2-11 give some
good examples of the psychological problems associated with collecting data
by interviewing methods.
BLANKENSHIP, A. B. Consumer and Opinion Research. New York: Harper &
Bros., 1943. Chapter 4 contains a brief discussion of the basic considerations
in using mail questionnaires, telephone interviews, and personal interviews.
BRADFORD, E. E. Marketing Research. New York: McGraw-Hill Book Co.,
Inc., 1951. Chapter 1 1 gives a bnef account of interviewing methods.
BROWN, L O. Marketing and Distribution Research. New York: The Ronald
Press Co., 1949. Chapter 22 has a fine discussion of various methods of data
collection.
FERBER, R "Which Mail Questionnaires or Personal Interviews?" Printers'
Ink, CCXXII (February 13, 1948), pp. 44 ff. A detailed evaluation is made
of the controversy between those who advocate personal interviews in prefer-
ence to mail questionnaires, and vice versa.
GOODE, W. J., and HATT, P. K. Methods in Social Research. New York: Mc-
Graw-Hill Book Co., Inc., 1952. Chapter 13 contains many interesting
examples of questions and answers that take place between interviewer and
respondent; the methods of handling interviewing problems are explained by
means of these examples.
HEIDINGSFIELD, M. S., and BLANKENSHIP, A. B. Market and Marketing Analy-
sis. New York: Henry Holt & Co., 1947. Major factors in methods of col-
lecting data are treated briefly in Chapter 9. Selection, training, supervision,
and compensation of interviewers are also covered.
LORIE, J. H., and ROBERTS, H. V. Basic Methods of Marketing Research. New
York: McGraw-Hill Book Co., Inc., 1951. Chapters 12-17 and 20-21 deal
in considerable detail with the problems of communication and observation
that are associated with collecting data, and with the problems of selection,
training, and supervision of interviewers.
LUCK, D. J., and WALES, H. G. Marketing Research. New York: Prentice-Hall,
Inc., 1952. Chapter 5 includes a schematic diagram of the relationships be-
tween various sources and methods of collecting data.
PARADISE, L. M., and BLANKENSHIP, A. B. "Depth Questioning/' Journal of
Marketing, XV (January, 19 5 1 ), pp. 274-88. The most comprehensive treat-
ment of depth questioning in marketing research problems is to be found
here.
PAYNE, S. L. The Art of Asking Questions. Princeton: Princeton University
Press, 1951. This entire book can be read with profit by research workers
164 MARKETING RESEARCH
who must prepare questionnaires or who need criteria for appraising the work
of others.
REED, V. D., et al. "Selection, Training, and Supervision of Field Interviewers
in Marketing Research/' Journal of Marketing, XII (January, 1948), pp.
365-78. An excellent treatment of the problems encountered in field inter-
viewing work.
WALES, H. G., and FERBER, R. Marketing Research Selected Literature.
Des Moines: Wm. C. Brown Co., 1952. A comprehensive bibliography on
interviewing techniques will be found on pp. 198-205.
Part V Cases
CASE I OBTAINING DATA ON THE
OPERATIONS OF A LAUNDRY
BACKGROUND
The Aswell Power Laundry Company was one of the three
major laundries serving the area within and surrounding a city
of about 100,000 population. Most of the company's revenues
came from laundry operations.
A number of smaller dry cleaning and laundry establishments
also operated in this area, but the principal competitors of the
Aswell Company were the two other major laundries, each of
which was about equal in size to the Aswell Company.
In 1946, the Aswell Company suffered a serious decline in
the volume of its operations. After the war, competing com-
panies had made substantial inroads into the business done by
the Aswell Company and, under the existing management,
there seemed to be little prospect of reversing the downward
trend.
This decline in sales, plus poor customer relationships, caused
the Aswell Company to discharge its manager and employ an-
other man to fill his place. The officials of the company gave
the new manager a clear mandate to do whatever was necessary
to lift the business out of its existing precarious position.
ACTION RECOMMENDED
Officials of the laundry were of the opinion that the major
problem faced by the company was to obtain enough business
to improve the profit situation. As a corollary to this, they also
insisted that it would be necessary to raise the profit margin
sufficiently to provide a more adequate return on the capital
165
166 MARKETING RESEARCH
investment in the business. In line with this objective, they
suggested that the following four questions concerning the
laundry's operations be investigated:
1. Is the profit margin of the business being influenced by poor
personnel relations between management and company
workers?
2. Is the equipment in the plant modern enough to handle a
sufficient volume to provide an adequate profit margin?
3. Do we need to increase the number of routes that we now
have in order to provide a more profitable operation?
4. Will the investment necessary to achieve some of these goals
be so high that it will be impossible to go ahead with an
expansion program?
THE PROBLEM
In searching for data with which to study these questions, the
new manager found that no formal reports of any type had ever
been prepared for officers of the company. Apparently, the
business had been operating rather successfully for a number of
years without any reporting system. Except for the usual laun-
dry sales slips, no records of any kind had been kept, and only
occasionally were profit and loss statements made up. The
company executives felt that it would be wise to engage in a
study of methods that could be used to provide adequate in-
formation for the proper management of the concern, at least
on those matters that had to do with policy. The board of
directors felt that its function should be that of determining
policy and that the manager should be the executive officer
in carrying out these policies.
ACTION NEEDED
In view of the above situation, the manager's initial task was
to determine (a) what data were required to enable a compre-
hensive study to be made of the laundry's present situation,
with specific reference to the four areas of investigation recom-
mended by the directors, and (b) in what manner these data
CASES IN COLLECTION OF DATA 167
should be collected. In other words, he had to prepare a sched-
ule listing:
1 . The data required for each area of investigation.
2. The methods and procedures by which such data were to be
obtained.
3. Other data that might prove useful in studying this general
problem and the methods by which they were to be ob-
tained.
CASE 2 DESIGNING FORMS FOR GATHERING
INTERNAL DATA
BACKGROUND
As part of a research study of the origins of branch store
patronage, it was planned to utilize the credit records of the
Boris Department Store. The data to be gathered concerned
credit purchases made by customers during specified time peri-
ods at the downtown parent store, at the suburban branch
stores, and in total.
From an alphabetical list of customers in the ledger books,
9,672 customer names had been drawn in clusters of 62 each.
A delivery department expert then went through these names
and addresses, separating out all those falling within the speci-
fied boundaries of the suburban branch store's trading area.
The total of names so selected was 1,360, though later elimina-
tion of accounts showing no activity during either time period
studied reduced the number to 1,276. This was the sample for
which internal purchase data were to be gathered.
NATURE OF THE RECORDS
Neither sales slips nor ledger records ("history" records) were
considered usable. Although sales slips offered the possibility
of securing additional worth-while information, particularly on
identification of items purchased, multiple unit sales, and iden-
tity of purchasers, the methods used in filing the slips made it
extremely burdensome and costly to locate and assemble the
168 MARKETING RESEARCH
number of slips involved. The ledger, or "history," records,,
showing monthly balances and payments for a period of five
years, were to be used for checking address changes during the
time periods studied, but they lacked the necessary identifica-
tion of departments patronized and individual purchases made.
The decision had been made, therefore, to use microfilm
records. These records reproduced the monthly statements
sent to customers (cycle billing not being used by the store).
Each month the statements of all those accounts in a ledger
(each ledger contained about one thousand names) against
which purchases had been made, past due bills rendered, or
credits issued, were microfilmed. The microfilm for the same
block (or ledger) of accounts the next month, and the next, was
spliced on later. Eventually, therefore, each microfilm stored
in a single box carried in three successive parts a record of all
account activity for accounts in a given ledger. As the store's
fiscal year began on February 1, the first roll of microfilm for a
given year and a given ledger would include all account activity
for, say, names beginning with Act to those beginning with As
for February, March, and April. Thus, Mrs. Bertram P. Arri-
more's account would occur on the film three times (if there was
activity each month) in three places and always in the same
sequence of names.
Because 1 56 clusters of 62 accounts each had constituted the
primary sample, and two yearly periods covering two quarters
each had been selected, this meant that 624 separate rolls of
microfilm would have to be used in the study. (Although this
appeared to be a large number of microfilms, it was far less than
the number that would be needed with use of a noncluster ran-
domly selected sample. The fact that placing microfilms in
the viewers, locating the names desired, and focusing the ma-
chines was the most time-consuming and costly operation, had
dictated, in large measure, the nature of the cluster sample de-
sign.) The periods chosen were November, 1949, to April,
1950, inclusive, and November, 1950, to April, 1951, inclusive.
The branch store had been opened early in October, 1950, and
a change in credit records made it impossible to distinguish be-
tween branch and parent store purchases after May, 1951.
CASES IN COLLECTION OF DATA 169
The monthly statements mailed to customers and reproduced
on microfilm carried the following data: a top section showed,
from addressograph plates, the name, address, and account
number of the customer. Column 1 indicated the number
of each department in which a credit purchase had been made
during the month. Column 2 identified the month, day, and
year of each departmental purchase (two or more sales slips from
one department, or multiple purchases on one sales slip, were
not always indicated separately). Column 3 gave an abbrevi-
ated description of the item or items purchased, such as hat,
men's white shirts. Column 4 showed the dollars-and-cents
totals for each purchase. Also, the statement showed balances
carried forward, credits, and outstanding balances, all of which
it had been decided to ignore for purposes of this study.
PROBLEMS FACED
Consideration of data desired and trial inspections of micro-
film records showed that the researchers faced a number of prob-
lems in establishing satisfactory procedures and designing forms
for recording and summarizing the purchase data.
Each department in the parent store was designated by a
three-digit number. Each department in the branch store had
the same number as its counterpart in the parent store except
that it was preceded by the number 1. Thus, the Men's Fur-
nishings Department in the parent store was designated as
Dept. No. 1 50 and in the branch store as Dept. No. 1 1 50. Un-
fortunately, many individual monthly statements when filmed
were allowed to slip slightly out of position. This resulted in
blurriness or in part of the department numbers in Column 1
actually being off the film. From an inspection of a sample
of films it was estimated that in nearly 10 per cent of cases
department numbers would be difficult or impossible to record
from the microfilms.
At least two methods could be used to assure, or increase,
accuracy of the department numbers. In nearly one third of
these cases of bad films, strangely enough the duplicate copy of
the microfilm in question made identification possible. More-
over, several members of the store's research staff were familiar
170 MARKETING RESEARCH
enough with items carried in both stores, with department num-
ber assignments, and with the usual methods of item identifica-
tion, that they could use the descriptions in Column 3 to make
accurate identifications of department numbers in nearly all
cases. These persons might be utilized either to refer back to
films or to examine the data abstracted from them by the orig-
inal data gatherers to make department identifications when
these were in doubt.
Another problem concerned the amount of data to be copied
directly from records with or without summaries. Ideally, the
data gatherers should copy all data desired without making any
computations. This would best assure accuracy of the work.
Individual accounts, however, varied widely in the amount of
activity. Although many showed only a few purchases, others
averaged two, three, or four purchases a month so that several
pages were needed to record all transactions for such accounts.
Whatever data might be abstracted initially from the microfilm
records would be usable finally only when summarized as fol-
lows:
For each month of the November, 1948, to April, 1949, period, the
total number of days on which purchases were made and the dollar
amounts (rounded to the nearest dollar);
For each month of the November, 1949, to April, 1950, period, the
total number of days on which purchases were made and amounts at the
parent store, at the branch store, and at both stores together; and for
both periods the grand total of "days purchased" and "amounts pur-
chased" for each breakdown.
Should data gatherers copy all data for each individual pur-
chase shown, with provision for comptometer operators, or
other specialists, to make the needed summaries later? Or
should forms and instructions be devised requiring data gather-
ers to make the appropriate summaries, or part of them, during
the data-gathering process? If the latter alternative is adopted,
could a method of internal checks be introduced to assure ac-
curacy?
It was estimated that some six hundred hours of work would
be required to abstract data from the microfilm records and to
CASES IN COLLECTION OF DATA 171
summarize these data. Plans were made to secure eight reading
machines to be used entirely by the data.gatherers.
ACTION NEEDED
At this stage it was necessary to:
1 . determine and lay out the work procedures to be followed,
2. design the recording forms to be used, and
3. prepare whatever instructions for staff workers were required.
CASE 3 MAIL QUESTIONNAIRE METHOD OF
SURVEY
BACKGROUND
To ascertain retailers' experience in the use of television as an
advertising medium, a mail questionnaire survey was made
among those furniture dealers who were using TV during the
third quarter of 1952. Also included in this survey were dealers
who used television in October, 1949, but who had discontinued
its use by October, 1952. This was done to obtain a more accu-
rate and impartial picture of dealer opinion of the usefulness of
TV advertising.
A list of TV advertisers in the retail furniture field was ob-
tained from a quarterly publication of the N. C. Rorabaugh
Company of New York City, The Rorabaugh Report for Tele-
vision Advertising. This publication was discontinued as of
September, 1952. It listed alphabetically by individual TV
markets all the local-retail advertisers active during the partic-
ular quarter covered.
From this report, 190 retail furniture dealers were found to
have used TV advertising at some time during the period from
July through September, 1952. Likewise, a check of those deal-
ers who had used TV advertising in October, 1949, but who no
longer were using it in October, 1952, revealed that 17 furniture
retailers fitted into this category.
To decide which firms should be considered as retail furniture
dealers, the classified sections of telephone directories were
used. Thus, firms reported by the Rorabaugh Company as
172 MARKETING RESEARCH
using TV were checked against those shown in the classified
sections as "Retail Furniture Dealers" before being included
in this study. This tended to exclude retailers who may have
had only one department handling furniture and who were
classified under some other more pertinent category in the tele-
phone directories.
Because of the nationwide distribution of firms using TV the
mail questionnaire method of securing data was used. The fact
that the questionnaire was relatively short and could be fitted
on one page facilitated the use of this method of data collection.
Of one hundred ninety questionnaires mailed, seventy-two (or
38 per cent) were returned; this includes responses received to
follow-up mailings. Of the seventeen furniture dealers who
stopped using TV, nine (or 53 per cent) returned their question-
naires.
THE PROBLEM
Because of the interest evidenced in this project, it was de-
cided to pursue a similar study in the fields of department store
advertising and financial organization advertising (banks and
savings and loan associations). This survey, like the preceding
one, was also to be on a nationwide basis, i.e., covering all the
TV markets. The problem therefore arose of identifying the
members of the population, that is, securing the names of de-
partment stores and financial organizations advertising over TV
during the period selected for study, which was the second
quarter of 1953.
A somewhat different way of solving this problem was called
for because of the discontinuance of the Rorabaugh Company
publication.
QUESTIONS TO BE ANSWERED
1. How could a listing of the members of this population be
obtained?
2. How could a listing be obtained of department stores and
financial organizations that had used TV as an advertising
medium but had discontinued it by the second quarter of
1953?
CASES IN COLLECTION OF DATA 173
CASE 4 SECURING RESPONDENT
COOPERATION
BACKGROUND
A marketing research class, as its project for the term, con-
ducted a study of patronage at local department stores in the
city. In this study, the class had the active cooperation of one
of the largest stores, with access to all store records.
Among the facts most important to the study planned, it was
felt, were the relative expenditures of different types of families
at each of the five largest department stores. Interviewing dur-
ing the preplanning stage led to the belief that it would be diffi-
cult to obtain accurate figures on such expenditures.
PLANNING
Two types of questionnaires were constructed a personal
interview questionnaire, to be used in interviewing city resi-
dents, and a mail questionnaire, to be sent to out-of-town cus-
tomers of the cooperating store. Several different questions
were to be tested to determine which would best provide an
accurate ranking of total cash and credit expenditures during
the past year at the five stores. As a further check on the accu-
racy of results, a special pre-interview mailing was devised. This
was to be sent several days before the date of interviewing. The
questionnaire requested information on department store pur-
chases made during October and November. Each purchase
by a member of the family was to be identified by date, store
name, article, price, and method of purchase (cash or credit).
The hope was that women would have to refer to statements,
sales slips, or other records, or that, at least, they would have
to reconstruct past expenditures with considerable accuracy.
Assurance that the credit purchases reported were accurate
when checked against store records would make it far safer to
assume that cash purchases (unverified) and expenditures re-
ported at other stores were valid. Respondents did not know
that any one store was particularly involved or that the answers
could be verified.
174 MARKETING RESEARCH
During the pretesting stage, the purchase-data form was
mailed to 100 picked credit customers of the cooperating store.
The following covering letter was used with this mailing:
EXHIBIT 1
Marketing Research Class
School of Business
Metropolis University
November 24, 19
Dear Madam :
We believe in learning by doing. We are conducting, therefore, an
actual and practical study as part of our class work.
The study concerns the services which department stores render you
and how they might serve you better. The five stores we are studying
in particular are .... We will greatly appreciate your telling us some
details concerning your patronage of any or all of these five stores. The
information we need is indicated on the attached page.
One of us will call on you during the week of November 29 to pick up
the facts you are giving us. We hope you will be willing then to answer
a few questions relating to your opinions of the services rendered you by
these stores. By helping us you will be helping yourself and other
shoppers.
Sincerely yours,
Members of Marketing
Research Class
RESULTS OF PRETEST
The pretest revealed several difficulties. When follow-up
interviews were attempted with these persons, fourteen said no
such letter had been received; twelve could not be located at
the address given; five persons said they had thrown away the
questionnaire; one was reported to be on her honeymoon; two
were dead; one was too sick to be interviewed; four said they had
lost it; six refused to fill it out; and nine said they would send it
in later.
CASES IN COLLECTION OF DATA 175
The forty-six completed questionnaires, however, when
checked, turned out to be highly accurate, They were consist-
ent with answers given in the personal interviews and, more
important, were very close to the purchases as shown by store
records. The only discernible error seemed to be the omission
of an occasional item or purchase.
Many student interviewers felt that the receipt of the letter
in advance had done much to encourage cooperation for the
personal interview. A few felt otherwise. It was decided to
use the premailing for the final interviewing but to make revi-
sions calculated to induce greater respondent cooperation.
THE SURVEYS AND RESULTS
The form for recording individual department store pur-
chases was individualized to some extent. Students were to fill
in the date, the inside address, and salutation (from interviews
assigned), and the date on which the interview was expected to
be held, as well as to sign each letter personally. The covering
letter used appears on p. 176.
Interviewing by previous classes had never received less than
90 per cent response. Results this time were severely disap-
pointing. Of the total city sample of eight hundred fifty-six
persons, eighty-seven had moved, or the address given was in-
correct; thirty persons claimed to be too sick to be interviewed
(a few were dead); one hundred five could not be located despite
as many as five callbacks (plus phone calls to arrange interviews);
eighteen widely scattered interviews were not attempted; and
one hundred thirty-three families refused to cooperate. The
final number of completed, reliable interviews was four hun-
dred sixty-seven, of which fifty-nine turned out to represent ac-
counts which were inactive at the cooperating store.
Unquestionably, the pre-interview mailing had backfired.
Apparently, when the family members discussed the form,
there was a sharp increase in their suspicion that this was really
some kind of credit investigation. At least it seemed to many
like "prying into personal affairs/' Reluctance to be inter-
viewed and on occasion actual hostility were reported. The
only fortunate aspect was that nonresponse cut across income
176 MARKETING RESEARCH
EXHIBIT 2
Marketing Research Class
School of Business
Metropolis University
January , 19
Dear Madam:
We believe in learning by doing. We are conducting, therefore, an
actual and practical research study as part of our class work.
The study concerns the services which department stores render you
and how they might serve you better. The five stores we are studying in
particular are .... We will greatly appreciate your telling us some de-
tails concerning your patronage of any or all of these five stores. The
information we need is indicated on the attached page. The facts that
you tell us will be treated confidentially.
I will call on you . . to pick up the facts you are giving us.
I hope you will be willing then to answer a few questions about your
opinions of the services rendered you by these stores. By helping us
you will be helping yourself and other shoppers.
Please be assured that I am not selling anything and that I have not
been hired by anyone. I am a student doing this as a part of required
classwork and not for money.
Sincerely yours,
lines and account sizes so that the sample shrinkage actually
caused little distortion by subgroups in proportionate terms.
The mail questionnaire, while setting no record in terms oJ
percentage of responses, worked well. Only one mailing was
made. Of the eight hundred twenty questionnaires sent tc
out-of-city families, 39 per cent were returned within sixteen
days which were reliable and usable. Final returns, including
those after tabulations were run, were 43 per cent. The cover
ing letter used with the mail questionnaire appears on p. 177.
REPRISE
Two years later, the marketing research class did a somewhal
similar study, this one on out-of-town patronage of local depart
CASES IN COLLECTION OF DATA 177
EXHIBIT 3
Marketing Research Class
School of Business
Metropolis University
January 10, 19
Dear Madam:
We are students doing an actual survey in order to learn market re-
search. Our success depends very much on you.
We are doing a practical, useful study of the five largest (name of
city) . department stores to learn the services these five stores
render to you and how they might serve you better.
Only a select few are being asked to cooperate by answering the ques-
tions on the attached page. It will take only a few minutes of your time,
but to us it means a critical part of our year's work.
We assure you that we have not been hired by anyone, that we are not
selling anything, and that none of us is being paid for this work.
Won't you please fill out the form now while it is before you.
Sincerely yours,
Members of Marketing
Research Class
ment stores. A two-page questionnaire was sent to 987 families
outside the city. The covering letter appears on p. 178.
One week after mailing (returns had exceeded 33 per cent
by that time) a follow-up mailing was used. This was sent only
to those who had not answered as yet. 1 Total returns at the
1 It was essential to the study that individual mail returns be identified so that
facts could be keyed with internal data from store records. A system of coding,
therefore, had been used. This consisted of: (1) assigning a number to each
name in the sample, (2) giving each letter on the questionnaire a number, (3)
putting pinpricks over a different letter on each questionnaire, and (4) being sure
that the correctly coded questionnaires got in the right envelopes. By using a
master code chart, identifications were quickly made of returned questionnaires,
and follow-ups could go only to those not answering.
178 MARKETING RESEARCH
EXHIBIT 4
Metropolis University
School of Business
January 3, 19
Dear Madam:
Experts are almost no help. Only you can tell us the facts.
We are students learning to do research by doing an actual, practical
consumer survey.
No matter how much you shop or how little, your preferences and
your actions determine how department stores try to serve you and other
shoppers. So in whatever amount or way you shop your answers are very
important to us.
We have arranged the questions about shopping so that you can an-
swer nearly every one with a simple check ( V ) mark. The many house-
wives who kindly helped us to make this questionnaire assure us it usu-
ally takes less than ten minutes to answer.
Those few minutes of your time, however, are vital to us. For on
your cooperation depends the success of nearly a full year's effort on our
part. We haven't much money to spend and can ask only a few repre-
sentative housewives. So we do need your help.
All you tell us will be kept confidential. No one will be bothering you
because of any facts you tell us.
Just in case you do not do most of the department store shopping in
your family, would you ask the person who does to answer the question-
naire?
We will be grateful indeed to you for checking your answers now and
mailing them to us in the stamped envelope enclosed. Thank you.
Sincerely yours,
Members of Marketing
Research Class
end of fifteen days were 55 per cent and (including returns after
the cutoff date) finally exceeded 62 per cent.
The follow-up consisted of a business-reply postcard. On the
reverse side of the card was this message:
CASES IN COLLECTION OF DATA 179
EXHIBIT 5
Dear Madam :
If you are not among the many housewives who have kindly helped us
by filling in and returning the questionnaire we sent, we would appre-
ciate very much your doing so now.
So that we may plan our classwork won't you please check one of the
blocks on the attached card and return it to us? Many thanks!
Sincerely,
The reply card was addressed to the instructor of the course
and contained this message:
EXHIBIT 6
Marketing Research Students:
(check one)
E] I have already mailed your questionnaire.
Q] I will fill it in and mail it very soon.
Q I have mislaid your questionnaire. If you will send me another
questionnaire and stamped envelope at the address below I will be
glad to fill it in.
Name
Address
An interesting aspect of this follow-up was that the great
majority of cards returned came in the stamped, self-addressed
return envelopes originally provided together with the com-
pleted questionnaires. The daily rate of returns, of course, was
rapidly falling off on the seventh day when the follow-up was
mailed. Returns immediately shot upward before again taper-
ing off. Relatively few persons were motivated to check Block
3, but the inexpensive follow-up unquestionably boosted re-
turns.
180 MARKETING RESEARCH
ACTION NEEDED
1. What other methods might have been used to encourage
respondent cooperation?
2. How could the appeals used be classified or characterized?
Were they the right appeals for the situations?
3. What reasons could be given for the failure of the pre-inter-
view mailing device used?
4. How might the letters used be revised?
5. What advantages, if any, does a school have as compared to
a company or a private research agency in securing coopera-
tion?
6. Would the appeals used have to be changed to adapt them
to company or agency use? If so, why and how?
CASE 5 SELECTING INTERVIEWERS FOR A
LIFE INSURANCE STUDY
BACKGROUND
The key questions from a preliminary questionnaire that was
to be used in a pilot personal interview survey of consumers are
reproduced on pp. 181-83. The purpose of the survey was to
investigate the extent to which consumers understood the uses
and kinds of life insurance that could be incorporated in a
family budget and to determine how well life insurance agents
were working with prospects and clients in a full utilization of
life insurance policies.
The sample for the study was to be selected by use of a ran-
dom sampling numbers table and a city directory published by
Polk & Co. This would provide a random sample which would
then make it possible to estimate the sampling error. It was
estimated that about seven hundred families would have to be
selected in order to provide an adequate sample for the study.
THE PROBLEM
A trial run with the questionnaire revealed that consumers
did not know what kind of policies they had, and that they f re-
CASES IN COLLECTION OF DATA 181
quently would ask questions of the interviewer that were of a
technical nature. In addition, when questions were asked con-
cerning the investment opportunities that life insurance offers,
many irrelevant answers appeared. Other questions that raised
problems were such as this: ''What kind of policy or policies do
you now have?" Answers were confusing because the respond-
ent might say, "I think it is twenty-payment life." When asked
to produce the policy, the respondent might have a twenty-year
endowment policy.
It was clear, therefore, that a very careful and conscientious
job of interviewing would have to be done in order to obtain
reliable information on the subject. Approximately 20 minutes
EXHIBIT 1
Questionnaire for Insurance Survey
I. Life Insurance as a Tool for Savings and Investment
A. (1) Do you operate on a family budget?
(2) Does this budget include a fixed allowance for savings?
(3) (If answer to 1 or 2 is no) Do you manage to save each
year?
B. (1) Where or in what forms do you think your savings are best kept?
(Number in order of importance)
( ) Savings Banks ( ) Real Estate
( ) Government Bonds ( ) Cash
( ) Corporate Secunties ( ) Insurance
( ) Own Business ( ) Building and Loan
( ) Other
(2) What in your opinion is the most important quality of a savings
medium?
( ) Safety of Principal ( ) Inflation Hedge
( ) Liquid"
( ) Yield
( ) Liquidity ( ) Others.
' ' Yield
C. (1) In your opinion, does life insurance offer an opportunity for saving
money? ( ) Yes ( ) No
(2) Is there any life insurance now in force on your life or on other mem-
bers of your family? ( ) Yes ( ) No
(3) How many different individual life insurance policies are in force on
the members of your family?
182
MARKETING RESEARCH
*3*
. 8 S3
;
**
*~* i
O
i
s-
CASES IN COLLECTION OF DATA 183
E. Do you feel that any of these policies represents a fund upon which you
can draw in an emergency? Yes
No
F. Do you feel that any of these policies represents a fund which you can use
for retirement purposes? Yes
No
G. Are you making payments on your house?
H. 1 . Have you made plans for financing your retirement?
2. (If answer is Yes) What plans have been made?
3. (If answer is No) Why haven't plans been made?
II. A. Has an insurance agent ever offered to coordinate your life insurance into
an over-all protection plan? Ye
If Yes
a) Was this plan made? a) Do you feel that such a service
Yes would be useful for you?
No Yes
b) (If Yes) Was any of No
your insurance bought b) (If No) Why not?
to fulfill this plan?
No
c) Has any agent offered to
bring your plan "up to
date?"
No
B. 1. (If there is no life insurance in force in family) Have other plans been
made for maintaining family financial security in event of the death
of the breadwinner? Yes No
2. (If answer is Yes) What are these plans?
3. (If answer is No) Why haven't plans been made?
were required for each interview, on the basis of the trial run.
After allowing for traveling time and callbacks, it was estimated
that as few as fifteen interviewers might complete the study
within the period allowed if each had a reasonably heavy load.
Alternatively, the number of interviewers could be increased,
say doubled, each assigned so many less interviews, and the
study could be completed so much faster.
184 MARKETING RESEARCH
DATA AVAILABLE
The following reference was found to contain material per-
tinent to this problem:
REED, V. D., et al. "Selection, Training, and Supervision of Field In-
terviewers in Marketing Research/' Journal of Marketing, XII (Janu-
ary, 1948), pp. 365-78.
ACTION NEEDED
1. What principle should guide the researchers in determining
the number of interviewers to be hired for this study fewer
interviewers and more interviews per interviewer, or more
interviewers, lighter interviewing loads, and speedier com-
pletion?
2. What qualifications should the interviewers have for this
assignment?
3. What sort of training should the interviewers be given?
CASE 6 INSTRUCTIONS FOR INTERVIEWERS
INTERVIEWER PREPARATION
The questionnaire for a local market study on luggage had
been pretested, revised, and then reproduced in quantity. The
.sample design had been laid out, and interviewer assignments
had been completed.
Interviewers listed in the files of the research group then were
called. A meeting was arranged for those hired. Nearly all
those selected had had considerable interviewing experience.
Many of them had worked on the pretest.
The practice of this organization was to provide training for
new interviewers. This included brief classes in interviewing
techniques, covering such problems as: how to approach re-
spondents, encourage responses, and maintain interest; causes
.and correctives for bias; how to record answers to various types
of questions, handle questions asked, and fill out reports; meth-
ods used to gauge the quality of work performed and to check
t>ack on interviews held; and compensation. New interviewers
CASES IN COLLECTION OF DATA 185
also might be accompanied by a field supervisor for early inter-
views on their first jobs. Interviewers were compensated for
time spent in such training.
On most assignments interviewers met together before the
field work began. They then were given their assignments,
sufficient copies of the questionnaire, and a sheet of instructions
for the particular job. The supervisor made a few remarks
about the purposes and methods of the study, any particular
problems that were anticipated, and deadlines and compensa-
tion. Sometimes a trial run of the questionnaire was made be-
fore the entire group. Most of the time, however, was spent
in answering questions concerning the questionnaires and in-
structions distributed for the job at hand.
THE SAMPLE
A combination area-quota sample had been designed. Using
the latest Federal Reserve Board figures on family income dis-
tribution, five income groups had been established for the city
being sampled: the A Group (15 per cent of all families) with
incomes of $5,000 or more; the B Group (28 per cent) with in-
comes of $3,000 to $4,999; the C Group (22 per cent) with in-
comes of $2,000 to $2,999; the D Group (20 per cent) with
incomes of $1,000 to $1,999; and the E Group (15 per cent)
with incomes under $1,000.
Using Census data, each income group was considered equiv-
alent to several average rental groups. Each of the 388 Census
subtracts in the city then was assigned to a rental group based
on the average rental for the tract. From these listings random
choices of tracts were made of sufficient number to be propor-
tional to the size of each income group. The 34 subtracts thus
selected became the primary sample units.
Using Census tables and Block Statistics maps for the city,
four city blocks were selected from each of the 34 subtracts
which had rentals within the class limits but which represented
the spread of rentals as well as the average rentals of all blocks
in the tract. Thus, 1 36 city blocks were chosen as secondary
sample units. Four interviews were to be made in each block
chosen.
186
MARKETING RESEARCH
The sample finally was stratified by family size. Each inter-
viewer was given an individual quota (varying among the in-
come groups) for the interviews to be made with families of
given sizes. These quotas, in total, were distributed as follows:
sixty families of one member, one hundred thirty-five families
of two members, one hundred nineteen families of three mem-
bers, one hundred families of four members, and one hundred
thirty families of five or more members. This distribution was
based on Census figures for income groups.
THE QUESTIONNAIRE
EXHIBIT 1
Questionnaire for Luggage Survey
1. Does anyone in your family, in this house, own luggage? Yes ] No
D.K.n
2. How many pieces of luggage are owned? Number _ D.K. Q]
. _
3, A Would you please identify these pieces on these photographs? The pic-
tures are the only ones we could get. They illustrate shapes and sizes
to help identify the types you own. The appearance does not matter.
3A
4B
TYPE
Number of each
type owned
Leather covering
Fabric covering
(airplane)
Cardboard covering
..covering
Matched set pieces
A B C D E F
G
H
Other
3. B What type of covering do these pieces have? ( cardboard, fabric, leather,
imitation leather, or other) RECORD IN TABLE ABOVE. D.K. Q
4. A Are any of these pieces parts of matched sets? Yes F] No F] D.K. Q
B If YES; Which types? RECORD ON LAST LINE OF TABLE.
..
5. A Which of these pieces you have identified was bought last? Type - .
D.K. Cl
B What type of covering was it? Type of covering - D.K. Q]
C When was it bought? Month _ Year _ D.K. []
D Whom was it bought for? Self Ph husband PI; or wife f]; other -
D.K.Q
CASES IN COLLECTION OF DATA 187
E Which person bought it? Self Q; husband Q; or wife Q; other
D.K. Q
6. A Have you, or others in the family, ever giv*en new luggage as a gift to
persons outside the home? Yes Q No Q D.K. p
B -If YES, ASK: When was the last gift bought? Month Year
D.K.Q
C Was trie gift for a relative Q or a friend p? D.K. Q
D For what occasion was it given? (Anniversary, birthday, Christmas,
going-away, graduation, wedding, Father's Day or Mother's Day.)
other . D.K.Q
E What type was it? Type D.K. Q. What type of covering
did it have? Type of covering DiK. Q
7. A Has anyone in the family ever received luggage as a gift from persons
outside the home? Yes Q No Q D.K.Q
B If YES, ASK: When was the last gift received? Month Year
D.K.Q
C For what occasion? ( Christmas, etc. ) D.K. Q
D What type was it? D.K. Q : What covering? D.K.Q
8. AWhich is the oldest piece of luggage in the family that is still used for
trips? Type. ~ ""
_ .
B What type of covering has this piece? Type of covering
D.K.Q
C In what year was it bought, or received as a gift? Year
,
9. A When was the last time any member of the family used luggage?
Month _ Year _ D.K. Q
B Which family member or members? All Q; wife Q; husband Q;
Other _ D.K.Q
C What kind of trip was it? (Vacation, business, weekend, or visit?)
Other _ D.K.Q
10 A Is any piece of luggage in the family used by more than one member of
the family? Yes Q No Q D.K. Q If YES, ASK: All pieces Q;
some pieces Q; or only one piece Q? D.K. Q
B Is any luggage ever lent to persons outside the home? Yes Q No Q
D.K. Q
11. A If you were to buy luggage soon what type would it be? Type -
D.K.Q
B What kind of covering? Covering - D.K.Q
C What color? (Brown, tan, black, grey, blue, white.) Other -
D.K. Q
D Approximately what price would you be willing to pay? Price $ -
D.K.Q
E How many years would you expect such luggage to last? Years -
D.K. Q
12. Does anyone in the family actually plan to buy any luggage this year?
YesQ NoQ D.K.Q
13. AWhat color is the family luggage? (Brown, tan, black, blue, grey)
D.K.Q Other _
B .How many pieces, if any, are striped? Number - D.K. Q
188 MARKETING RESEARCH
14. Would you tell your personal preferences as to using luggage with :
A Compartments for many different items F] or only a few compart-
ments p? D.K. [~| N.P. n
B Cloth linings with cloth pockets IT] or without cloth pockets
D.K.n N.P.Q
C- Your initials on it Q, or without initials p? D.K. Q N.P. [J]
1 5. Is there anything about your present luggage which you dislike or which you
have found of no use? Yes Q No D D - K - D If YES > ASK: What?
IDENTITY BE SURE TO COMPLETE THIS SECTION IN
EVERY CASE.
Census Tract and Block Number Fa . Type
Name of person interviewed: Mr., Mrs., Miss
Address Estimated Age
How many persons are there in your family (in this household)?
Wife Q] Husband Q] children 18 or over, number ; other
children, No. Fl; other adults, No. n. Total No
No. No.
Comments (Use back of paper also) :
EXHIBIT 2
Instructions for Interviewers Luggage Survey
Please read carefully and fully. Success in this project depends upon your
complete understanding of questionnaire procedures.
1. Purpose of the survey. The purpose of our survey is to determine the size
and nature of the market for luggage, and to measure the effect on luggage pur-
chases of income, family size, travel, gift-buying, joint-use, durability, and color,
material and construction preferences. The survey is expected to aid manufac-
turers and merchandisers in designing, promoting, distributing, and selling
luggage.
2. Who is to be interviewed. Interviews should be made with heads of house-
holds, either men or women. Presumably most interviews will be with house-
wives. Avoid interviewing other persons, or more than one person at a time.
3. When interviews should be made. Tuesday, Wednesday, and Thursday
between 10 a.m. and 4 p.m. are the best days and hours, but successful interviews
are possible at nearly all times.
4. Where interviews are to be made. Each base assignment includes 16 inter-
views in four blocks adjacent to or close to each other.
CASES IN COLLECTION OF DATA 189
It is very important that interviews be made in the blocks specified on the
individual assignment sheets. To be certain of identification, and in order to
choose typical dwellings, you should walk around each block before beginning
the interviews.
Wherever possible four interviews should be made in each block. The inter-
views should be distributed also. Make one interview on each side of the block
unless the typical structure of the four-sided block is not located on all sides.
Where you find two-story structures to be typical, half of the interviews should
be with families on the upper floor. Make only one of the sixteen interviews per
basic assignment in a corner house.
The family-size quota may be distributed among the four blocks in any way
found easiest. A recommended way of finding families of the size required is ta
ask interviewees about neighboring families. Please be certain, however, that
the quota is accurately filled. The block-assignment sheet may be used for a con-
venient check sheet. When any one quota is full be sure to ascertain family
size before beginning each of the later interviews.
5. Reliability of interviews. Try to complete every interview that is successfully
begun. If you believe any interviews to be unreliable, mark them as unreliable
together with a brief comment. All interviews, however, count as part of your
quota.
6. Entries on questionnaire before interviewing. Enter the census tract and
block number from your assignment sheet and the family type (A, B, C, D, or
E) on the line provided in the final Identity section.
7. Opening the interview. The following formula usually is successful:
"How do you do? I am , a research worker ( . ).
I am making a survey. Will you give me your opinions on a few questions? It
will not take long."
If any hesitancy is shown you might add: 'The questions are about luggage"
showing the photographs at once "It will take only a few minutes of your
time. May I come in?"
You may find by experiment that another opening works best for you.
Try to conduct all interviews inside the house.
When the person agrees to cooperate, give the following definition of luggage.
"The kind of luggage we are interested in is the luggage in which you put
clothes or personal belongings when going on a trip r and which you can carry
with you (not trunks)."
Read each question exactly as it appears on the questionnaire. Ignore blocks
Sin reading the questions. Do not rephrase questions. Do not use introduc-
y or explanatory remarks. If question is not understood, read it again. If
still not understood, pass on to the next question. (Words in capital letters are
for your aid and should not be read to respondents. )
Use a soft pencil. Fill in blocks with a small "x." Lines are provided for
write-ins. Answers to questions 6-D (on occasions for giving gifts), 9-C (type of
trip taken), 11-C (on color preferences), and 13-A (on color of luggage owned)
may be recorded by underlining the occasion or colors listed.
Any spontaneous comments about luggage which seem interesting or impor.
tant to you should be recorded almost verbatim on the back of the ques-
tionnaire.
190
MARKETING RESEARCH
8. How to handle individual questions.
Q-l . If answer is Don't Know
(1) The person may be confused. Repeat the definition for luggage.
(2) You may be talking to the wrong family member try to see the
housewife. If you can get only a DK answer, the interview is no
good. Try elsewhere.
If answer is NO. This is a valid answer. Repeat the definition for
luggage to minimize chance of confusion. If answer is still "NO" BE
SURE to get name, address, estimated age, and family size. Report all
such interviews.
Q-2. Accept estimates or guesses as to number. If person delays in answering
or takes too much time deciding, go right on to Questions 3 A and B. If
more than 6 pieces are owned, identify only as total number of leather,
fabric, or other coverings. Do not try to identify each type as to A, B,
etc. Record in "Other" space of block.
Q-3. This is a crucial question. Be as careful as possible in making accurate
records.
If persons say a piece of luggage is "not exactly like" any of those shown,
say "Would you tell me which type shown your piece of luggage most
resembles?" Be especially careful to do this if a "Gladstone" is men-
tioned. Straps are unimportant details.
In recording use the following system of digit recording. EXAMPLE :
to record ownership of two type A (one of genuine leather and one of
fabric) and one type C (fabric)
A
B
C
vv
V
V
V
V
Q-4. If matched sets are owned (even if only 1 piece) record types (as A, B,
etc.) by digit marks in the last line of the Table.
Q-5. The purpose of this question is to secure data on the last purchase made
by the family and for the use of the family. Gifts, however, may cause
confusion. A purchase for a gift to a person outside the home should
be recorded in Question 6. The receipt of a gift should be recorded in
Question 7.
NOTE: In all cases when identity of type and/or covering is certain, do not
repeat the question for 5-B, 6-E, 7-D, 8-B, 10-A for type and covering
identity.
Q-6-D. Read the parenthetical listing of occasions only if the person answers
"Don't Know" or is confused.
CASES IN COLLECTION OF DATA 191
Q-7. If answer to 7-A is NO, proceed immediately to question 8-A.
Q-8-C. If respondent says "about 1924," or can only estimate the year, indi-
cate by such notation as following: 1924 (about) or 1924 (cert.).
Q-9-A. The last use of luggage for any kind of trip or visit away from home
is desired.
Q-10-A. Any form of joint use should be counted: separate use of same piece
or sharing of one piece at one time.
Q-ll. Ask every part of this question, irrespective of the answer to 11 -A.
The question should cover actual expectation if a purchase for any
reason is actually contemplated. If no purchase is contemplated,
then the question should express the respondent's personal prefer-
ences and opinions.
Q-13-A. Underline each color represented in the family luggage. Do not
record the number of each color or seek to find the predominant
color.
Q-14. Distinguish between "Don't know" and "No preference" in record-
ing answers. Write in the answer if the respondent says to 14-A
"No compartments" or to 14-B "No cloth linings."
Q-l 5. Indicate all pertinent remarks and record as many complaints as the
respondent mentions. Use the comment section at bottom of page,
or the back of the page, if more space is needed.
Identity
Census tract, Block Number, Family Type: Enter these data before the inter-
view. Check against the assignment sheet immediately after leaving.
Get name whenever possible either by question or from mailbox, etc.
Estimate (do not ask) age of respondent.
Be sure to get family size and membership identification accurately.
EVALUATING THE INSTRUCTIONS
Various criteria are used * to evaluate and prepare instruc-
tions. Some points to be covered, as given by Luck and Wales, 2
are:
1 . What the survey is all about.
2. When the survey is to start and be finished. When to call
on respondents.
1 See J. H. Lorie and H. V. Roberts, Basic Methods of Marketing Research.
(New York: McGraw-Hill Book Co., Inc., 1951 ), pp. 391-96, and L. O. Brown,
Marketing and Distribution Research (New York: The Ronald Press Co., 1949),
pp. 500-6.
1 D. J. Luck and H. G. Wales, Marketing Research (New York: Prentice-
Hall, Inc., 1952), pp. 246-47.
192 MARKETING RESEARCH
3. How many persons are to be interviewed, where and how
to select them, and what to do about persons not at home.
4. How to introduce one's self and initiate the interview.
5. How each question should be asked and which ones are
contingent on the answers to other questions.
6. Methods of probing, encouraging responses, and aiding
memory.
7. If any items are to be observed, what is to be noted.
8. How each questionnaire is to be studied and corrected be-
fore returning.
9. What to do with completed questionnaires.
10. When and how she (the interviewer) will be paid for her
work.
11. The exact basis on which the quality of her work will be
appraised.
Bennett 3 reports that frequent criticisms made by field work-
ers about instructions are: "too wordy/' "omissions/' "too
fussy/' "ambiguous," and "insults intelligence."
How do the instructions for this survey rate in the light of
such criteria and criticism?
CASE 7 DEPTH INTERVIEWING
BACKGROUND
From German processes secured through the Alien Property
Custodian the Rilms Chemical Company had developed, and
soon after the war introduced, TONE-U into the market dom-
inated by the aspirins and seltzers and occupied by many other
preparations, or home remedies, for headaches, stomach pains,
and other minor ailments.
The product manager for TONE-U felt that the advertising
copy was too imitative of that of the leaders in the field. More-
over, he was not certain that anyone really understood the phys-
8 A. S. Bennett, Report on Researching Researches (New York: A. S. Bennett
Associates, 1948), p. 12, as summarized in L. O. Brown, Marketing and Distribu-
tion Research (New York: The Ronald Press Co., 1949), p. 500.
CASES IN COLLECTION OF DATA 193
ical and psychological characteristics of users or their ways of
thinking of their ills and their self-medication of them. Al-
though TONE-U was still a small factor in the market, its sales
were increasing steadily, and the company was willing to devote
considerable funds to building volume. The product manager
felt confident of larger advertising appropriations to come. He
was willing, therefore, to spend several thousands of dollars on
marketing research if he could be convinced that a practical
study of consumer psychology could be made.
Together with an executive of the advertising agency he had
several conferences with a research agency which had consider-
able experience in psychological and social studies. Encour-
aged by these discussions he paid a fee of several hundred dollars
so that the agency could prepare for him a specific proposal for
a study.
THE STUDY PROPOSED
Purpose:
We will study three groups of people: (1) users of TONE-U to deter-
mine what characteristics they have in common, (2) users of competitive
products to ascertain any important differences, and (3) nonusers. The
essential question is why users buy TONE-U and what it is in their feel-
ings, habits, and attitudes which accounts for their purchase. Knowing
these things, the manufacturer and advertiser can direct their work to
communicate more effectively with the users of TONE-U, to compete
more effectively with related remedies, and to attract those nonusers
who can be expected to find it helpful.
Sample:
We shall interview approximately 300 persons from the main stream
of American life, divided approximately equally among the three groups
mentioned.
Methods:
We shall ask these persons questions about the kinds of medicines
they keep in their homes; what kinds they use only when a doctor tells
them to; what kinds they use on their own judgment; what they think
about home remedies and what experience they have had with them;
what they think about the way these preparations work; why so many
people use aspirin; what kinds of people they know who do (and who
194 MARKETING RESEARCH
don't) use medicine and to describe a particular person as an illustration;
what kinds of disease they think are the worst; and to tell us a story about
a picture which we shall show them. We also shall get information from
them about their occupations, living quarters, education, and where they
came from. Our interviewers will take notes on the kind of furniture
they have and the impressions their homes give to a visitor.
We shall listen to and take notes on what they think about doctors,
medicines, illness, doctors' patients, self-diagnosis, sick children, and all
the related topics that come up some because we ask about them, some
because questions we ask bring other things to mind and we want to
know what these other things are. Then we shall read over the notes
about the interviews to interpret the findings, thinking about the atti-
tudes which have been expressed on the many topics, the likes and dis-
likes of respondents, their beliefs about who goes to a doctor and who
does not, and their explanations about how TONE-U works. Also, we
shall concern ourselves with the question of why they say things, why
they believe the things they do, and why they keep or do not keep medi-
cine in the home.
Then we shall attempt personality interpretations by reading their
stories over with an eye to the kind of things they imagine; what they
do or don't imagine can be done about these personal concerns; how they
feel about people; the diseases, the virtues, and the inadequacies they
perceive; the wishes, the dreads, the catastrophes they conceive; and how
they put them to work in the situation. In short, we shall find out
something of what their private worlds are like. We shall match this
with a social analysis from what we learn of their homes, jobs, ages, occu-
pations, and their parents. With that we shall know what kind of per-
son each one is. Finally, we shall put them into groups people who
have the same social worlds, people who have the same personalities,
people who buy a lot of TONE-U, people who buy none, people who
buy some, people who buy only other remedies. From these we can
build up composite pictures to answer the questions of the research.
Application:
From such a study of the personalities of users we can learn what they
need, and what they find in TONE-U. This knowledge will establish
the basic appeals which can be used by the advertiser. From the study
of social classes involved we can learn what they listen to, which will be
information to the advertiser about how to state his appeals so his audi-
ence will hear and accept his words.
CASES IN COLLECTION OF DATA 195
The Questionnaire
The following is a draft of the type of questionnaire we would pro-
pose to use. Introductory remarks are omitted.
1. a) Do you now have any of these in your home? Which ones?
(Respondents will be handed a list of brands.)
b) Have you had any of them in the past?
(Interviewers will record comments offered in addition to yes or
no for the above questions. )
c) (If the answer is yes to any of the above, the interviewer will ask
the following questions.)
Who made use of the preparations?
For what purpose were they used? For what ills? Was the aim
to prevent, to cure, to relieve?
How often did you use it? Just once? Over a period of time for
a given difficulty?
How often do you use it from time to time for different difficul-
ties?
How effective did you find the (product name)?
d) (If the respondent says yes only to aspirin, and no to the others,
the interviewer will ask the following.)
Do you keep any medicines in the house (and/or office, where
applicable)?
What kinds are they? That is, were they prescribed by your
doctor?
What kinds do you have that were not bought by prescription?
What kinds of medicine do you give to the children (if any)?
What do you do for minor illnesses? That is, for yourself or
other members of the family?
2. I would like you to think of the heaviest user of the preparations
mentioned earlier (brand names) that you know.
Why does (s)he use them? What for? How often?
What is your opinion of the way (s)he uses it? Does (s)he seem to
feel it helpful?
What kind of person is (s)he? Could you tell me something of
what (s)he is like?
3. I would like you to think of someone you know who practically never
uses any medicine.
Why doesn't this person take any medicine?
What kind of person is (s)he? What's (s)he like?
Do you think (s)he really doesn't take medicine, or do you think
196 MARKETING RESEARCH
maybe (s)he takes real pride in being healthy and doesn't like to
admit taking anything?
4. What kind of people prefer to take care of themselves when they are
sick? (If respondents find this hard to respond to spontaneously,
interviewers may suggest: children, old people, men, women, work-
ers, smart people, scared people, etc.)
Do you think of some particular person who is this way?
What is (s)he like otherwise?
5. What kind of people prefer to go to a doctor?
6. How much home medication do you think a mother ought to give
her children? What kinds of measures might she take for what
kinds of illness or complaints?
7. What's your opinion on why aspirin is used by so many people for
so many different things?
8. Why do remedies like TONE-U work on different bodily troubles?
(If the answer is scanty, the interviewer will ask something along
this line: What kind of process goes on in the body when remedies
like TONE-U are taken?)
9. What kind of illness or bodily disease is worst in terms of the hard-
ships and handicaps it leads to?
What kind do you think is most unpleasant?
10. Now there's just one more thing left. I'm going to show you a
picture and I would like you to make up a story about it. Tell -what
the people are doing, how they feel, what they are thinking about,
-what has happened to them in the past, and what you think is going
to happen to them in the future. People tell all kinds of stories
about pictures like this, so just make up any story that the picture
suggests to you. (Various pictures will be used on various persons
with differing apparent pains or troubles. )
ACTION NEEDED
The product manager was impressed by the many possibil-
ities of this study. However, he had some doubts about the
reactions which some members of the company might have to
such a study, some doubts about the adequacy of the sample,
the quality of the interviewers to be used, and the worth of
some particular questions. Should he accept the proposal?
Should he accept it subject to some changes and, if so, what
changes?
CASES IN COLLECTION OF DATA 197
CASE 8 HOW CAN THE DATA BE OBTAINED?
BACKGROUND
In 1951, a study was undertaken to evaluate the reliability
and consistency of replies of individual family members on
sample surveys. The background of this study and the prob-
lems it sought to answer are given by the following excerpts
taken from the original project proposal:
The relative influence of each of the sexes on consumer purchases is
of considerable interest and importance to marketing, sales, and adver-
tising people alike. About ten scattered studies have been made during
the past thirty years in attempts to answer this question for certain com-
modities, for certain areas, or for certain classes of people. These sur-
veys have generally taken the form of asking one member of a household
to state who influences most the purchases of particular commodities.
In some cases the respondent will be asked to state who does the actual
purchasing as well. The fact that the respondent happens to be, un-
avoidably, a family member himself (or herself), and is therefore inti-
mately concerned on such questions has long been recognized as a
serious potential bias, how serious nobody knows.
About the only attempt made to correct for this bias has been to re-
quest the opinion of college students as to which of their parents exerts
the dominant influence on the purchases of particular products. How-
ever, this procedure eliminates one unknown bias (and then only to the
extent to which the students themselves are not concerned in the reply)
at the expense of introducing another bias; namely, that due to the stu-
dent's lack of knowledge as to which of the parents wields the power of
the purse on particular items. Again the probable magnitude of this
bias has never been determined. In all the previous studies only one
member of a family has been questioned and no attempt has been made
to verify the respondent's replies by securing answers to the same ques-
tions from the other members of the family. This is obviously the only
means of ascertaining the reliability that can be placed in the replies of
the respondents.
It is therefore proposed to conduct a survey that would answer the fol-
lowing two questions:
1. What is the degree of agreement between men and women as to
the relative influence of the sexes on consumer purchases?
2. To what extent do children's opinions of the relative influence of
198 MARKETING RESEARCH
each of their parents on family purchases agree with those of their par-
ents?
At the same time, it is also proposed to test the reliability of various
other types of questions, e.g., questions on the labor force, income, in the
same manner.
THE QUESTIONNAIRE
The questionnaire designed for the study is reproduced on
the following pages. It was contemplated that all adult (six-
teen years of age and over) members of the sample families were
to be interviewed, and that one copy of the questionnaire would
be filled out for each family member.
EXHIBIT 1
Study of Sample Variability
1 . a) Has anyone in your household bought any durable goods since last Sep-
tember, that is, items similar to those listed on Card A?
Yes 1 No 2 Not sure 3
b) If yes, what was bought? (Please list items purchased below and answer
parts c through / of this question for each item.)
(a) (b) (c) (d) (e)
c) Which member or members of the family made the actual purchase?
d) Now, as we all know, the person buying an article is not always the same
as the one who is most interested in having the purchase made. In the
case of each of the above items, which member or members of the family
were most instrumental in getting the purchase to be made?
e) How was each of these purchases financed?
/) Why was each of these purchases made at that particular time?
2. a) Are any of you planning to buy any durable goods between now and next
September?
Yes 1 No 2 Notsure 3
b) If "yes" or "not sure/' what is each of you planning to buy?
(a) (b) (c) (d) (e)
Item
Member(s)
CASES IN COLLECTION OF DATA 199
c) Do you know when it will be bought?
d) Do you know where it will be bought? ( Name of store if possible )
e) How do you plan to finance it?
3. It is commonly known that the wife usually acts as purchasing agent for the
family although she may be acting after consulting with other members.
For example, the wife may buy all the linens though various towels and
handkerchiefs may be bought to satisfy the preferences of the husband or
of some of the children. Can you give us a rough idea of the relative influ-
ence, in percentage terms, that each member of your family has upon the
purchase of each of the following items? Make sure that the total of each
row is 100 per cent.
Others (please specify)
Husband Wife Son Daughter Total
Car 100%
Living-room
furniture 100%
Table model
radio 100%
Phonograph
records 100%
Refrigerator 100%
Dentifrices 100%
Breakfast
cereal 100%
Linens 100%
4. c) Do you have a family budget in the sense of either
(1) keeping breakdowns of all expenditures by the week or month?
Yes 1 No 2 Notsure 3
or
(2) planning future durables expenditures on the basis of some schedule?
Yes 1 No 2 Notsure 3
b) If you have a budget, which members of the family assist in preparing it?
5. Do you feel that your family is financially better able or worse able to buy
durable goods now than, say, a year ago?
Better 1 Worse 2 Not sure 3
200 MARKETING RESEARCH
Is this because:
a) The family income has changed over the past year? Yes No 1
Not sure 4 If yes, how? Has your income been rising 2
falling? 3
b) Your liquid savings (cash plus money in bank plus U. S. bonds) are now
different? Yes No 1 Not sure 4 If yes, how?
Have your savings risen 2 fallen 3.
c) Other reasons
6. In the coming six months do you expect your family income to be higher,
lower, or about the same as in the last six months?
Lower 1 Higher 2 About same 3 Not sure 4
7. Is there anything that you believe will be hard to get during the next six
months because of our mobilization program?
Yes 1 No 2 Not sure 3
8. Do you expect our relations with the Communist nations to get better,
worse, or not to change during the next six months?
Better 1 Worse 2 Not change 3
9. a) How many members are there in the family?
b) How many family members are under 14 years of age? .
10. a) What is your position in the household
b) W T hat is your employment status?
Housework School Looking for work
Employed Other
c) If employed, what do you do?
1 1 . Please indicate your age bracket below.
Under 20 1 20-34 2 35-49 3 50 4
12. a) In which of the brackets shown in Card B does your income fall? .
b) How about the income of the entire family?
THE PROBLEM
The collection of data for such a study poses a major prob-
lem; because of the objective of the study, the data must be
obtained from the individual family members in such a way
that each family member has no knowledge of the replies of
the other family members at the time of the interview. Obvi-
ously, if the members of a particular family heard of the nature
of the survey beforehand and had an opportunity to discuss
their opinions of the subject, the replies of that family would
be valueless. The same would be true of a questionnaire of
an individual hearing in advance of the scope of the survey.
CASES IN COLLECTION OF DATA 201
The problem is, therefore, essentially one of securing the in-
formation from the family members rtot only separately but
seemingly, and paradoxically, simultaneously as well. Tact is
obviously a major consideration in securing these data, for any
irritation of sample respondents may well lead to a refusal.
This matter is of particular importance in explaining the pur-
pose of the survey to the sample families. For what reason can
be given for wanting to ask the same questions of the different
family members?
ACTION NEEDED
The sample design has been laid out, the sample families
have been chosen, the questionnaire has been completed, and
the interviewers have been trained. The question now be-
comes one of determining how to approach the sample mem-
bers. In other words, a complete set of instructions is needed
describing:
1 . How the initial contact is to be made.
2. How the purpose of the survey is to be explained to the
respondents.
3. How the actual interviews are to be conducted.
Part VI
EDITING AND TABULATION
Although editing and tabulation are generally considered as
the beginning of the "winding up*' phase of a research study,
to be done after the data have been collected, these operations
may at times be carried out advantageously concomitantly with
the collection of the data. This is particularly true in situations
where the possible answers can be determined in advance as in
the case of dichotomous * and multiple-choice questions and
where the survey is relatively small and the staff working on it
relatively few, so that specialization of operations is not feasible.
A Review of Some Basic Considerations in Editing
For the most efficient results, editing should be the respon-
sibility of a number of individuals and not exclusively of the
one with the title "editor." Specifically, editing concerns the
following persons :
1. The interviewer who should review his questionnaires for
vague or inconsistent replies, illegible answers, and omis-
sions.
2. The field supervisor who should check that the interview-
ers have followed instructions and that answers have been
provided in the prescribed manner.
3. The office staff who should:
a) Verify the internal consistency of the questionnaires in
so far as possible.
* Questions where there are basically only two possible, mutually exclusive
answers, e.g., yes-no, male-female.
203
204 MARKETING RESEARCH
b) Rectify errors and omissions.
c) Ensure comparability between questionnaires.
d) Determine the classifications to be used in the subse-
quent tabulations and analyses of the data.
Preliminary classifications are generally set up at the time the
questionnaire is constructed, but the final determination or
check is best left till the data have been collected. It is a good
idea to check the final classifications by tabulating a small ran-
domly selected sample of the returns. Such a pretabulation
enables corrections to be made to reduce the frequency of near-
zero categories and of heavy concentration of replies in other
categories.
The Case for Preceding
Whenever the answers possible on a questionnaire can be
listed in advance, the advisability of preceding may well be con-
sidered. Placing code numbers directly on the questionnaires
can be a considerable timesaver. In the case of manual tabula-
tion, this procedure facilitates the preparation of worksheet
tables direct from the questionnaires, eliminating the inter-
mediate step of coding and transcribing the replies to posting
sheets. In the case of machine tabulation, proper arrangement
of the questionnaire combined with preceding enables punch
cards to be made directly from the questionnaires.
On the other side of the ledger, preceding is not advisable if
the answers to be classified cannot be foreseen with a high
degree of assurance, or in the case of long, probing interviews
where undesirable forcing of replies might be necessary in order
to categorize them. In addition, preceding must appear unob-
trusively on mail questionnaires, lest the respondent be con-
fused by the profusion of numbers.
Manual vs. Machine Tabulation
A summary of the main considerations entering into the
selection of a manual or machine tabulation method is provided
on the next page.
EDITING AND TABULATION 205
Manual Machine
1 . Useful in small numbers of ques- 1 . Useful* for a large number of ques-
tionnaires with relatively few ques- tionnaires, with numerous items
tions, lists of items to be checked, and cross classifications.
and relatively few cross tabulations.
2. Useful in depth interviewing in 2. Generally useful for items that can
which a formal questionnaire is not be classified into meaningful cate-
used. gories.
3. Useful in pilot tabulations for plan- 3. A combination of both manual and
ning machine tabulations. machine may be useful if detailed
analysis, such as for brand prefer-
ence, is needed.
4. Time-consuming in the posting op- 4. Time-consuming in the coding and
eration. punching stages.
Mark Sensing. This is a mechanical method somewhat sim-
ilar to punch card work, except that instead of punching the
information on the card it is marked with an electrically sensi-
tive pencil. These cards are passed through special tabulating
machines which pick up the markings electrically and tabulate
the results. This method possesses the advantage of elim-
inating the punching operation, but can be used only where
clear-cut answers are involved, e.g., yes no and multiple-choice
questions. Among other uses, it is a popular device for mark-
ing examinations composed of questions of these types.
Keysort Tabulation. This is a semimechanical method
based on the use of cards with circular holes in the margins.
These cards can be punched in such a way that when a long
needle is inserted, the punched cards remain on the needles.
Counting must be done by hand.
Findex Method. This is based on the use of cards with holes
already punched in them. The information to be recorded is
punched on the card by connecting two adjacent holes. A
special filing drawer is used to hold the cards; when a needle
or rod is thrust through the front of the file drawer, the holes
in the cards that have been slotted will be elongated so that
when the drawer is turned over, all the cards punched for a
particular hole will drop down into position. The cards are
then locked in this position, the drawer is turned over to its
original position, and the cards are then counted.
206 MARKETING RESEARCH
Keysort and Findex methods are mainly useful in studies
based on several hundred cases where the information obtained
for each case is not too detailed or comprehensive.
The Cases in This Part
The five cases in this Part provide practice materials in all
three of the aspects of the research operation reviewed on the
preceding pagesediting, coding, and tabulation. Here the
emphasis is, if anything, more on detail work than on general
policy considerations, on the actual devising of codes, methods
of tabulation, and tabular layouts, because it is only after doing
a considerable amount of work of this type that the beginning
researcher becomes aware of the various problems arising at this
stage of the research operation. The first two cases deal with
editing and coding. The first case deals with the editing of
responses and devising a coding system for the replies to one
part of a frozen food questionnaire. The second case involves
the preceding of a questionnaire for a bacon survey containing
questions representing a fairly broad range of coding problems.
Tabulation is the subject of the next three cases. Case 3 in-
volves the preparation of forms for manual tabulation of a
jewelry survey where time is the main consideration.
Case 4 poses the frequently encountered problem of choosing
a method of tabulation, the subject here being a rather lengthy
insurance questionnaire. The question of editing the question-
naire is another part of this case.
The last case carries the research operation a bit closer to
completion, involving the specification of cross-classification
tabulations and the preparation of tentative summary tables
for a department store survey. The student is asked not only
to list the tables but to lay out each in detail.
BIBLIOGRAPHY
AMERICAN MARKETING ASSOCIATION. The Technique of Marketing Research.
New York: McGraw-Hill Book Co., Inc., 1937. Chapter 14 devotes the
entire discussion to machine tabulation methods.
BENJAMIN, K. "Problems of Multiple-Punching with Hollerith Machines."
Journal of the American Statistical Association, XLII (March, 1947), pp. 46-
EDITING AND TABULATION 207
71. A technical knowledge of punch card operations will be useful in under-
standing the problems described and their solutions.
BOYAJY, J. S., et d. "Tabulation Planning and * Tabulation Techniques/'
Journal of Marketing, XIII (January, 1949), pp. 330-55. A summary of the
steps that must be considered in planning tabulations; some questionnaire
problems are also discussed.
BRADFORD, E. S. Marketing Research. New York: McGraw-Hill Book Co.,
Inc., 1951, pp. 188-96. A very brief statement on editing and tabulation
procedures.
BROWN, L. O. Marketing and Distribution Research. New York: The
Ronald Press Co., 1949. Chapter 23 contains a summary of the most
frequently used methods of editing and coding.
EDWARDS, T. I. "The Coding and Tabulations of Medical and Research Data
for Statistical Analysis." Public Health Reports, LVII (1942), pp. 7-20.
An example of the way in which written information as well as punched data
can be incorporated on the same punch card.
GREENBERG, A. "A Method for Coding Questionnaires in Market Surveys/'
Journal of Marketing, XIV (October, 1949), pp. 456-58. The methods by
which multiple choices can be coded are explained and illustrated.
HALL, R. O. Handbook of Tabular Presentation. New York: The Ronald
Press Co., 1946. This is a standard reference work on the rules for prepara-
tion of all forms of tables.
HEIDINGSFIELD, M. S., and A. B. BLANKENSHIP. Market and Marketing Analy-
sis. New York: Henry Holt & Co., 1947. Editing and tabulation methods
are summarized in Chapter 10. Some helpful hints are given for the prepara-
tion of tables which will be useful to beginning students of marketing re-
search.
LORIE, [ . H., and H. V. ROBERTS. Basic Methods of Marketing Research. New
York: McGraw-Hill Book Co., Inc., 1951. Chapter 22 discusses problems
and methods of preceding, editing, final coding, and tabulation.
LUCK, D. J., and H. G. WALES. Marketing Research. New York: Prentice-
Hall, Inc., 1952. Chapter 12 contains a detailed treatment of editing and
tabulation methods.
PARTEN, M. B. Surveys, Polls, and Samples. New York: Harper & Bros., 1950.
Chapters 13 and 14 contain a treatment of editing and coding.
Part VI Cases
CASE I EDITING RESPONSES
BACKGROUND
The rapid expansion of the frozen food industry in the 'for-
ties had caused severe problems of quality control. Many pur-
chasers of lesser known brands had been dissatisfied with the
poor quality of products received. Processors with well-known
brands believed that this problem had been greatly alleviated
both by improved techniques and experience and by the with-
drawal from the field of many small companies. Nevertheless,
the problem seemed sufficiently important to be included
among other topics in a personal interview survey on frozen
foods being planned. It provided a possible explanation for
the limited use of frozen foods, the brand preferences of con-
sumers, and consumers' attitudes on price. The informal in-
vestigation, moreover, had shown that many women still
remembered these past disappointments.
For the questionnaire constructed for testing in the field, no
satisfactory classification of complaints by consumers had been
found. For testing purposes, therefore, it had been decided to
ask an open-end, exploratory question. The question used was
worded as follows:
(A vegetables
6. Have any frozen (B fruits ever spoiled?
ANS. A. VEG. YES NO NO ANSWER
B. FR. YES NO NO ANSWER
IF YES, ASK: What was the cause of spoilage?
This question followed one on methods of storage used in the
home. Interviewers were expected to ask about either vege-
208
CASES IN EDITING AND TABULATION 209
tables or fruits or both, depending on which the respondent
already had indicated that she purchased. *
RESPONSES SECURED
A. For those indicating vegetable spoilage :
1. One package of corn was "soggy." "I'm not exactly sure but I
think it must have been poor quality because I thawed it just like
other vegetables."
2. Poor handling by the maid.
3. Once spoiled after three days when refrigerator broke. Won't
spoil "when kept properly."
4. Got asparagus twice that was "gritty." "It was so bad we
wouldn't eat it."
5. Has lost several packages at different times. Not sure whether
they thawed before she put them in the refrigerator.
6. Tried broccoli once. Forgets which brand. Was "rotten."
Her grocer gave her her money back.
7. Just last week had some mixed vegetables that didn't "taste
right." Her family refused to eat them.
8. "You have to be awful careful not to let them thaw and then
freeze again." Uses hers right after buying.
9. Doesn't know why but just doesn't like the taste of (brand)
green beans. Won't buy again.
10. Cube compartment was full and a package thawed out.
11. Says threw lots away when had to buy "strange kinds." Buys
only (brand) now and has no trouble.
12. Didn't like the looks of wax beans once. "I said, Into the gar-
bage with them.' "
1 3. Not really spoiled but didn't seem "worth the price."
B. For those indicating fruit spoilage :
1. Thinks they were spoiled before she bought them. Probably
"low grade" fruit.
2. Peaches often get discolored when they thaw out.
3. Reluctant to say directly that anything had "spoiled," but said
she had to throw out frozen strawberries because they "did not
look edible" after thawing. "They were shriveled up and wa-
tery."
4. One package of grapefruit segments turned watery. This re-
sulted in no more purchases of that fruit.
210 MARKETING RESEARCH
5. One brand, she forgets which, of strawberries "looked simply
terrible." Discarded them.
6. Some peaches looked fine but "tasted like they had acid in
them."
C. For those indicating both vegetable and fruit spoilage:
1. Lima beans were spoiled in store and not noticed 'til home;
peaches too salty and made them taste very poor.
2. Thawed out before they were put in freezer cabinet as they were
purchased downtown and lady didn't get home as soon as she
expected to.
3. Complains that frozen foods not always clean. Can see "little
bits of all kinds of things" in them.
4. Says that peaches, especially, sometimes taste sour. A package
of peas she left in too long was no good.
5. Asks if I don't think frozen spinach "tastes funny" sometimes.
Had some "funny colored" peaches she refused to serve her
family.
6. Wonders what "they put in" frozen foods. Has to cook them so
long to taste right. Grapefruit once had some "green" in it and
looked "poisonous."
7. Corn was "all water" once. Likes (brand) strawberries but
wouldn't buy other kinds "any more."
8. Learned you can't keep any frozen foods too long unless you get
them "right by the coils" after coming from store.
9. Only lost one "thawed out" package of peas once. Thinks
most frozen fruits have a peculiar taste.
10. Fruits tend to "come apart." Had lima beans that were "full
of spots."
These were the only "complaints" registered by the 143 re-
spondents. Approximately 35 per cent reported that they did
not use frozen foods. Many more families purchased vege-
tables than fruits. In nearly half of all families frozen foods
were served the same day that they were purchased. Only a
small percentage kept them on hand more than two days.
Among comments of those not reporting spoilage were these:
"Always keep food frozen and in cube compartment"; "It you
handle them properly they won't ever spoil"; "I hestitate to keep
fruits any longer than a day because they start to run if I do";
"I can keep them for a week without spoiling." Interviews
CASES IN EDITING AND TABULATION 211
were conducted among widely scattered neighborhoods, and
no formal sample design had been followed.
ACTION NEEDED
1. What sort of system of headings or subheadings might be
devised as a basis for classifying these answers? How should
each answer have been classified?
2. Based on the evidence obtained, should the question on
spoilage be eliminated? Should it be revised, and, if so,
how?
CASE 2 PREPARING AND PRECODING
A QUESTIONNAIRE
THE SURVEY
A survey on the consumption habits of consumers with regard
to packaged bacon was planned for a medium-sized city in 1952.
After due deliberation among the staff members of the research
organization planning the study, the following list of questions
to be included in the questionnaire was prepared:
EXHIBIT 1
1 . a) Do you use bacon? Yes No Not Sure
b) If yes, is it packaged bacon or slab bacon or both?
Packaged Slab Both N.S
2. If you were buying a package of bacon right now, would you look for any
particular brand? Yes No N.S
If so, which one (s)?
Armour Oscar Mayer Wilson
Cudahy Swift Other .
3. What (other) brands can you name? None N.S
Armour Oscar Mayer Wilson
Cudahy Swift Other.
4. Which of these brands have you heard of? (Show card)
Armour Swift Reese _
Cudahy Wilson None .
Oscar Mayer Smith
5. How often do you use packaged bacon?
Every day Less than once a week
Every other day Not sure
Twice a week
Once a week
212 MARKETING RESEARCH
6. At what meals do you use packaged bacon?
Breakfast Noon Evening N.S. .
Do you use packaged bacon by itself, with another type of meat, as season-
ing for vegetables, or in more than one of these ways?
By itself With meat With veg N.S
7. Do you usually buy packaged bacon from a meat market or combination
store?
Meat market Combination store Both Other
N.S
8. Do you prefer any specific brand? Yes No N.S
If so, which one?
Armour Oscar Mayer Wilson
Cudahy Swift Other
9. Why do you prefer this brand? .
10. (1) If the store that you go to is out of stock of bacon, what do
you do: (a) ask for another brand (b) accept a substitute brand only
after looking in one or two other stores for bacon, or (c) wait
until bacon is in stock again? (a) (b) (c)
N.S
(2) (If a OT b) If so, which brands?
Armour Oscar Mayer Wilson
Cudahy Swift Other
N.S
1 1 . What qualities do you like your bacon to have?
12. How often do you buy fresh meat?
Every day Other
Every other day Don't buy
Twice a week Not sure
Once a week
13. Family size
14. Number of children
1 5. Occupation of head of family:
Professional Unskilled .
Skilled White Collar .
16. Age of housewife: Under 20 20-34 35-49 50+-
17. Income level:
Under $1,499 $4,200-6,599 .
$1,500-2,599 $6,600-10,399_
$2,600-4,199 $10,400 and over.
18. Address
19. Interviewer
20. Date
The questions in this list are not necessarily in the most
logical order, as they were noted primarily in the order in which
they were suggested.
The nature of the information desired was such that it was
estimated that approximately six hundred completed inter-
CASES IN EDITING AND TABULATION 21?
views would be required. Because of this large number of in-
terviews, a mechanical means of tabulation was to be used, the
data being transcribed to punch cards and then tabulated by
machine. To save time and expense, the questionnaire was to
be precoded. For analytical purposes, it was considered highly
desirable if the answer to such questions as 2, 3, 4, and 8 could
be designated by a single code number, so that a glance at the
code number for any particular question would indicate which
brands had been named. In this connection, it should be noted
that slab bacon is not usually sold under any particular brand
name.
ACTION NEEDED
1. Prepare a working questionnaire from the questions in the
above list, putting them into some logical order.
2. Precede the questionnaire, showing all codes directly on the
questionnaire form, making sure that allowance has been
made for all possible answers to each question.
CASE 3 PREPARATION OF FORMS FOR
MANUAL TABULATION
BACKGROUND
In 1948, a study of the operating results of retail jewelry stores
was conducted by the Business Research Bureau of a leading
university. The purpose of this investigation was to provide
a standard by which a businessman could compare his operating
experience and methods with those of similar enterprises in his
field. Because data of this type had not been presented for
jewelry stores within the past twenty years, it was felt that a
retail jeweler would be keenly interested in a study which would
allow him to compare his own operating results with those of
other jewelry stores of comparable size. All information sup-
plied was to be entirely confidential.
The plan was instituted in 1948 with the expectation that it
would be continued each year. Questionnaires (one of which
is reproduced on page 214) were to be mailed to cooperating
214 MARKETING RESEARCH
members of a State Retail Jewelers' Association about the 1 5th
of January each year with a request that the statistics be in the
hands of the research bureau by February 10 of the same year.
The reason for the early date of returning the questionnaire
was to make it possible to give a preliminary report on the re-
sults to the annual convention of the Jewelers' Association
about February 20. The detailed report to the association
members was to be issued in June of each year.
EXHIBIT 1
CONFIDENTIAL CODE NO
Survey of Retail Jewelers
For detailed instructions and meaning of terms see attached paper. If you
have more than one store, please enter figures for each on a separate page.
Enter figures (unless otherwise indicated) for full year. If you operate on a
fiscal year, rather than a calendar year ending December 31, please give the clos-
ing date of your fiscal year here: .
OPERATING RESULTS, 1947 and 1948: 1947 1948
1. Sales (including sales taxes) $ $
2. State O. E. Tax
3. Federal Excise Taxes
4. Total Sales Taxes (items 2 plus 3) . . $ $
5. Sales, excluding sales taxes (item 1
minus 4 )
6. Net cost of merchandise sold
7. Gross margin (item 5 minus 6) $ $_
8. Total expenses . . ... _
9. Income or loss, before income taxes
(7 minus 8) $ $_
EXPENSE BREAKDOWN: 1947 1948
10. Salaries of owners or managers (see note
below) $ $
1 1 . Wages and salaries, other
12. Advertising
13. Rent (if building owner, write "owned"
AND fill in item 14.)
14. Building maintenance, depreciation, in-
surance
15. Heat, light, water and power
16. Credit (bad debt) losses
17. Other expenses
18. Total expenses (item 8 in "Operating
Results") $ $
CASES IN EDITING AND TABULATION 215
(Note on salaries: How many owners and/or managers receive salaries?)
If any owners work in store, but draw no salary, how many?
INVENTORIES:
As of December 31, 1946 (or end of nearest fiscal year) $_
As of December 31, 1947 " " " " " " $_
As of December 31, 1948 " " " " " " $_
-/o
Of
REPAIR AND OPTICAL WORK:
What per cent of your total sales is repair work (including engraving)?.
What per cent of your total sales (item 1 ) is optical work?
Are these amounts estimates? (yes or no)
ADVERTISING:
Give percentage of your advertising expense for these media in 1948:
Is your breakdown of advertising expense estimated? (yes or no)
Newspapers %
Radio %
Direct mail %
* Other advertising %
Total advertising expense 1 00 %
CREDIT SALES: Give the per cent of your total 1948 sales made in the follow-
ing ways. Is this breakdown of sales an estimate? (yes or no)
Cash %
Charge accounts %
Installment plan %
Total Sales 100%
MERCHANDISE LINES: Please check merchandise lines listed below which
you earned in 1948, and if possible, give the per cent of your merchandise
sales which is in these lines. Add other lines you carry which are not listed.
Check here
if carried:
Precious stones and rings %
Other jewelry (including costume jewelry) %
Clocks and watches %
Silverware %
China and glassware %
Other goods (describe) %
* What other advertising media did you use? .
216 MARKETING RESEARCH
THE PROBLEM
The tabular information needed for the project included:
average operating statement for jewelry stores
number of stores reporting by size of city and volume of sales
average operating statement by volume of sales
average operating statement for the 25 per cent most profitable
stores reporting
merchandise turnover by size of store and size of city
inventory and cost of goods sold by sales volume and size of city
sales percentages for optical work
advertising media used by size of store and size of city
credit sales by size of store and size of city
percentage of sales that are cash, charge account, and installment
rent expense by size of store and size of city
owners' and managers' compensation as a percentage of sales
number of employees by sales volume
store hours by size of city
Manual tabulation was to be used. However, because of the
time element involved, it was necessary to develop forms for
tabulation that would permit accurate final results to be ob-
tained as quickly as possible.
DATA AVAILABLE
H. G. WALES and R. FERBER. Marketing Research Selected Literature.
Dubuque: Wm. C. Brown Co., 1952, Part V.
K. PHELPS. "A Flexible Method of Hand Tabulation/' Journal of Marketing,
III (January, 1939), pp. 265-68.
ACTION NEEDED
1. What would seem to be the best form for expediting the
tabulations? Would it be desirable to set up forms per-
mitting the data to be tabulated as the returns come in?
If yes, how?
2. Prepare directions for tabulating the data on these forms.
3. Devise a system of easily made checks to guarantee accuracy
in the tabulations, including some means of ready reference
back to the original returns.
CASES IN EDITING AND TABULATION 217
CASE 4 SELECTING A METHOD OF
TABULATION
BACKGROUND
In 1949, the representatives of several life insurance com-
panies indicated an interest in having a study made on the ex-
tent to which life insurance is used to cover the financial losses
that result from the death of key personnel in a business. For
example, when a company executive dies, expense is incurred
in replacing the man who has undoubtedly developed special
skills in handling his work. These costs arise from promoting
and training another man to take the place of the one who has
died. Hence, life insurance has been used by some businesses
to compensate them for losses arising from the death of key
executives.
THE PROBLEM
A research organization was asked to investigate the extent
to which business life insurance is used in practice. A plan was
outlined for the work, questionnaires were prepared, interviews
completed and the data were ready for editing and tabulation.
The problem then arose of selecting a method of tabulation,
a problem to which relatively little thought had been given in
the earlier stages of the project.
THE QUESTIONNAIRE
A rather lengthy questionnaire was used in the study. It is
reproduced in full below.
EXHIBIT 1
A CONFIDENTIAL QUESTIONNAIRE TO DETERMINE THE EXTENT OF THE
USE OF LIFE INSURANCE BY BUSINESS IN SELECTED AREAS
A. Business data
1. Operating name of business
2. Location of principal office
3. Number of employees
218 MARKETING RESEARCH
4. Type of business (check one) ( ) Manufacturing ( ) Wholesaling
( ) Retailing ( ) Service
5. Form of business unit (check one) ( ) Sole proprietorship
( ) Partnership ( ) Corporation ( ) Closed corporation
6. Description of business
7. Volume of business (Gross sales)
B, For sole proprietorships only (Business interest liquidation)
1. Have arrangements been made for the continuation of the business after
the owner's death? ( ) Yes ( ) No
2. (If the answer to question 1 is Yes ) What arrangements have been made?
( ) Owner's wife will operate the business
( ) Owner's son will operate the business
( ) Other relative of the owner will operate the business
( ) Owner has a definite agreement with a particular person or group of
persons to purchase the business upon death of the owner at a fair
and equitable price
( ) Other arrangements
3. If there is a definite agreement with a particular person or group of per-
sons to buy the business at owner's death, is it reinforced by the use of
life insurance on the owner's life paid for by this person or group of per-
sons to finance the purchase of the deceased interest?
( )Yes ( )No
a) Who drew up the agreement? ( ) Attorney ( ) Life Insurance
Agent
b) How much life insurance is in force for this purpose?
c) What plan is used?
d) How was the amount arrived at?
( ) Value of the business agreed upon in the buy and sell agreement
( ) Estimated value of business
( ) A percentage of the value of the business to be used as down
payment
( ) Arbitrary amount
( ) Other
e) Do you feel that the plan now used adequately meets the needs?
( ) Yes ( ) No ( ) Uncertain
/) How did you first become interested in the plan you now use?
( ) Approached by Life Insurance Agent
( ) Approached by Life Insurance Agent and referred to Home Of-
fice specialist
( ) Formulated the plan without motivation outside of the business
( ) Received suggestion from accountant
( ) Received suggestion from attorney
( ) Received suggestion from general insurance salesman
( ) Other
4. If there is no plan financed by life insurance involving the sale of the
business in case of the death of the owner, please complete the following:
CASES IN EDITING AND TABULATION 219
tf) Have you ever had explained the use of life insurance plans to liquidate
the business interest in case of the owner's death? ( ) Yes ( ) No
b) (If the above answer is Yes) What are the reasons for not entering
into such a plan?
c) (If the answer to part (a) is No) Do you think there is a place for life
insurance in your business? ( ) Yes ( ) No ( ) Undecided
C. For partnerships only (Business interest liquidation)
1. Have arrangements been made for the continuation of the business after
the death of one of the partners? ( ) Yes ( ) No
2. (If the answer is Yes) What arrangements have been made?
( ) Take the heirs in as partners
( ) Take the buyers of the heirs' interest as partners
( ) Sell out to the heirs and have them continue the business
( ) Purchase the interest of the heirs
( ) Surviving partners will inherit the business
3. If the plan is to have the surviving partner purchase the interest of the
heirs in case of death of one of the partners, is there in force a "buy and
sell agreement" to be financed by insurance on the lives of each partner?
( )Yes ( )No
a) Who drew up the agreement: ( ) Attorney ( ) Life Insurance
Agent
b) How much life insurance is in force for this purpose?
c) What plan is used? Type of insurance?
Who pays premium? Beneficiary?
d) How was the amount arrived at?
( ) Value of business interest as agreed upon in the buy and sell
agreement
( ) Estimated value of business interest
( ) A percentage of the value of the partnership interest to be used
as down payment
( ) Arbitrary
( ) Other
e) Do you feel that the plan now used adequately meets the need?
( )Yes ( )No ( ) Undecided
/) How did you first become interested in the plan?
( ) Approached by Life Insurance Agent
( ) Approached by Life Insurance Agent and referred to Home Of-
fice specialist
( ) Formulated the plan without motivation outside of the business
( ) Received suggestion from accountant
( ) Received suggestion from attorney
( ) Received suggestion from general insurance salesman
( ) Other
4. If there is no plan financed by life insurance involving the sale of the part-
nership interest in case of the death of one of the partners, please complete
the following:
220 MARKETING RESEARCH
a) Have you ever had explained the use of a life insurance plan to liqui-
date the interest of a partner in case of death? ( ) Yes ( ) No
b) (If the answer is Yes) What are the reasons for not entering into such
a plan?
c) (If the answer to part (a) is No) Do you think there is a place for life
insurance in your business? ( ) Yes ( ) No ( ) Undecided
D. For closed corporations only (Business interest liquidation)
1. Have arrangements been made for the continuation of the business after
the death of one of the stockholders? ( ) Yes ( ) No
a) Who drew up the agreement? ( ) Attorney ( ) Insurance Agent
2. (If the answer is Yes) What arrangements have been made?
( ) Surviving active stockholders will inherit the deceased's interest
( ) Accept heirs of the deceased as active stockholders
( ) Allow the heirs to draw dividends but to take no active part in the
management
( ) Allow the heirs to sell their interest to outsiders
( ) An agreement whereby the heirs sell their interest to the surviving
stockholders
( ) Other
3. If the plan is to have the surviving stockholders purchase the interest of
the heirs in case of death of one of the stockholders, is there in force a
"buy and sell agreement" to be financed by insurance on the lives of each
stockholder? ( ) Yes ( ) No
a) How much life insurance is in force for this purpose?
b) What plan is used? Type of insurance?
Who pays premiums? Beneficiary?
c) How was the amount arrived at?
( ) Value of business interest as agreed upon in the buy and sell agree-
ment
( ) Estimated value of business interest
( ) A percentage of the value of the corporation interest to be used
as down payment
) Arbitrary
) Other
d) Do you feel that the plan now used adequately meets the needs?
( )Ycs ( )No ( ) Undecided
e) How did you first become interested in the plan?
( ) Approached by Life Insurance Agent
( ) Approached by Life Insurance Agent and referred to Home Of-
fice specialist
( ) Formulated the plan without motivation outside of the business
( ) Received suggestion from accountant
( ) Received suggestion from attorney
( ) Received suggestion from general insurance salesman
( ) Other
4. If there is no plan financed by life insurance involving the sale of the
CASES IN EDITING AND TABULATION 221
stockholder interest in case of the death of one of the stockholders, please
complete the following:
a) Have you ever had explained the use of a life insurance plan to liqui-
date the interest of a stockholder in case of death? ( ) Yes ( ) No
b) (If the answer is Yes) What are the reasons for not entering into such
a plan?
c) (If the answer to part (a) is No) Do you think there is a place for life
insurance in your business? ( ) Yes ( ) No ( ) Undecided
E. For all business units
1 . Does the business carry insurance on the lives of the officers of the com-
pany? ( )Yes ( )No
a) Which ones: ( ) President ( ) Vice President ( ) Secretary
( ) Treasurer ( ) General Manager ( ) Other
2. Does the business carry insurance on the lives of any of the employees who
would be difficult to replace except at a financial loss? ( ) Yes ( ) No
a) Type of employee
3. (If the answer to questions 1 or 2 is Yes)
a) What were the principal motives leading to the purchase of this in-
surance?
( ) Tax advantages
( ) Protection needed by the business
( ) Retirement plan for officers
( ) Other
b) How many policies do you have and what are their amounts? Plan?
Policies Amounts Plan
c) How did you arrive at these amounts?
( ) A percentage of salary paid
( ) An estimate of the real contribution to the profits of the business
by the insured employee
( ) An arbitrary amount
( ) Other
d) Do you feel that your plan adequately meets your needs? ( ) Yes
( )No. ( ) Doubt
e) How did you first become interested in the plan which you now have
in force?
( ) Approached by Life Insurance Agent
( ) Approached by Life Insurance Agent and referred to Home Of-
fice specialist
( ) Formulated the plan without motivation outside of the business
) Received suggestion from accountant
) Received suggestion from attorney
) Received suggestion from general insurance salesman
) Other
4. (If the answer to questions 1 and 2 are both No)
a) Have you ever been offered an explanation of the use of life insurance
to protect business values? ( )Yes ( ) No
222 MARKETING RESEARCH
b) (If the answer to the above question is Yes) What is your reason for
not covering the lives of the key men?
c) (If the answer to question (a) is No) Do you think there is a place for
life insurance in your business? ( ) Yes ( ) No ( ) Undecided
F. Comments
OTHER CONSIDERATIONS
About five hundred and fifty interviews were obtained with
this questionnaire. The answers to some of the questions were
almost entirely of a free response type. Most of the responses,
however, were definite enough to permit classification by dis-
tinct groups.
Only two persons were available for three months' service to
do the tabulating work. Both were relatively inexperienced at
manual tabulation work but in spite of this, the director of the
research project wanted to keep these people busy, so he sug-
gested that manual tabulation be used. It would cost approx-
imately $425.00 per month to keep these two workers on the
payroll for parttime work on this and other projects.
A considerable number of cross-classifications would have
to be derived from the data obtained in answer to each question.
The task of tabulation also would be made more arduous by
the substantial number of subclassifications that would be
necessary in the initial tabulation work.
The schedule called for the tabulation work to begin in Sep-
tember and be completed by December 1. At each stage of
the work, careful analysis of the tabular information was to be
made in order to ascertain if significant results were being
brought to light, and dummy tables were prepared to guide the
persons who were immediately responsible for the tabulation
work.
ACTION NEEDED
1. A selection of some method of tabulating the data gathered
with this questionnaire had to be made. What should that
method bemechanical, manual, or semimechanical (Key-
sort, Findex, etc.)? Why?
CASES IN EDITING AND TABULATION 223
2. What is the nature of the editing work needed to prepare
this questionnaire for tabulation by the chosen method?
3. Prepare a set of instructions for tabulating the data from
this questionnaire.
CASE 5 CROSS-CLASSIFICATION
TABULATIONS
BACKGROUND
The Langdale Department Store had completed gathering
data on a customer survey directed at discovering why so many
credit accounts of their customers were small. (See Part III,
Case 5, for details on hypotheses and types of data gathered.)
Personal interviews made with customers living within the
city had produced slightly under five hundred usable returns.
A mail questionnaire to customers outside the city had pro-
duced slightly over three hundred usable returns. Those re-
turns had been edited, coded, and the data entered by the store
through manual tabulation on so-called tally or working sheets
(illustrated on pp. 226-28). In addition, data from store
records, principally dollar expenditures, made against each
credit account during the past year, had been recorded.
THE PROBLEM
What summary tables should be drawn from the data com-
piled on the tally sheets? Numerous factors were involved in
resolving this question.
The problem, of course, had been anticipated. During the
planning stage, attempts had been made to visualize the sort
of tables that would be needed to test the formulated hypoth-
eses. Descriptions of about thirty tables had been prepared
before the final questionnaire form was drawn. For a number
of the more important tables, statistical findings had been anti-
cipated and tentative figures entered in the tables. This pro-
cedure provided a test of the size of minor breakdowns to be
expected and of the usefulness of the data-gathering forms for
producing the needed data.
224 MARKETING RESEARCH
Plans had called for a sample of a thousand cases. Because
returns were only eight hundred, some loss in minor universes
size and in cross-classification opportunities resulted.
Consideration of the kind of summary tables desired also had
entered into the design of the tally sheets. The decision had
teen made that the characteristics appearing most frequently
in the cross-classifications should be customer income and size
of customer accounts. A procedure had been adopted, there-
fore, in advance of tabulation to simplify the construction of
summary tables, including such cross-classifications. This was
to be accomplished by sorting questionnaires into groups of
Upper, Middle, and Lower incomes and into subgroups of
Large, Medium, and Small accounts, and then by recording
the responses of each subgroup on a separate tally sheet. (How
this simplified the work of making summary tables will be
clearer after inspection of the illustrations given later.)
The classification of data as recorded on questionnaires or
on the tally sheets also was a factor. The classifications did not
determine what summary tables could be made, but they did
restrict the breakdowns of data which could be made without
direct reference back to the original questionnaires. No matter
what preplanning of tables had been done or the means to be
employed for drawing off summary tables, later analysis could,
of course, be expected to reveal the desirability of tables not pre-
viously anticipated.
Another factor influencing decisions was the desirability of
simple, one-dimensional tables, or straight runs, as against two-
dimensional (cross-classification) tables, which are almost es-
sential for correlation analysis, and as against more complex
tables, those concerned with more than two dimensions.
A last factor involved in determining what summary tables
to construct was the relative preferences for "storehouse" tables
{convenient summaries of many data but not necessarily well
adapted to analysis), for working tables (useful relationships),
and for presentation tables (the sort that might appear in a final
report presenting the results of the study).
CASES IN EDITING AND TABULATION 225
DATA AVAILABLE
The basic plan of the tally sheet was to reproduce the ques-
tionnaire vertically down the left side of a large sheet of ruled
accounting paper, making provision under each question for
all classifications of data desired. Each questionnaire, after
editing and coding, was then entered on this tally sheet on a
separate vertical column. This method is illustrated on pp.
226-28, using for this purpose and this entire case the ques-
tionnaire as proposed after the pretest described in Part III,
Case 5, pp. 91-100. The questionnaire used is shown on pp.
98-100.
The tally sheet illustrated on pp. 226-28 was a part of the
master counter sheet covering all the respondents (60) whose
family income was in the "Upper" group and whose accounts
at the Langdale Store the previous year had been "Large."
Similar sheets assembled such other groups as Upper Income,
Medium Accounts, and Lower Income, Small Accounts in a
similar fashion. Each would have the same vertical columns
on the left. Each questionnaire had a code, or sample, number,
as shown by the column headings on pp. 226-28, which was
keyed to a list containing respondent names and addresses,
thereby permitting checks to be made on the tabulations and
reference back to the original data, should this be necessary.
A question such as No. 4, it will be noted, provides no tab-
ulating problem. The answers Yes, No, and D.K. (D.K. stand-
ing for Don't Know or Unknown) cover all possible answers
to the question asked. In contrast, a question such as No. 1
can be troublesome. Answers might be classified for each year
separately, although for most purposes the detail would be
meaningless as well as extremely complex and the number of
cases in each class would be ridiculously small. On the other
hand, all answers might be put into only three categories, such
as 5 years or less, more than 5 years, and inevitably, D.K.
If the researcher employed broad breakdowns, as illustrated
above, and in analyzing the data discovered a high correlation
between size of account and length of residence, he probably
would want further breakdowns, assuming the group was large
I
I
I
O
r-H
OO
s
s
^c
s
Ol
Oi
QQc^Q
226
$
a
C/D
a
1
"c
t
i
a
It
s
227
I
I
EC
oo
I
8,
s
*
1 ill
ex. ex, cu cu o* o*
rf Q - *-(
OOOOOOQ
228
CASES IN EDITING AND TABULATION
229
enough to permit such an operation. To obtain these break-
downs, however, would require a tabulation separate from the
original questionnaires. Should the real answer lie in the realm
of fractions of a year, then it would be too late; the question-
naire makes no provision for such breakdowns. (It is the work
of the preliminary investigations, planning, and pretesting to
avoid such errors.) In general, it is best, in tabulating, to pro-
vide for more numerous breakdowns than seem necessary at
first; for it is easy to merge classifications, but costly to split
them after original tabulations are completed.
From the classifications shown above, the types of summary
tables shown on pp. 229-30 might be constructed.
Summary Table A is of the one-dimension, or straight-run,,
type, Its limited usefulness for analysis is apparent; and it will
be noted that such one-dimensional data constantly reappear
in the "Total" columns and rows of all summary tables. For
these reasons, such summary tables often are omitted in analyt-
ical work, although the facts may be important in presenting
results and these tables may appear in the final published report.
Summary Table B is the correlation, or two-dimen$ion y type
EXHIBIT 2. ILLUSTRATIONS OF SUMMARY TABLES
SUMMARY TABLE A SIZE OF CUSTOMER ACCOUNTS
Size
Large
Medium
Small
Total
Number of Accounts
200
264
336
800
SUMMARY TABLE B SIZE OF CUSTOMER ACCOUNTS, BY INCOME GROUPS
Size of
accounts
Income Groups
Upper
Middle
Lower
Don't Know
Total
Large
Medium
Small
Total
84
95
111
89
143
143
23
16
46
4
10
36
200
264
336
290
375
85
SO
800
230 MARKETING RESEARCH
SUMMARY TABLE C SIZE OF CUSTOMER ACCOUNTS, BY INCOME
GROUPS AND AGE GROUPS
Income and
Age Groups
Size of Account
Large
Medium
Small
Total
Upper Income
44 or younger
4 5 or older
Age unknown
__
...
All
84
95
111
290
Middle Income
44 or younger
45 or older
Age unknown
All
89
143
143
375
Lower Income
44 or younger
45 or older
Age unknown
All
23
16
46
85
Income Unknown
44 or younger
45 or older
Age unknown
All
4
10
36
50
All Incomes
44 or younger
4 5 or older
Age unknown
Grand Total
200
264
336
800
of table, being extremely useful in revealing relationships and
the most common of all types of analytical tables. The table
shown could have been run, and usually would be run, to show
seven income breakdowns. To simplify the illustration,
Groups A and B were merged into Upper Income, Groups C
and D into Middle Income, and Groups E and F into Lower
Income.
Summary Table C exemplifies the more complex, three-di-
mensional table. The complexity of multidimensional tables
would be even clearer if all seven possible income breakdowns
plus the six possible age groups had been used in the illustration.
Even in the more simple form shown, it will be noted that the
CASES IN EDITING AND TABULATION 231
analyst would have the chore of splitting up sixteen totals into
minor categories. For much larger samples, zeroes or single-
digit numbers tend to appear for some breakdowns in such
tables.
The complexity of three-dimensional tables, however, is
mostly a matter of complexity of analysis. A moment's reflec-
tion will reveal that they are not much more complex or time-
consuming to construct than are many two-dimension tables.
They also have a marked advantage as a storehouse for a great
deal of data (note how Table B could be constructed quickly
from the total columns of Table C) and as clues to the inter-
relationship of market factors in a world that after all is not two-
sided.
ACTION NEEDED
1. How many summary tables should be constructed from the
data gathered with this questionnaire and of what types?
2. Design all column headings and line designations for the
most important tables, including at least two complex-type
tables.
Part VII
ANALYSIS AND INTERPRETATION OF DATA
The questions that generally arise in the course of analyzing
a set of data may be grouped under the following headings:
The reliability of the data
Formulating and testing working hypotheses
Interpretation of results
Verification of findings
Translation of findings into recommendations for action
Some pertinent remarks on each of these questions are of-
fered on the following pages.
RELIABILITY OF DATA
An evaluation of the reliability of a set of data not only aids
the reader or client in arriving at a more complete judgment of
the validity of the findings, but also serves as a determining fac-
tor in the selection of analytical techniques, and may even alter
the findings themselves. Thus, if the data appear to be subject
to wide margins of error, it is generally foolish to apply highly
refined and time-consuming analytical techniques to the data
or to dwell on the apparent implication of relatively minor dif-
ferences. All one can do in such a case is to concentrate on
over-all trends revealed by the data and to pass over the fine
points.
A set of findings may be altered by an evaluation of data
reliability if the data are found to possess some bias that can
be corrected. In such a case the data should be adjusted before
the substantive analysis is begun, 1 though such corrections
1 For an example of how this was done, see R. Ferber and H. G. Wales,
"Detection and Correction of Interviewer Bias," Public Opinion Quarterly, XVI
(Spring, 1952), pp. 107-27.
233
234 MARKETING RESEARCH
should be made with extreme caution because they may at
times alter substantially the nature of the final results.
Now, what does an evaluation of data reliability include? In
the case of data taken from other sources, there is not much to
be done other than, perhaps, to transcribe from that source
whatever information on the question might be deemed perti-
nent. If the data are compiled by the researcher himself but
are not based on a sample survey, such as company sales esti-
mates by territories, the study should include an investigation
of possible sources of error or bias in the data ranging from the
origin of the data to the last step involved in obtaining the esti-
mates. Thus, in judging the reliability of company territorial
estimates, one might look for possible errors or bias in the man-
ner in which the salesmen file their reports, how and when the
reports are compiled in the district offices, possible discrepancies
between the date a sale is made and the date it is reported, and
others.
If the data are obtained from a survey designed especially
to collect these data, a thorough investigation of their reliability
must include the two main aspects of the subject, sampling
error and bias, and, in addition, an attempt may be made to
validate the data.
Pros and Cons of Validation
To validate sample data means essentially to locate informa-
tion from other sources on the same characteristics covered by
the sample, which is then used as a yardstick against which to
gauge the accuracy of the sample data. If in a consumer prefer-
ence study in City Z, for example, classifying information were
compiled on marital status, occupation, age, and sex, data on
the same characteristics might be obtained for that city from
recent Census material or from other sources. If the sample
data exhibit much the same distributions for these character-
istics as the Census or other data such as falling well within
the range of sampling error the sample is said to be "vali-
dated." It is perhaps needless to note that a validation analysis
has no meaning if carried out on characteristics that were used
ANALYSIS AND INTERPRETATION OF DATA 235
as sample controls or on characteristics closely related to those
used as sample controls.
From an operational viewpoint the applicability of validating
procedures is restricted by the difficulty of finding accurate data
regarding the particular population under study. Census-of-
population data are compiled every ten years only, and for small
areas may be outdated by the time of release, three or four years
after the Census date. Other data relating to the same popula-
tion studied, the same time, the same characteristics, and the
same definitions are also generally difficult to find. Hence, a
validation analysis must be interpreted with considerable cau-
tion, as it is often difficult to ascertain whether differences un-
covered by the validation procedure are due to a possibly biased
sample or to lack of comparability between the various sets of
data.
Despite these limitations, validation can prove very useful
in pointing up weak spots in a sample survey, though validation
cannot serve as a substitute for measurement of sampling error
or evaluation of bias. At the least, it will indicate the direction
of the bias; it provides no measure of sampling error, and little,
if any, evidence as to the nature of the bias. In short, validation
to a researcher is in some ways analogous to the function of a
thermometer for a doctor: it acts as a warning signal when
something is out of order, but like a thermometer it is not a
foolproof indicator.
Formulating and Testing Working Hypotheses
General hypotheses as to the nature of the results that might
be obtained from the study have undoubtedly been formulated
at the preliminary investigation and planning stages. At this
stage, it remains to convert these hypotheses into explicit forms
that can be tested directly on the data. Where statistical tests
can be applied, this means setting up the so-called null hypoth-
esis of statistics to the effect that the observed differences do
not really exist in the population which is then tested with the
aid of appropriate significance tests. At the same time it is a
good idea to review all the data that have been collected with
an eye toward detecting relationships that might point to addi-
236 MARKETING RESEARCH
tional hypotheses or to revising the hypotheses originally form-
ulated. It is a rare study indeed where all pertinent hypotheses
have been postulated before the data are collected. 2
Interpretation of Results
A number of pitfalls face the researcher in interpreting the
results of an analysis. Probably the biggest one is the danger
of unwarranted generalizations. The urge to generalize is un-
derstandably great, for the value of a set of findings often in-
increases with its scope. Nevertheless, the possibility of gen-
eralizing is invariably restricted by the fact that a study is made
at a particular time, at a particular place, and in a certain way.
Broadening the scope of the results therefore necessitates care-
ful prior consideration of how they might be affected by the
change of time and place, not to mention the effect of different
methods of obtaining the same data. Even different ways of
asking for the same information can completely alter the pic-
ture. 8
An additional problem involved in generalizing is the danger
of going beyond the range of the observations. Thus, a study
may find that savings are linearly related to income for families
earning between $2,600 and $6,400 per year. To generalize
solely from this finding that savings are linearly related to in-
come for all families in that area would not be justified, since
these data provide no basis for believing that the same relation-
ship holds for those earning below $2,600 or above $6,400 per
year. This is one of the basic problems involved in forecasting
from past relationships, for when one has to consider conditions
outside the range of the observations, as is often the case, there
can be no assurance that the same relationships will continue.
Another pitfall is the tendency to identify correlation with
causation. If a close association exists between the stork popu-
* In this connection, rearranging of the data by doing percentaging in vari-
ous ways, regrouping classifications, etc. often brings to light hitherto unsus-
pected relationships. Numerous aids for carrying out such analyses are provided
in the Zeisel reference in the bibliography at the end of this Part, a necessary
reading for beginning marketing researchers.
* For an example, see S. Payne, The Art of Asking Questions (Princeton,
N. J.: Princeton University Press, 1951), chap. 7.
ANALYSIS AND INTERPRETATION OF DATA 237
lation and the number of births in the United States, nobody
is likely to conclude that the former influences the birth rate.
But if the relationship is a more plausible onesuch as, say, that
preference for prefabricated houses rises as income level falls
the urge to label one factor as a cause of the other is often
overwhelming. It can only be repeated here that lack of corre-
lation invariably proves absence of causation, but the presence
of correlation by itself does not prove causation. More positive
evidence is needed to establish causation, and it is here that a
priori reasoning plays an important role.
The confirmation or rejection of hypotheses on the basis of
statistical tests presents a similar pitfall. If the outcome of the
tests contradicts the hypothesis, it can be pretty safely rejected.
But if the outcome is in agreement with the hypothesis, the
validity of the hypothesis is not necessarily established, for the
same reason that correlation does not always signify causation:
the coincidence may be largely fortuitous. It is for this reason
that most analysts will not say in such a case that ''the tests con-
firm the hypothesis/' but rather they will take a more cautious
attitude and say either that "the tests do not disprove the hy-
pothesis" or "the results of the tests are in accord with the
hypothesis."
Verification of Findings
This step is in many ways analogous to validating survey data.
The question that one seeks to answer here is: how do the find-
ings obtained in the present study compare with those obtained
in similar past studies? If they agree with past studies, so much
additional support is provided for the validity of this study. If
they do not agree, an attempt should be made to reconcile the
differences. In either case, where similar studies do exist (a
fact that should already have been ascertained at the prelim-
inary investigation stage), a section on verification should be
included in the analysis. Such a section may provide strong
substantiation for the findings, and if it does not, this fact
should in all fairness be presented anyway. It is much better
for the researcher who conducted the study to point out possi-
238 MARKETING RESEARCH
ble inconsistencies in the results than to have someone else dis-
cover them at a later time.
Translation of Findings into Policy Recommendations
Many studies, especially those for business concerns, call for
a section on the practical implications of the findings, on what
sort of action might be indicated by the study. For the re-
searcher simply to translate his findings directly into the actions
they might imply cannot only prove dangerous, but may be
fatal from the point of view of ensuring favorable reception of
the study. For in translating the findings into recommenda-
tions for action, the researcher must take account (and show
that he is doing so) of established practices in the company as
well as of the feasibility of the actions indicated by the findings.
A study may point to the desirability of altering and combin-
ing a firm's sales routes, but existing personnel relationships and
the desire to maintain the morale of its staff may not render it
practicable to make such changes except gradually, as salesmen
leave the company or work their way up. Another study may
point to an opportunity for increased profits by enlarging the
scope of operations, but the company may simply not be in a
position to finance such an endeavor at the time.
A frequent procedure in cases where a clear conflict exists
between the nature of the findings and their practicability is to
present, first, the implications of the findings in the absence
of practical difficulties (sometimes placed under the heading of
"long-range objectives"), second, a demonstration of the effect
of practical difficulties as the findings are converted to policy
recommendations, and third, the policy recommendations
themselves. Whatever the method used, a tempering of the
findings in the light of actual conditions is essential for present-
ing policy recommendations.
If the findings are not conclusive enough to warrant submis-
sion of specific policy recommendations, an additional section
may be appended outlining what further work is required be-
fore policy recommendations could justifiably be made. Many
research reports include such a section more or less as a matter
ANALYSIS AND INTERPRETATION OF DATA 239
of course to point out for future reference the areas in which
further research would seem most fruitful.
A useful checklist of the main practical considerations that
should be taken into account in preparing policy recommenda-
tions is provided by Brown. 4 It is reproduced below:
1 . The condition of the company and the market should be taken into
account.
2. The probable opposition of executives should be analyzed.
3. Company officials should be given an active part in the interpretation.
4. The required course of action should be clearly stated. 5
5. A series of progressive changes should be recommended.
6. Recommendations should be interpreted into concrete gains from
following the proposed course of action.
The Cases in This Part
The cases in this part concentrate largely on the interpreta-
tion of data and partly on the translation of the findings into
policy recommendations. The only exception is the first case,
which raises the problem of validating a consumer survey in the
common situation where few data are available for the purpose.
The other five cases are concerned with the interpretation of
different types of data. Cases 2, 3, and 4 are problems dealing
solely with interpretation, all of them relating to sample data.
In each case, the researcher is faced with a mass of data and the
problem is to derive some meaning from them. Case 2 in par-
ticular, on the extent of use of business life insurance, presents
a hodgepodge of information, which may or may not be relevant
to the subject, and the researcher is asked to bring some order
out of it.
In Case 3, on listenership to radio programs, the data have
been organized, but the classifications and breakdowns pre-
sented are so numerous that the problem of interpretation is
rather complex and presents the danger of unwarranted gen-
eralizations.
4 L. O. Brown, Marketing and Distribution Research (New York: The Ron-
ald Press Co., 1949), pp. 566-71. This list is elaborated upon in these pages.
6 And even more clearly justified!
240 MARKETING RESEARCH
Case 4, on the effectiveness of advertising, poses a problem
familiar to all advertisers and also raises the question of possible
remedial action.
The data in Case 5 are secondary source material and present
an intermediary type of situation in which recommendations
for future research are needed on the basis of the material com-
piled to that time with regard to when retail stores should be
open.
The situation in Case 6 is one in which research concerning
the characteristics of a men's apparel store has been completed
and the problem is to interpret the results and present policy
recommendations.
AMERICAN MARKETING ASSOCIATION, The Technique of Marketing Research.
New York: McGraw-Hill Book Co., Inc., 1937. Chapter 16 presents useful
pointers on interpreting data, and Chapter 19 reviews briefly various statisti-
cal methods used in analyzing marketing data.
BLANKENSHIP, A. B. (ed.). How to Conduct Consumer and Opinion Research.
New York: Harper & Bros., 1946. Chapters 4, 5, 8, and 9 in particular con-
tain illustrations of analysis of marketing data.
BRADFORD, E. S. Marketing Research. New York: McGraw-Hill Book Co.,
Inc., 1951. A brief section on analysis and interpretation of data is included
on pp. 192-97.
BROWN, L. O. Marketing and Distribution Research. New York: The Ronald
Press Co., 1949. The last part of Chapter 23 and all of Chapter 24 discuss
the problems involved in drawing inferences from statistical data and contain
much useful information on interpreting research findings into business
policy.
HEIDINGSFIELD, M. S., and A. B. BLANKENSHIP. Market and Marketing Analy-
sis. New York: Henry Holt & Co., 1947. Of special interest to readers of
chapter 11 is a discussion concerning the role of the researcher in making
recommendations. Some writers advocate that research workers should work
only with inferences and conclusions gathered from data and avoid making
too many recommendations, leaving the recommendations to the manage-
ment to make. Others believe that making recommendations is part of the
researcher's responsibility.
HOBART, D. M. (ed.). Marketing Research Practice. New York: The Ronald
Press Co., 1950. The practical examples upon which this book is built con-
tain many interesting and useful illustrations of analyzing and interpreting
marketing data.
LORIE, J. H., and H. V. ROBERTS. Basic Methods of Marketing Research. New
York: McGraw-Hill Book Co., Inc., 1951. In the treatment of scientific
method in marketing research (Chapter 4) the authors give a brief treatment
of interpretation problems.
ANALYSIS AND INTERPRETATION OF DATA 241
LUCK, D. J., and H. G. WALES. Marketing Research. New York: Prentice-
Hall, Inc., 1952. An extensive discussion of Analysis of data and table ar-
rangements is contained in Chapter 13. Interpretation of data is treated in
some detail in Chapter 14 with examples in Chapters 17-19.
ZEISEL, H. Say It with Figures. New York: Harper & Bros., 1948. This book
contains much material throughout on analyzing and interpreting numerical
data and is well worth reading for this reason alone.
Part VII Cases
CASE I VALIDATION OF A CONSUMER
SURVEY
BACKGROUND
The sample design described in Case 3 of Part IV, pp. 1 30-
36, was used for a personal interview survey of families in Phila-
delphia in 1947 concerning their purchase and use of a class of
food products. Sampling results were as follows:
1 . Noncooperation of respondents plus the discarding by the
editor of several unreliable questionnaire returns reduced the
number of usable interviews to 479, out of the 533 planned, and
the number of areas in which interviews were made from 1 52
to 144.
2. Families not interviewed fell almost proportionately
among the rental groups. To adjust the quartiles no more than
three families had to be shifted from the edge of one group to
the edge of another, although distortion in some subrental
groups was larger. (Interviewers did not ask respondents for
either income or rentals, nor did they make estimates; respond-
ents were classified on the basis of the rental groups of the
blocks in which they lived; see page 132).
3. The subsample of apartment dwellers comprised 52 inter-
views, or 10.9 per cent of the total. (The 1940 Housing Census
for Philadelphia showed 10.1 per cent of families living in apart-
ment houses.)
4. The 70 planned interviews among nonwhites were com-
pleted. Interviewers also found that three blocks identified as
predominantly, or wholly, white in 1940 were solidly nonwhite
in 1947. The final results, therefore, included 87 usable inter-
views with nonwhite families, or 18.1 per cent of the total.
(The 1940 Census showed 12.9 per cent of nonwhites in the
city.)
Because of difficulties in sample construction and because of
these results, serious doubts were raised as to the validity of the
242
CASES IN ANALYSIS AND INTERPRETATION 243
sample, particularly as to whether or not it constituted a repre-
sentative sample of Philadelphia families.
METHODS OF VALIDATION
The principal methods of validation used were to check the
sample data against data that were available for the population
for characteristics not employed as sample controls.
Refrigeration ownership, for example, had not been used in
any way as a sample control or stratification. Possession of
mechanical refrigeration was thought to be importantly related
to the purchase and use of the food products being studied.
Were the families interviewed representative of Philadelphia
families with regard to refrigeration ownership?
Of the families interviewed, 326 reported having electric or
gas refrigerators, 1 32 had iceboxes, and 21 had none, other types,
or gave no answer. The 1940 Census estimated that 57 per cent
of Philadelphia families owned mechanical refrigerators. The
Edison Electric Institute Bulletin of August, 1945, summar-
izing a national study made earlier in the year, showed a gain
of 6 million units over the 15,364,000 mechanical refrigeration
units reported owned in 1940 by the Census. Thus, mechan-
ical refrigeration among all U. S. families had increased from
43 per cent in 1940 to more than 50 per cent in 1945. The 1945
figures given for refrigeration equipment in occupied urban
dwellings were: mechanical refrigerators, 68,3 per cent; ice
boxes, 25.8 per cent; other or none, 5.9 per cent.
Family size also had been uncontrolled. The family size dis-
tribution of the total sample as compared with the correspond-
ing Census data for 1940 for Philadelphia was as follows:
EXHIBIT 1
FAMILY SIZE DISTRIBUTION, ALL FAMILIES
Source
Members in family All families
1
2
3
4
5 or
more
N.A.
Total
Census 1940
Sample 1947
7.8
5.2
23.3
24.8
22.1
24.2
18.9
21.1
27.9
24.4
0.3
100%
100%
244 MARKETING RESEARCH
Family-size distributions for nonwhite families were:
EXHIBIT 2
FAMILY SIZE DISTRIBUTION, NONWHITES
Source
Members in family Nonwhite families
1
2
3
4
5 or
more
N.A.
Total
Census -1940
Sample 1947
11
8
25
25
19
13
15
18
30
36
100%
100%
Users of the food product being studied had been reported
by other researchers as comprising: 34 per cent of San Francisco
families (Call-Bulletin Study) in 1941; 23 per cent of all urban
families (General Foods) in 1942; and 55.9 per cent of Phila-
delphia families (Philadelphia Evening Bulletin) in 1946. The
latter figure was produced as part of a survey of many items.
The definition of the class of food product being compared
was unavailable, except that the earlier study had included all
varieties of this class of product whereas the latter had not.
The present study revealed that 64 per cent of families used
the product some time within the past year. Excluding families
whose last purchase had been more than one month previous
to the interview, user families were 56.1 per cent of the total.
In addition to variance in these percentages because of differ-
ences in definition, seasonal difference was not checked, the
time of interviewing for the 1946 survey not being given.
The Bulletin survey gave the following use-of-product per-
centages by rental groups (renters and owners had "reported"
rental values to interviewers) :
Rentals
Over $50
$40 to $49.99
$30 to $39.99
Under $30
Per cent of users
72.8
62.3
53.7
40.4
CASES IN ANALYSIS AND INTERPRETATION 245
The present study, using 1940 rental values, showed:
Rentals Per cent of users
$37 and over 85 ( 1st quartile)
$28.50 to $36.99 73 (2nd quartile)
$23.01 to $28.49 58 ( 3rd quartile)
$23 and under 38 (4th quartile)
ACTION NEEDED
1. What light do these data throw on the validity of the survey?
2. Someone suggested that the reliability of the data might be
increased by weighting the results. Evaluate the desirability
of this procedure in the present case.
CASE 2 EXTENT OF USE OF BUSINESS
LIFE INSURANCE
BACKGROUND
The investigation, which was to determine the extent to
which business life insurance is used by various types of busi-
nesses, (Part VI, Case 4) had reached the analysis stage. A re-
search worker had been assigned the job of working out an
interpretation and analysis of a series of tables that had been
prepared by one of the statistical clerks who was working on
the study.
Two samples had been selected one of business firms and
the other of life insurance salesmen. The first sample was
selected to ascertain the extent to which various types of busi-
nesses carried business life insurance. The sample of life in-
surance agents was selected to determine the types of policies
sold by agents for business life insurance purposes and other
purposes. Exhibits 1 through 8 are based on data collected
from business firms, and Exhibits 9 and 10 are based on data
gathered from agents.
The immediate reaction of the worker on studying the tables
was, "How do I organize the tables so that an understandable
interpretation can be made?" After making several futile at-
246
MARKETING RESEARCH
tempts at the assignment, the worker finally got in touch with
one of the directors of the project and asked for help.
EXHIBIT 1
NUMBER OF FIRMS INTERVIEWED, BY FORM OF BUSINESS
Form of business
Number of firms
Corporations .
323
Partnerships
125
Sole proprietorships
132
Total
580
EXHIBIT 2
NUMBER OF FIRMS INTERVIEWED, BY FORM OF BUSINESS AND CITY SIZE
Form of business
Number of
firms
Percentage
of group
Percentage of type
in entire sample
A. In two large cities
Corporations
Partnerships . . .
Sole Proprietorships
Total, all types
213
70
65
61.2
20.1
18.7
65.9
56.0
49.2
60.0
348
100.0
B. In five small cities
Coq^orations
Partnerships . .
Sole Proprietorships . . .
Total, all types
110
55
67
47.4
23.7
28.9
34.1
44.0
50.8
40.0
232
100.0
EXHIBIT 3
NUMBER OF FIRMS INTERVIEWED, BY TYPE OF BUSINESS ACTIVITY
Type of business activity
Number of
firms
Percentage
of total
Manufacturing
159
27.4
Manufacturing and merchandising
55
9.5
Wholesaling
64
11.0
Retailing
241
41.6
Wholesaling and retailing . ...
61
10.5
CASES IN ANALYSIS AND INTERPRETATION 247
EXHIBIT 4
NUMBER OF FIRMS INTERVIEWED, BY TYPE OF BUSINESS ACTIVITY
AND CITY SIZE
Type of business activity
Number of
firms
Percentage
of group
Percentage of
type in entire
sample
A. In two large cities
Manufacturing
Manufacturing and
merchandising
Wholesaling
Retailing
Wholesaling and retailing
Total, all types . .
B. In five small cities
Manufacturing . . .
Manufacturing and
merchandising
Wholesaling
Retailing
Wholesaling and retailing
Total, all types
108
37
46
117
40
31.0
10.6
13.2
33.7
11.5
67.9
67.3
71.9
48.5
65.6
60.0
32.1
32.7
28.1
51.5
34.4
40.0
348
51
18
18
124
21
100.0
22.0
7.8
7.8
53.3
9.1
232
1000
EXHIBIT 5
SIZE OF SAMPLE FIRMS, BY CITY SIZE
Number of
employees
Seven cities
Two large cities
Five small cities
Number
of firms
Per cent
of total
Number
of firms
Per cent
of total
Number
of firms
Per cenl
of total
1-5
82
116
157
91
69
62
3
14.1
20.0
27.1
15.7
11.9
107
0.5
36
60
90
62
50
49
1
10.3
17.2
25.9
17.8
14.4
14.1
0.3
46
56
67
29
19
13
2
19.8
24.1
28.9
12.5
8,2
5.6
0.9
6-10
11-25 . . .
26-50 . .
51-100
Over 100
Not disclosed
Total, all groups
580
100.0
348
100.0
232
100.0
248
MARKETING RESEARCH
EXHIBIT 6
SIZE OF INCORPORATED BUSINESSES IN SEVEN SELECTED CITIES
Number of
employees
Number
of firms
Number of
corporations
Per cent of
size-group
incorporated
Per cent of
all corpora-
tions in
size-group
1-5
82
11
13.4
3.4
640
116
37
31.9
11.5
11.25
157
81
51.6
25.0
26-50
91
70
76.9
21.7
51-100
69
63
91.3
19.5
Over 100
62
60
96.8
18.6
Not disclosed
3
1
33.3
0.3
Total, all groups
580
323
55.7
100.0
EXHIBIT 7
SIZE OF SAMPLE FIRMS, BY FORM OF BUSINESS
Number of
employees
Corporations
Partnerships
Sole
proprietorships
Num-
ber
Per cent
in size-
.group
Num-
ber
Per cent
in size-
group
Num-
ber
Percent
in size-
group
1-5
11
37
81
70
63
60
1
3.4
11.5
25.0
21.7
19.5
18.6
0,3
26
41
35
16
5
2
20.8
32.8
28.0
12.8
4.0
1.6
0,0
45
38
41
5
1
2
34.1
28.8
31.1
3.8
0.8
0.0
1.4
6-10
11-25
26-50
5MOO
Over 100
Not disclosed
Total, all groups
323
100.0
125
100.0
132
100.0
CASES IN ANALYSIS AND INTERPRETATION 249
EXHIBIT 8 *
BUSINESS LIFE INSURANCE CARRIED BY SIZE OF FIRM INTERVIEWED
Firms that reported carrying
Key man
Business
continuation
Either
type
Number of
Number
insurance
insurance
or both
employees
of firms
Per cent
Percent
Per cent
Num-
in size-
Num-
in size-
Num-
in size-
ber
group
ber
group
ber*
group
1-5.
82
10
12.2
6
7.3
16
19.5
6-10 . . .
116
16
13.8
15
12.9
31
26.7
11-25..
157
36
22.9
15
9.6
50
31.8
26-50 ..
91
32
35.2
9
9.9
40
44.0
51-100
69
27
39.1
9
13.0
34
49.3
Over 100
62
41
66.1
3
4.8
43
69.4
* In some cases these figures differ from the sum of figures for the two types
because some concerns have both types and are therefore listed in both columns.
EXHIBIT 9
REASONS GIVEN BY SALESMEN FOR BUYING RESISTANCE OF CLIENTS
Reason given
Num-
ber
of
clients
Number of cases assigned
specific rank
1
2
3
4
5
6
Client unable to reach a de-
cision
62
53
56
57
39
16
20
22
19
15
10
21
10
19
24
7
2
18
13
13
13
8
5
1
5
1
5
7
4
2
3
3
3
3
1
4
2
Client unable to understand
the need
Cannot afford it ....
Future of business uncertain .
Definitely no need for it
Doesn't believe in life insur-
ance
250
MARKETING RESEARCH
EXHIBIT 10
SALES OF BUSINESS LIFE INSURANCE, BY EXPERIENCE OF AGENT
Years of
experience
Percentage
of
answers
Percentage who reported selling
Business
continuation
Key man
Over 10
42.6
24.4
23,0
5.3
4.7
48.5
25.7
18.5
2.1
5.2
53.1
28.3
11.1
1.2
6.1
3-10
1-3
0-1
Not stated
THE PROBLEM
The worker had some questions to be answered before any
semblance of reasonable action could be undertaken. His first
question was, "What is the objective of the study?" The an-
swer given by the director was, 'To study the extent to
which business life insurance is used by various types of
business."
The next question was, "How many of the tables shall
be used in the interpretation and should others be pre-
pared?"
The answer to this question by the director was, "Use the
tables and prepare any others that are necessary to fulfill the
objectives of the study."
The director added, "Discard any tables that do not seem
to be suitable for the task assigned to you."
ACTION NEEDED
Prepare a one thousand word interpretive report utilizing the
tables that seem to be appropriate on the Extent of the Uses of
Life Insurance by Business, with particular reference to:
1. Characteristics of the segment of the market studied.
2. Analysis of purchasers of insurance by size, business opera-
tion, type of operation, and city.
CASES IN ANALYSIS AND INTERPRETATION 251
CASE 3 WHO LISTENS TO RADIO
PROGRAMS?
BACKGROUND
In the fall of 1946, a study of radio listenership in Champaign
County, Illinois, was made by means of the diary method.
This county contains the twin cities of Champaign-Urbana, a
number of small villages, and a substantial farm population. A
sample of 528 adults (18 years old or more) was selected from
the population of this county, and each adult was requested to
keep a diary of his radio listening for the week of November
3-9, 1946. The diary provided space for recording station lis-
tenership by 15-minute intervals from 6:00 A.M. to midnight
for seven consecutive days. In addition, classifying informa-
tion was secured on these adults in an initial personal interview
on such characteristics as sex, age, education, and place of resi-
dence.
THE PROBLEM
The objective of the study was to determine some of the fac-
tors that influence listenership to different types of radio
programs. In other words, what characteristics of an individual
seem to be most closely related to listenership of various types of
radio programs? To throw some light on this question, tabula-
tions were made of female listenership to eighty-eight individual
programs by three characteristics of the listener: education, area
of residence, and age. Study of female listenership was empha-
sized in the study because most men are not able to listen to a
radio when many programs are on. These eighty-eight pro-
grams were classified under six major headings: news, religion,
music, drama, audience participation, and homemaking.
The results of these tabulations are shown in the accompany-
ing exhibits. The percentages shown refer to the proportion
of women in that particular category the total number being
shown in the row labeled "number in sample" reported listen-
ing to each program. The last exhibit provides a special break-
down of sports listenership by sex and education groups.
252 MARKETING RESEARCH
EXHIBIT 1
PERCENTAGE OF WOMEN LISTENING TO SELECTED PROGRAMS, BY
EDUCATION GROUPS
T 1 * A
Education
Program
Time
Grade
school
High
school
Col-
lege
66
138
76
News
Single Commentators
Drew Pearson
Sun.
5:00 P.M.
21.2%
12.9%
11.8%
Paul Harvey
Sun.
10:00 P.M.
4.5
7.2
5.3
Gabriel Heatter
Sun.
9:00 P.M.
7.5
1.4
1.4
Gabriel Heatter
Wed.
8:00 P.M.
8.2
8.4
6.6
Round Table Discussions
Northwestern Reviewing
Stand
World Front
Sun.
Sun.
10:30 A.M.
11:OOA.M.
*
*
*
1.4
2.6
1.3
Town Meeting of the Air .
Gossip Columnists
Louella Parsons
Thurs
Sun.
7:30 P.M.
8:15 P.M.
4.5
2.3
5.5
13.8
9.5
9.2
Jimmy Fidler
Sun.
8: 30 P.M.
46
10.1
6.6
Walter Winchell
Sun.
8:00 P.M.
17.8
26.3
11.9
Religion
Little Brown Church ....
Local Church Service
Hymns of All Churches .
Salt Lake City Choir . . .
Music
Classical
New York Philharmonic . .
NBC Symphony
Chicago Theater of the Air .
"400" Hour
Sun.
Sun.
M-F
Sun.
Sun.
Sun.
Sat.
M-Sat
9:15 A.M.
11:00 A.M.
9:30 A.M.
8:15 A.M.
2:00 P.M.
4:00 P.M.
9:00 P.M.
7:00 A.M.
16.7
4.5
14.5
3.0
*
1.9
3.4
#
9.4
6.7
11.4
1.5
3.3
*
2.2
13
4.4
5.3
4.8
2.6
11.4
5.9
4.3
4.3
Semiclassical
RCA Victor Show . . .
Sun.
1:00 P.M.
2.3
*
6.6
Harvest of Stars
Hour of Charm
Sun.
Sun.
1:30 P.M.
3:30 P.M.
.1.5
*
1.4
6.2
7.9
17.1
Voice of Firestone
Mon.
7:30p.M.
2.3
9.8
13.1
Telephone Hour
Mon.
8:00 P.M.
2.3
10.5
15.8
Contented Program . , , .
Mon.
9:00 P.M.
3.8
5.4
7.9
* Less than 1 per cent
CASES IN ANALYSIS AND INTERPRETATION
EXHIBIT 1 Continued
253
1
Education
Program
rime
Grade
school
High
school
Col-
lege
66
138
76
Popular
Shaeffer Parade
Sun.
2:00 P.M.
5.3%
1.8%
8.5%
Manhattan Merry-Go-
Round
Sun.
8:00 P.M.
6.8
11.9
15.1
American Album of
Familiar Music
Sun.
8:30 P.M.
1.5
12.0
15.8
Your Hit Parade
Sat.
8:00 P.M.
9.1
11.8
9.2
Fred Waring
Folk Music
Gene Autry
M-F
Sun.
10:00 A.M.
6:00 P.M.
1.2
2.3
3.6
5.8
6.8
2.0
Prairie Ramblers; Sage
Riders
M-F
7:00 A.M.
11.4
6.5
4.2
WLS Barn Dance
Sat.
7:30 P.M.
18.2
17.9
11.8
Dinner Bell Time
M-F
12:30 P.M.
17.4
6.1
8.0
Drama
Good Drama
Cavalcade of America . . .
Lux Radio Theater ....
Theatre Guild on the Air. .
Family Type Drama
Blondie .
Mon.
Mon.
Sun.
Sun.
7:00 P.M.
8:00 P.M.
9:00 P.M.
6:30 P.M.
1.5
12.1
4.6
8.4
7.3
19.1
4.0
13.4
6.5
12.1
11.5
7.9
The Great Gildersleeve . . .
A Date With Judy
Wed.
Tues.
7:30 P.M.
7:30 P.M.
7.6
6.1
14.5
12.0
9.2
11.8
The Aldrich Family
Thurs.
7:00 P.M.
7.6
12.7
5.9
Baby Snooks
Fri.
7:00 P.M.
8.3
10.8
3.9
Detective Shows
Counterspy
Sun.
4:30 P.M.
4.6
4.0
1.3
Crime Doctor
Sun.
7:30 P.M.
4.6
5.8
1.3
Inner Sanctum
Mon.
7:00 P.M.
6.8
7.6
2.6
Mr District Attorney
Wed.
8:30 P.M.
7.6
1.5
6.6
This Is Your FBI
Fri.
7:30 P.M.
5.3
8.7
2.0
The Lone Ranger
MWF
6:30 P.M.
14.1
11.9
5.2
Comedy and Variety
Jack Benny Show
Sun.
6:00 P.M.
10.6
25.7
21.1
McCarthy and Bergen ....
Fred Allen
Sun.
Sun.
7:00 P.M.
7:30 P.M.
13.6
37.3
26.5
21.7
39.5
30.3
Victor Borge
Mon.
8:30 P.M.
3.0
9.1
13.8
Bing Crosby
Wed.
9:00 P.M.
12.9
13.7
11.8
254
MARKETING RESEARCH
EXHIBIT I Continued
]
Education
Program
Time
Grade
school
High
school
Col-
lege
Number in sample
66
138
76
Housewife Variety
Breakfast Club
M-F
8-00 A.M.
14.7%
16.6%
6.4%
Breakfast in Hollywood . .
Family Serial
One Man's Family . . .
M-F
Sun.
10:00 A.M.
2 -30 P.M.
17.6
6.1
19.8
6.9
9.2
10.5
Lum and Abner
M-F
7-00 P M.
19.6
13.8
7.4
Soap Operas
Our Gal Sunday
M-F
11 -45 A.M.
4.9
3.9
1.5
Guiding Light . ...
Life Can Be Beautiful . .
Lorenzo Jones
M-F
M-F
M-F
12:45 P.M.
2:00 P.M.
3:30 P.M.
6.7
4.3
13.3
5.5
8.0
6.2
3.4
3.9
3.4
Portia Faces Life . . .
M-F
4:15 P.M.
10.3
4.2
3.2
Audience Participation
Quiz Shows
Quiz Kids
Sun.
3-00 P.M.
10.6
5.8
11.8
Take It or Leave It
Sun.
9:00 P.M.
7.6
11.2
7.9
Dr I Q
Mon.
8:30 P.M.
6.8
8.7
3.9
Break the Bank
Fri
8:00 P.M.
6.1
4.4
1.3
Twenty Questions . .
Sat
7:00 P.M.
1.5
*
5.9
Studio Interviews
Queen for a Day
M-F
1:00 P.M.
*
2.3
*
Bride and Groom
M-F
1:30 P.M.
12.8
7.9
4.0
Ladies Be Seated
M-F
2: 00P.M.
7.1
7.5
2.1
Amateur Shows
Sachs* Amateur Hour
Sun.
12:30 P.M.
20.8
13.6
5.9
Homemaking
Food Magician
M-F
11:00 A.M.
1.5
1.5
1.6
Betty Crocker
M-F
1:30 P.M.
9.6
5.5
3.4
Mary Lee Taylor
Sat.
9:30 A.M.
1.5
2.5
*
CASES IN ANALYSIS AND INTERPRETATION
EXHIBIT 2
PERCENTAGE OF WOMEN LISTENING TO SELECTED PROGRAMS, BY
AREA OF RESIDENCE
255
Program
Time
Area of residence
Urban
Village
Farm
Number in sample
150
74
75
News
Single Commentators
Drew Pearson
Sun.
5:00 P.M.
12.0%
10 8%
20.0%
Paul Harvey
Sun.
10:00 P.M.
5.3
* v/w /o
6.8
5.3
Gabriel Heatter
Sun.
9:00 P.M.
2.0
#
4.0
Gabriel Heatter
Wed.
8:00 P.M.
7.3
12.0
10.7
Round Table Discussions
Northwestern Reviewing
Stand
Sun.
10:30 A.M.
1.3
<c
1.3
World Front
Sun.
11:00 A.M.
*
1.3
1.3
Town Meeting of the Air . .
Thurs
7:30 P.M.
5.3
10.6
10.3
Gossip Columnists
Louella Parsons
Sun.
8:15 P.M.
167
4.0
9.3
Jimmy Fidler
Sun.
8:30 P.M.
12.0
4.0
5.3
Walter Winchell
Sun.
8:00 P.M.
27.3
21.6
28.0
Religion
Little Brown Church
Sun.
9:15 A.M.
8.5
7.7
12.0
Local Church Service ....
Sun.
11:00 A.M.
9.3
5.4
*
Hymns of All Churches . .
M-F
9:30 A.M.
5.3
10.8
14.7
Salt Lake City Choir
Sun.
8:15 A.M.
4.0
3.3
$
Music
Classical
New York Philharmonic
Sun.
2:00 PM.
7.1
6.9
*
NBC Symphony
Sun.
4:00 P.M.
3.4
*
2.0
Chicago Theater of the Air
"400" Hour
Sat.
M-Sat
9:00 P.M.
7:00 A.M.
4.1
2.5
3.7
1.7
#
2.6
Semiclassical
RCA Victor Show
Sun.
1:00 P.M.
4.7
*
#
Harvest of Stars
Sun.
1:30 P.M.
5.6
*
*
Hour of Charm
Sun.
3: 30 P.M.
11.0
6.7
4.0
Voice of Firestone
Vfon.
7: 30p.M.
10.0
5.4
10.7
Telephone Hour
Vlon.
8:00 P.M.
11.9
5.4
9.3
Contented Program
Mon.
9:OOp.M.
8.9
6.1
*
Popular
Shaeffer Parade
Sun.
2:00 P.M.
5.0
4.1
2.7
* Less than 1 per cent.
256
MARKETING RESEARCH
EXHIBIT 2 Continued
Program
Time
Area of residence
Urban
Village
Farm
Number in sample '.
150
11.0%
12.3
14.6
5.3
4.7
3.4
13.6
5.7
5.3
17.7
8.5
13.0
12.9
11.3
14.4
10.1
4.7
3.3
9.2
11.6
6.3
8.3
29.0
34.7
34.7
11.9
11.3
7.4
10.6
74
15.6%
12.9
6.2
2.7
2.7
5.3
15.3
6.1
4.0
16.5
7.1
8.8
8.0
10.9
16.4
8.9
4.1
6.7
2.8
12.0
*
6.8
16.2
25.0
19.0
2.7
20.7
19.3
23.0
75
7.4%
3.3
5.3
#
4.0
15.3
25.0
15.4
9.3
10.6
1.3
6.7
12.0
10.0
10.7
4.7
2.0
4.0
1.3
8.7
11.4
15.3
13.4
14.0
8.0
8.0
9.3
14.0
19.4
Music continued
Manhattan Merry-Go-
Round
American Album of
Familiar Music
Sun. 8:00 P.M.
Sun. 8: 30 P.M.
Sat. 8:OOp.M.
M-F 10:00 A.M.
Sun. 6:00 P.M.
M-F 7:00 A.M.
Sat. 7: 30 P.M.
M-F 12: 30 P.M.
Mon. 7:00 P.M.
Mon. 8:OOp.M.
Sun. 9:00 P.M.
Sun. 6: 30 P.M.
Wed. 7: 30 P.M.
Thurs. 7: 30 P.M.
Thurs. 7:00 P.M.
Fri. 7:00 P.M.
Sun. 4:30 P.M.
Sun. 7:30 P.M.
Mon. 7:00 P.M.
Wed. 8: 30P.M.
Fri. 7: 30 P.M.
MWF 6: 30P.M.
Sun. 6:00 P.M.
Sun. 7:00 P.M.
Sun. 7:30p.M.
Mon. 8: 30p.M.
Wed. 9:OOp.M.
M-F 8:00 A.M.
M-F 10:00 A.M.
Your Hit Parade
Fred "Waring . . . .
Folk Music
Gene Autry . . .
Prairie Ramblers; Sage
Riders
WLS Barn Dance
Dinner Bell Time
Drama
Good Drama
Gavalcade of America .
Lux Radio Theater
Theatre Guild on the Air. .
Family Type Drama
Blondie .
The Great Gildersleeve . . .
A Date With Judy
The Aldrich Family
Baby Snooks
Detective Shows
Counterspy
Crime Doctor
Inner Sanctum
Mr. District Attorney
This Is Your FBI
The Lone Ranger
Comedy and Variety
Jack Benny Show
McCarthy and Bergen ....
Fred Allen
Victor Borge
Bing Crosby
Housewife Variety
Breakfast Club
Breakfast in Hollywood . . .
CASES IN ANALYSIS AND INTERPRETATION
EXHIBIT 2 Continued
257
Program
Time
Area of residence
Urban
Village
Farm
Number in sample
150
9.7%
6.0
6.0
6.0
4.6
2.6
3.3
11.4
12.6
8.0
3.3
4.3
1.3
4.0
3.3
11.7
1.3
6.6
*
74
3.4%
19.7
8.1
4.1
4.1
6.8
4.1
5.4
8.1
7.5
1.3
1.9
1.3
6.8
6.8
5.6
2.7
4.1
1.9
75
10.0%
30.7
9.3
4.0
9.3
13.3
12.0
8.0
7.4
5.3
7.3
*
2.7
13.3
7.4
23.0
2.7
2.7
*
Family Serial
One Man's Family
Sun.
M-F
M-F
M-F
M-F
M-F
M-F
Sun.
Sun.
Mon.
Fri.
Sat.
M-F
M-F
M-F
Sun.
M-F
2:30 P.M.
7:00 P.M.
11:45 A.M.
12:45 P.M.
2:OOp.M.
3: 30P.M.
4:15 P.M.
3:00 P.M.
9:00 P.M.
8: 30 P.M.
8:00 P.M.
7:00 P.M.
1:00 P.M.
1: 30P.M.
2:00 P.M.
12: 30P.M.
11:OOA.M.
1: 30P.M.
9:30 A.M.
Lum and Abner
Soap Operas
Our Gal Sunday
Guiding Light
Life Can Be Beautiful . .
Lorenzo Jones
Portia Faces Life
Audience Participation
Quiz Shows
Quiz Kids
Take It or Leave It
Dr. I. Q.
Break the Bank
Twenty Questions
Studio Interviews
Queen for a Day
Bride and Groom
Ladies Be Seated
Amateur Shows
Sachs* Amateur Hour
Homemaking
Food Magician
Betty Crocker
M-F
Sat.
Mary Lee Taylor
258 MARKETING RESEARCH
EXHIBIT 3
PERCENTAGE OF WOMEN LISTENING TO SELECTED PROGRAMS, BY AGE GROUPS
A
ge
Program
Time
18-24
years
25-39
years
40-59
years
60
years
and
over
Number in sample
44
97
108
43
News
Single Commentator
Drew Pearson . ...
Sun. 5:00 P.M.
2.3%
9.3%
19.5%
20.9%
Paul Harvey ....
Sun. 10:00 P.M.
4.5
5.2
8.3
2.3
Gabriel Heatter . .
Sun. 9: 00 P.M.
*
2.1
2.8
7.0
Gabriel Heatter
Wed. 8: 00 P.M.
*
9.3
11.1
4.7
Round Table Discussions
Northwestern Review-
ing Stand . .
Sun. 10: 30 A.M.
*
2.1
*
*
World Front .
Sun. 11:00 A.M.
*
3.1
*
*
Town Meeting of the
Air
Thurs. 7: 30 P.M.
1.7
5.7
7.0
11.6
Gossip Columnists
Louella Parsons
Sun. 8: 15 P.M.
6.8
7.2
16.7
16.3
Jimmy Fidler . ...
Sun. 8: 30 P.M.
2.3
7.2
11.1
9.3
Walter Winchell . .
Sun. 8:00 P.M.
11.4
17.5
35.2
34.9
Religion
Little Brown Church
Sun. 9: 15A.M.
4.5
5.9
10.8
18.6
Local Church Service
Sun. 11:00 A.M.
2.3
3.1
8.3
12.2
Hymns of All Churches
Salt Lake City Choir..
M-F 9: 30A.M.
Sun. 8: 15 A.M.
*
*
15.5
2.1
5.6
2.8
9.3
5.8
Music
Classical
New York Philhar-
rnonic
Sun. 2:00 P.M.
5.3
5.1
12.5
5.8
NBC Symphony
Sun. 4:00 P.M.
4.5
1.0
2.1
4.1
Chicago Theater of
the Air ....
Sat. 9:00 P.M.
2.3
3.4
3.9
1.7
"400" Hour
M-Sat 7:OOA.M.
1.7
1.8
3.0
2.3
Semiclassical
RCA Victor Show .
Sun. 1:00 P.M.
6.8
1.0
2.3
2.3
Harvest of Stars . . .
Sun. 1:30 P.M.
6.8
1.0
3.9
2.3
Hour of Charm
Sun. 3: 30 P.M.
20.5
5.2
7.4
3.5
* Less than 1 per cent.
CASES IN ANALYSIS AND INTERPRETATION
EXHIBIT 3-- Continued
259
A
ge
Program
Time
60
years
18-24
25-39
40-59
and
years
years
years
over
Number in sample
44
97
108
43
Music continued
Voice of Firestone . . .
Mon. 7:30 P.M.
6.8%
11.3%
7.4%
14.0%
Telephone Hour
Mon. 8:00 P.M.
11.4
9.8
7.4
16.3
Contented Program
Mon. 9:00 P.M.
6.8
7.7
4.1
9.3
Popular
Shaeffer Parade
Sun. 2:00 P.M.
4.5
3.1
5.6
5.8
Manhattan Merry-Go-
Round
Sun. 8:00 P.M.
18.2
13.9
9.2
11.6
American Album of
Familiar Music
Sun. 8: 30 P.M.
12.5
11.9
9.7
10.5
Your Hit Parade . . .
Sat. 8:00 P.M.
11.4
8.9
9.6
2.3
Fred Waring ....
M-F 10:00 A.M.
6.8
5.2
*
*
Folk Music
Gene Autry
Sun. 6:00 P.M.
9.1
6.2
1.8
10.5
Prairie Ramblers; Sage
Riders
M-F 7: 00 A.M.
4.5
8.2
4.6
2.3
WLS Barn Dance .
Sat. 7: 30 P.M.
15.0
18.8
16.7
14.0
Dinner Bell Time . .
M-F 12: 30P.M.
1.1
6.2
12.0
9.3
Drama
Good Drama
Cavalcade of America
Mon. 7:00 P.M.
4.5
10.3
4.6
7.0
Lux Radio Theater
Mon. 8:00 P.M.
14.8
14.7
16.2
4.7
Theatre Guild on the
Air ....
Sun. 9:00 P.M.
8.5
6.2
5.8
4.7
Family Type Drama
Blondie
Sun. 6.30 P.M.
15.9
13.9
7.8
8.1
The Great Gildersleeve
Wed. 7: 30p.M.
15.9
9.8
83
18.6
A Date With Judy
Thurs. 7: 30 P.M.
10.3
15.5
7.4
7.0
The Aldrich Family .
Thurs. 7:00 P.M.
10.3
21.6
9.3
14.0
Baby Snooks
Fri. 7:00 P.M.
3.4
12.4
9.8
4.7
Detective Shows
Counterspy
Sun. 4: 30 P.M.
6.8
1.0
5.1
3.5
Crime Doctor ....
Sun. 7: 30 P.M.
9.1
6.2
2.8
1.2
Inner Sanctum
Mon. 7:00 P.M.
6.8
5.2
5.5
23
Mr. District Attorney
Wed. 8: 30 P.M.
6.8
21.1
6.5
2.3
This Is Your FBI . .
Fri. 7: 30 P.M.
4.5
8.2
5.1
4.7
The Lone Ranger
MWF 6: 30p.M.
4.5
18.5
5.6
8.1
260
MARKETING RESEARCH
EXHIBIT 3 Continued
A*
1*
Program
Time
18-24
years
25-39
years
40-59
years
60
years
and
over
Number in sample
44
97
108
43
Comedy and Variety
Jack Benny Show
McCarthy and Bergen
Fred Allen
Victor Borge
Sun. 6:00 P.M.
Sun. 7:00 P.M.
Sun. 7: 30 P.M.
Mon. 8: 30 P.M.
15.9%
22.7
18.2
4.5
24.7%
28.3
25.8
10.3
15.3%
27.3
21.7
6.5
33.7%
27.9
22.1
16.3
Bing Crosby
Wed. 9:00 P.M.
13.6
17.0
11.5
5.9
Housewife Variety
Breakfast Club
Breakfast in Hollywood
Family Serial
One Man's Family . .
Lum and Abner
M-F 8: 00 A.M.
M-F 10:00 A.M.
Sun. 2:30p.M.
M-F 7:00 P.M.
1.2
6.8
6.8
6.8
22.4
22.7
6.2
14.4
10.4
11.6
6.5
13.9
4.7
10.4
22.1
16.3
Soap Operas
Our Gal Sunday
M-F 11:45 A.M.
4.5
3.1
2.8
2.3
Guiding Light
M-F 12:45 P.M.
2.3
2.1
*
2.3
Life Can Be Beautiful
Lorenzo Jones ....
Portia Faces Life . .
Audience Participation
Quiz Shows
Quiz Kids
M-F 2:00 P.M.
M-F 3: 30P.M.
M-F 4: 15 P.M.
Sun. 3: 00 P.M.
6.8
4.5
2.3
4.6
6.2
3.1
5.2
2.6
1.9
7.4
3.7
11.1
11.6
11.6
14.0
22.1
Take It or Leave It
Dr. I. Q.
Sun. 9:00 P.M.
Mon 8 -30 P.M.
11.4
4.5
7.2
8.8
10.1
7.4
7.0
16.3
Break the Bank ....
Twenty Questions . . .
Studio Interviews
Queen for a Day ....
Bride and Groom ....
Ladies Be Seated
Amateur Shows
Sachs* Amateur Hour.
Homemaking
Food Magician
Fri. 8:00 P.M.
Sat 7:00 P.M.
M-F 1:00 P.M.
M-F 1:30 P.M.
M-F 2:OOp.M.
Sun. 12: 30 P.M.
M-F 11:00 A.M.
*
2.3
*
*
#
12.5
*
7.2
*
1.0
9.3
9.3
11.6
3.1
*
3.7
1.9
8.3
2.3
13.2
1.9
4.7
4.7
2.3
2.3
2.3
7.1
2.3
Betty Crocker
Mary Lee Taylor . .
M-F 1:30 P.M.
Sat 9: 30A.M.
6.8
*
4.1
2.6
2.8
1.4
14.0
*
CASES IN ANALYSIS AND INTERPRETATION 261
EXHIBIT 4
PERCENTAGE OF LISTENING BY WOMEN AND MEN TO SELECTED
SUNDAY PROGRAMS, URBAN LISTENERS
Program
Time
Women
Men
Number in sample
150
112
News
Single Commentators
Drew Pearson
5:00 P.M.
12.0%
10.7%
Paul Harvey
10:00 P.M
5.3
7 1
Gabriel Heatter
9:00 P.M.
2.0
2.7
Round Table Discussions
Northwestern Reviewing Stand . .
World Front
10:30 A.M.
11:OOAM
1.3
*
*
#
Gossip Columnists
Louella Parsons
8:15PM
16.7
14 3
Jimmy Fidler
8:30 P.M.
12.0
12.5
Walter Wincbell .
8:00 PM
27.3
22 3
Religion
Little Brown Church
9:15 A.M.
8 5
39
Local Church Service . .
11:00 AM.
9 3
5 8
Salt Lake City Choir
8:15 A.M.
40
1 4
Music
Classical
New York Philharmonic
NBC Symphony ....
2:00 P.M.
4-00 PM
7.1
34
4.a
29
Semiclassical
RCA Victor Show
1:00 P.M.
47
3.6
Harvest of Stars
Hour of Charm
1:30 P.M.
3:30 P.M.
5.6
11.0
1.8
6.8
Popular
Shaeffer Parade
2:00 P.M.
5.0
3.2
Manhattan Merry-Go-Round ....
American Album of Familiar Music
Folk Music
Gene Autry
8:00 P.M.
8: 30 P.M.
6:00 P.M.
11.0
12.3
4.7
8.5
9.8
*
Drama
Good Drama
Theatre Guild on the Air
9:00 P.M.
8.5
6.0
* Less than 1 per cent.
262
MARKETING RESEARCH
EXHIBIT 4 Continued
Program
Time
Women
Men
Number in sample
150
112
Family Type Drama
Blondie
6:30 P.M.
13.0%
13 8%
Detective Shows
Counterspy
4:30 P.M.
4.7
45
Crime Doctor
7:30 P.M.
3.3
63
Comedy and Variety
Jack Benny Show
6-00 PM
290
25 5
Jack Parr
6:00 P.M.
McCarthy and Bergen .
7-00 p M
347
29 1
Alec Templeton
7:00 P.M.
Fred Allen
7:30 P.M
34.7
308
Family Serial
One Man's Family
2:30 P.M.
9.7
32
Audience Participation
Quiz Shows
Quiz Kids
3:00 P.M.
114
36
Take It or Leave It
9:00 P.M.
12.6
103
Amateur Shows
Sachs' Amateur Hour
12:30 P.M.
11.7
69
Sports
College Football (Sat.)*
Sat P M
158
183
Professional Football b
Sun. P.M.
1.7
6.4
Baseball
Sun. P.M.
* Broadcast on Saturday, Nov. 9. Games included were Army vs. Notre
Dame and Indiana vs. Northwestern.
b Bears vs Packers.
c Included three games, Cubs, Cardinals, White Sox and opponents.
ACTION NEEDED
1. What inferences do these tables allow to be made regarding
the relationship of the characteristics studied to program lis-
tenership?
2. Judging by these results, what additional tabulations would
it seem desirable to obtain?
CASES IN ANALYSIS AND INTERPRETATION
EXHIBIT 5
PERCENTAGE OF LISTENING TO SELECTED SPORTS PROGRAMS BY SEX
AND EDUCATION GROUPS
263
Sex and Education
Type of sports program
College
football
Professional
football
Baseball
Women
Grade school
12.1%
16.7
19.1
12.7
11.8
29.4
4.0%
3.2
*
10.5
10.3
4.4
3.9%
9.2
5.5
11.0
15.8
6.7
High school
College
Men
Grade school . . .
High school
College
* Less than 1 per cent.
CASE 4 EVALUATING THE EFFECTIVENESS
OF ADVERTISING
THE PROBLEM
One of the principal objectives of the survey on public atti-
tudes toward prefabricated housing (Part IV, Case 3), it will be
recalled, was to ascertain the effect of advertising and other
prefabricated housing literature on these attitudes. The in-
formation gleaned on this question from the survey was con-
densed into a number of analytical tables and charts, which
are reproduced on pp. 264-70. The problem is to determine
what inferences on this subject would seem warranted from
an examination of these data.
DATA AVAILABLE
Two aspects of the subject are covered in these tables and
charts, namely, the degree to which literature and material on
prefabrication is noted, i.e., recalled, and the effect of this
material on the reader's attitudes toward the product when this
264
MARKETING RESEARCH
advertising is noticed. (In this respect, seeing prefabricated
houses was treated as a form of advertising, which, of course, it
is.) Exhibits 1-7 relate to the first question, presenting statistics
on the percentage of respondents in particular population
groups reporting notice of any or none of four different forms
of prefab material: articles, advertisements, manufacturers' pro-
motional literature, and visual inspection.
The second question, the effect of prefab literature or sight
of prefab houses on the respondent's attitude, is treated in the
charts on pp. 267-69 and Exhibits 10 and 12. The chart on
page 267 portrays a percentage breakdown of the persons notic-
ing each form of prefab material according to its effect on them,
EXHIBIT 1
PERCENTAGE OF RESPONDENTS REPORTING NOTICE OF PREFAB HOUSES
OR LITERATURE, BY TYPE OF RESIDENCE
Type of Notice
Type of present residence
Standard
house
Prefab
Apart-
ment
Total
Article
46%
38
19
12
5
25
76%
69
31
100
52%
46
26
10
26
48%
41
21
17
4
23
Advertisement . .
Promotional literature
Inspection *
Not stated
No notice
Total
145%
345
276%
30
160%
64
154%*
439
Number of respondents
* Percentages total more than 100 because many respondents gave more than
one reply.
1 Only one out of every nine non-prefab dwelling families reported having
seen a prefab house. However, there seems to have been some confusion about
the meaning of this particular question, many persons not realizing that they
were being asked whether they had seen prefab houses. The only persons who
acknowledged having seen prefabs, except prefab dwellers, were apparently those
who had inspected model prefab homes, and possibly not even all of this group.
A follow-up survey of a subsample of 71 families revealed that actually about
three fourths of the respondents had seen prefabs at one time or another, at least
from the outside. Some slight adjustments in the interpretation of the attitudes
of those who had seen prefab houses was made on the basis of information ob-
tained from these follow-up interviews.
CASES IN ANALYSIS AND INTERPRETATION 265
EXHIBIT 2
PERCENTAGE OF RESPONDENTS REPORTING NOTICE OF PREFAB
HOUSES OR LITERATURE, BY SEX
Type of notice
Male
Female
Total
Article .
60%
45%
51%
Advertisement
49
37
42
Promotional literature
29
18
22
Inspection
6
14
11
Not stated
4
4
4
No notice
19
26
23
Total
167%
144%
153%
Number of respondents
168
270
438
EXHIBIT 3
PERCENTAGE OF RESPONDENTS REPORTING NOTICE OF PREFAB HOUSES
OR LITERATURE, BY MARITAL STATUS
Type of notice
Single
Married
Other
Total
Article
64%
51%
35%
51%
Advertisement
58
41
35
42
Promotional literature . ...
22
24
11
22
Inspection ....
16
10
13
11
Not stated . . ....
2
4
4
4
No notice
31
20
37
23
Total
193%
150%
135%
153%
Number of respondents ...
45
347
46
438
EXHIBIT 4
PERCENTAGE OF RESPONDENTS REPORTING NOTICE OF PREFAB
HOUSES OR LITERATURE, BY OCCUPATION
Type of notice
Prof.,
Mgr.,
Prop.
Cleri-
cal,
sales
Skilled,
semi-
skilled
Un-
skilled
Other
Total
Article . .
62%
50%
37%
28%
50%
50%
Advertisement ....
Promotional literature .
Inspection
48
27
11
50
23
12
30
22
12
30
8
22
40
18
3
41
22
11
Not stated
4
4
6
2
2
4
No notice
12
21
31
30
27
22
Total
164%
160%
138%
120%
140%
150%
Number of respondents . .
168
48
90
40
88
434
266 MARKETING RESEARCH
EXHIBIT 5
PERCENTAGE OF RESPONDENTS REPORTING NOTICE OF PREFAB HOUSES OR
LITERATURE, BY SIZE OF HOUSEHOLD
Type of notice
One
Two
Three
Four
Five or
more
Total
Article
35%
54%
54%
52%
39%
50%
Advertisement
Promotional literature . .
Inspection
28
12
12
48
24
13
45
27
8
39
20
12
29
14
4
41
22
11
Not stated
2
4
6
3
2
No notice
22
26
20
19
12
21
Total
111%
169%
160%,
145%
98%
148%
Number of respondents
40
135
109
100
49
433
EXHIBIT 6
PERCENTAGE OF RESPONDENTS REPORTING NOTICE OF PREFAB HOUSES OR
LITERATURE, BY AGE OF RESPONDENT
Type of notice
Under 21
21-34
35-49
50 and over
Total
Article
75%
63%
47%
39%
51%
Advertisement
62
52
36
33
42
Promotional literature
Inspection
12
32
8
19
12
14
14
22
11
Not stated
12
4
2
4
4
No notice
12
15
24
33
23
Total
173%
174%
140%
137%
153%
Number of respondents ....
8
175
100
153
436
EXHIBIT 7
PERCENTAGE OF RESPONDENTS REPORTING NOTICE OF PREFAB
HOUSES OR LITERATURE, BY INCOME
Type of notice
Under
$1500
$1500-
$2599
$2600-
$4199
$4200-
$6599
$6600
and over
Total
Article
42%
42%
54%
52%
74%
53%
Advertisement . . .
Promotional literature .
Inspection
25
11
14
32
12
8
47
19
12
49
32
12
49
32
4
44
22
11
Not stated
3
7
3
4
6
4
No notice
33
25
26
13
13
21
Total . . .
Number of respondents
128%
36
126%
60
161%
129
162%
114
178%
47
155%
386
CASES IN ANALYSIS AND INTERPRETATION 267
as ascertained by direct questioning. The relative breakdowns
by type of material noticed of answers to the question, "What
do you think are the advantages and disadvantages of prefabri-
cated houses against standard houses?" are presented in the
chart on page 268. The category "written material" includes
all three forms of prefab literature, inasmuch as little difference
was apparent by the individual forms.
The number of advantages per disadvantage noted in the
chart on page 268 and the percentage of respondents preferring
to purchase a prefabricated rather than a standard house if they
had only $8,000 to spend is cross-classified against the effect of
prefab literature in Exhibit 1 0. The base for the percentages in
the last column in the table is, in each case, the total number
of respondents reporting the particular "notice and effect" of
prefab literature. Relative breakdowns by notice of prefab
material of the preference for prefabricated houses or standard
EXHIBIT 8
WHAT PEOPLE THINK OF PREFAB MATERIAL
ARTICLE
AO
MFR'S
PROMOTIONAL
SEEN
PREFABS
NOT AFFECTED
FAVORABLE UNFAV.
10 20 30 40 50 70 80 9O tOO
PERCENT
268
MARKETING RESEARCH
I
o
1
o
CASES IN ANALYSIS AND INTERPRETATION 269
EXHIBIT 10
ADVANTAGES PER DISADVANTAGE AND PERCENTAGE PREFERENCES BY
EFFECT OF PREFAB LITERATURE
Notice and effect
Ratio of advan-
tages to
disadvantages
Per cent preferring
prefab at $8,000
No contact with literature .51 11.8
Not impressed either way .76 16.5
Favorably impressed 1.17 41.8
Unfavorably impressed .77 25.5
EXHIBIT 11
PREFERENCE FOR PREFABS AT $8,000 PRICE LEVEL
NOTICE
OF MATERIAL
NONE
ARTICLE
AD
mrs/sssss^
iiBM;*iii*S;i;i;!;S;i;S;i;S;S;S;!*S;!;S:S;S;!;S;!;S;S;!;!;i;!;S;
Mro. MOM Y//////////M, I jljij: j! jiji JM^
^MS:S:ii:?:i;i;t;i:i:i;S:i;!:o!?i?!S!i!>!!!ii!S!;!?s!B^^
3EEN PREFABS
10 20 30 40 $0 60 70 60 90 100
PERCENT
270
MARKETING RESEARCH
houses at the $8,000 price level are shown in the chart on page
269. A request for individuals to state their preferences be-
tween prefabricated housing and standard housing at a $12,000
price level resulted in almost 100 per cent preference for stand-
ard houses irrespective of notice and effect of prefab literature.
The last table in this series, Exhibit 12, presents information
on the notice and effect of prefab material on the main brands
of prefabricated homes in the area. For each brand, the table
indicates the number of respondents acquainted with it and the
percentage distribution of this number by brand preference.
There is a fair amount of overlapping between columns be-
cause many of the respondents were acquainted with more
than one brand.
EXHIBIT 12
PERCENTAGE OF PREFAB PREFERENCES STATED BY RESPONDENTS
ACQUAINTED WITH PREFAB BRANDS
Brand Preference
Brand Acquaintanceship
None
National
Gunnison
Lustron
Others
Not
Sure
None
100
10%
12
33
26
9
10
10%
4
38
24
8
16
10%
2
19
50
5
14
10%
2
21
26
31
10
11%
89
National
Gunnison
Lustron
Other
Not stated
Total
Total replies
100%
236
100%
82
100%
125
100%
123
100%
48
100%
9
In assessing the significance of Exhibit 12, it is pertinent to
note that the area sampled contained about five times as many
National homes as Gunnisons, and about five times as many
Gunnisons as Lustrons. Only a negligible number of each of
a variety of other brands of prefab houses existed in the area at
that time, although the total of these other brands was rea-
sonably large. All in all, it was estimated that prefabricated
housing constituted about 7 per cent of the permanent one- or
two-family housing units in the area.
CASES IN ANALYSIS AND INTERPRETATION 271
ACTION NEEDED
An analysis is required of these tables and charts to determine
the effectiveness of prefabricated advertising and other liter-
ature. Specifically:
1. How widespread is the notice of written and other material
dealing with prefabricated housing?
2. How effective is this material in altering the individual's
attitude in favor of prefabricated houses?
3. On the basis of the answers to the preceding two questions,
what recommendations would you be prepared to make to:
(a) increase the notice of prefabricated material, and (b) im-
prove its effectiveness?
CASE 5 WHEN SHOULD RETAIL STORES
BE OPEN?
BACKGROUND
A number of merchants in a city of approximately 50,000
population in Central Illinois asked the manager of the local
Chamber of Commerce to review the hours at which their
stores should be open. Discussion with the merchants did not
offer any solution to the problem because of the great variations
in opinions. Nevertheless, there appeared to be four major fac-
tors in the store hour problem that emerged from the discussion,
namely, competition, employees, efficiency, and customers. It
was realized that each of the factors had a number of ramifica-
tions and that agreement on the matter of store hours would
not be very simple to obtain. The consensus on each of the
four factors is outlined briefly below:
Competition. Rivalry between stores always is a leading con-
sideration, often the only one. Merchants who are open longer
or at different hours may attract more shoppers, and thus do a
greater volume of business than those who follow another
policy. The problem applies both to competing stores in the
same shopping center and to competition between various cen-
ters for local and immediate rural trade.
272 MARKETING RESEARCH
Employees. Retailing is only one of the many fields in which
competent workers may be employed, and if other employment
offers more desirable working hours, retailing may suffer by ob-
taining lower grades of personnel. Long and wearisome hours
of work also affect employee efficiency and morale, which in
turn affect cost, labor turnover, and customer satisfaction.
Efficiency. The profitability of a store to a great extent de-
pends upon the efficiency of the workers and the control of
expenses relative to sales. On the one hand, there is the matter
of utilizing the fixed investment in the retail plant and inven-
tory by maximizing the sales volume done with that investment.
Yet, this must be balanced against the higher variable expenses
incurred by being open longer hours. Also, if store hours are
longer than those worked by employees, staggering hours when
clerks are at work involves additional cost, which varies among
stores, being heaviest for the smaller stores, particularly those
with some specialization among their clerks.
Customers. The first three factors are all related to attract-
ing customers: longer hours enable more persons to enter stores,
competent and cheerful sales persons make stores more desir-
able places to shop, and efficient operation tends to lower prices.
Beyond these factors, there are those related to the consumers'
real needs and desires regarding store hours, which must be
determined before hours can properly be decided. The ulti-
mate goal of the merchant is customer satisfaction, and this
must be a dominant factor in his decision regarding store hours.
The type of shopping center obviously is an important con-
sideration in studying the customers' needs. Shopping centers
may be classified as: downtown districts in metropolitan areas;
smaller cities in which city trade dominates; smaller cities in
which country trade dominates; and villages whose stores deal
mainly in convenience goods for a relatively small community.
In reaching their store-hour decisions, merchants must bal-
ance relevant factors against one another. For instance, it may
be decided that the labor situation is so tight that employee
relations should have foremost consideration. On the other
hand, the district may be one with strong competition between
stores and this may be the deciding factor.
CASES IN ANALYSIS AND INTERPRETATION 273
If a short-run view prevails, it is unlikely that the decision
will be best in the long run. If merchants give competition the
main consideration and do not join their fellow retailers in
planning store hours, some will probably decide to adopt hours
longer or different from the others. Competitors are likely to
retaliate by extending their own hours. This results in raising
expenses and causing discontent among employees, and may be
of little benefit to anyone in the long run.
If, on the contrary, retailers take the long-run view, they will
give customer and employee interests the weight which is due
them and will study carefully all factors involved. They will
understand that lack of uniform hours is a nuisance from the
shoppers' viewpoint, and that it is not mere number of hours
but their suitability which brings maximum business. They
will recognize that store hours really are a community problem
to be worked out cooperatively, and not a private matter for
competitive snap judgment.
THE PROBLEM
The manager of the Chamber of Commerce discussed the
matter with the retail merchants and found that an impasse
had been reached on the problem concerning the hours at which
the retail stores should be open. The crux of the matter was
that all merchants agreed that a uniform policy should be
adopted, but no specific policy was favored by a substantial
majority in the absence of empirical evidence. It was agreed,
however, that the manager of the Chamber of Commerce
should see what data could be obtained on the subject and then
make some recommendation to the merchants, if possible, on
store hours, or, if not possible, on deriving whatever additional
information he might deem necessary for a decision to be made
on the subject.
DATA AVAILABLE
After investigating the subject, the manager unearthed the
data shown in the following six tables. No other data on the
subject could be found.
274 MARKETING RESEARCH
EXHIBIT 1
ORDINARY STORE HOURS IN SELECTED CITIES IN ILLINOIS
City, population group
Number of
stores which
observe
hours stated
Total
hours
open
Time of
opening
Time of
closing
Chicago metropolitan area:
Downtown Chicago
Most stores
8
9:45
5:45
Outlying centers (77) . . .
20
1
8
8*4
9:30
9:30
5:30
5:45
1
8V2
9:00
5:30
22
8V2
9:30
6:00
30
9
9:00
6:00
1
9
9:30
6:30
2
More
than 9
Suburban cities (12) .
5
4
8
8V4
9:30
9:00
5:30
5:30
3
9
8:30 or 9:00
5:30 or 6:00
Downstate cities and towns
(68):
Population more than
30 000 (13)
7
IVi
9:30
5:00
3
8
9:00
5:00
3
8
9:30
5:30
Population 10,000-
30,000 (20)
1
IVi
9:30
5:00
3
8
9:00
5:00
1
8
9:30
5:30
1
8*4
9:15
5:30
2
8V2
8:30
5:00
9
8V2
9:00
5:30
3
9
8:30
5:30
Population less than
10,000 (35)*
2
8
9:00
5:00
3
8Vi
9:00
5:30
5
9
8:00
5:00
6
9
8:30
5:30
1
9
9:00
6:00
12
9V2
8:00
5:30
1
9V2
8:30
6:00
1
10
7:00
5:00
4
10
8:00
6:00
Seven of these places had populations of 3,000 or less, and the smallest had
only 1,400 inhabitants.
CASES IN ANALYSIS AND INTERPRETATION 275
EXHIBIT 2
1
EVENING STORE OPENINGS IN SELECTED CITIES IN ILLINOIS
City, population group
Number
whose stores
are open
Number of
evenings
open
Days when stores
open in evening
Chicago metropolitan area:
Downtown Chicago
Most stores
1
Monday
Outlying centers ( 77 )
1
3
Tuesday, Thursday,
Saturday
45
2
Monday, Thursday
1
2
Tuesday, Thursday
17
2
Thursday, Saturday
2
1
Monday
6
1
Thursday
5
1
Saturday
Suburban cities (13)
8
2
Monday, Thursday
1
1
Thursday
2
1
Friday
2
o
Downstate cities and
towns (71) :
Population more than
30,000(13)
7
Monday
2
Friday
3
Saturday
1
Population 10,000-
30,000 (21)
4
Monday
1
Friday
13
Saturday
3
Population less than
10,000(37)
1
1
Friday
28
1
Saturday
8
276 MARKETING RESEARCH
EXHIBIT 3
MORNING CLOSING, COMBINED WITH EVENING OPENING,
SELECTED CITIES IN ILLINOIS
City, population group
Number
combining
evening
opening with
morning or
afternoon
closing
Number of
days plan is
practiced
Days when plan
is used
Chicago metropolitan area :
Downtown Chicago . .
Outlying centers
Most major
stores
1
1
2
Monday
Monday, Thursday
Suburban cities
1
20
2
2
1
2
Monday, Saturday
Monday
Monday, Thursday
Downstate cities and towns:
Population more than
30,000 . ..
1
2
7
1
1
1
Monday
Friday
Monday
Population 10,000-
30,000
2
2
1
1
Friday
Monday
Population less than
10 000
None
EXHIBIT 4
MORNING OR AFTERNOON CLOSING, NOT COMBINED WITH EVENING
OPENING, SELECTED CITIES IN ILLINOIS
Number
City, population group
closing
morning or
Day selected
Morning or
afternoon
afternoon
Chicago metropolitan area:
Outlying centers ....
2
Wednesday
Afternoon
Suburban cities
4
Wednesday
Afternoon
Downstate cities and towns:
Population more than 30,000 ....
1*
Wednesday
Afternoon
Population 10,000-30,000
2
Thursday
Morning
Population less than 10,000
1
Wednesday
Morning
1
Wednesday
Afternoon
9
Thursday
Morning
1 Not all stores close
CASES IN ANALYSIS AND INTERPRETATION
EXHIBIT 5
TOTAL STORE HOURS PER WEEK, SELECTED "CITIES IN ILLINOIS
277
Hours per
week
Chicago metropolitan area
Downstate cities, by popula-
tion group
Down-
town
Chicago
Out-
lying
centers
Subur-
ban
cities
More
than
30,000
10,000
to
30,000
Less
than
10,000
46
46V2-48
'i
3
14
7
21
23
7
1
'6
1
1
3
4
4
2
2
1
*2
1
4
2
5
1
'i
1
2
6
6
9
4
3
2
1
48V4-50
50V2-52
52^2-54
54V^-56
56V2-58
58V2-60
60V2-62 . ...
62W64
68 ....
EXHIBIT 6
TIME OF DAY WHEN CONSUMERS LAST SHOPPED, AS PERCENTAGE OF TOTAL
IN ECONOMIC GROUP, BLOOMINGTON-NORMAL AND DANVILLE, ILLINOIS
Economic
Group
Morning
Afternoon
Evening
Bloom-
ington-
Normal
Danville
Bloom-
ington-
Normal
Danville
Bloom-
ington-
Normal
Danville
A
60
36
31
30
27
34
41
22
40
58
60
57
73
66
56
78
6
9
12
3
B
C
D
ACTION NEEDED
On the basis of the information contained in these tables, what
should be the manager's recommendation to the local merchants?
Should he recommend particular store hours to the merchants? If
so, which hours, and why? If not, what additional data should he
recommend be obtained on the subject? Why should these data
be obtained, and how?
278 MARKETING RESEARCH
CASE 6 ANALYSIS OF PRICE AND
INCOME DATA
BACKGROUND
Brown Brothers was a long-established, exclusive men's ap-
parel store somewhat on the famous Brooks Bros, model. For
generations it had catered to a select clientele of professional
men, business owners, and business executives. The store en-
joyed great prestige in its area. It also had, management be-
lieved, an exceptionally loyal clientele.
In recent years, however, sales volume had been somewhat
disappointing and the profit margin had fallen. Through the
urging of a young executive, the store eventually hired a research
agency to examine its buying, pricing, and merchandising pol-
icies.
STORE POLICIES
The store owners and executives were confident that the
store could increase volume at any time by invading the mass
market for men's clothing. They had no intention, however,
of making any such radical departure from store traditions.
The settled policy was to handle only quality clothing and to
operate on a full-service, conservative basis. The store had
been highly successful over the years in cultivating a particularly
desirable segment of the market. It desired only to adapt its
policies as might be necessary to maintain its position in this
special market.
In characterizing this market, it was estimated that the bulk
of business for higher-price lines came from men with incomes
of $11,000 or more per year, and that the bulk of business for
the medium-price lines came from men with incomes of at least
$5,000 or $6,000. (The most recent income data at that time
revealed that 24 per cent of families in metropolitan areas of
the U. S. had incomes of $5,000 or more. Six per cent of spend-
ing units had incomes of $7,500 or more. ) Because of those in-
comes, it was assumed that relatively few customers were single
men and that most customers were older men.
CASES IN ANALYSIS AND INTERPRETATION 279
Despite the high incomes which most customers were be-
lieved to have, it was thought that the store was heavily depend-
ent on sales of medium-price lines. As defined by the store,
high-price lines for men's suits (excluding summer suits) began
at $95, medium-price lines began at $65, and low-price lines
began at $45. The division of dollar sales for men's suits was:
low-price, 20%; medium-price, 55%; high-price, 25%. In ad-
dition, there was a tailoring department offering a wide selec-
tion of materials for personally fitted clothes. Volume was
not large, however, and the store had placed only five adver-
tisements of tailormade clothes in the past year.
The advertising budget was distributed, with relatively minor
exceptions, in proportion to sales volume. The $65 line of
men's suits, for example, got the biggest appropriation. The
bulk of expenditures was in newspapers, although all types of
media, except magazines and transit ads, were used. It was
believed that few men really shop for clothes, their minds hav-
ing been made up by the time they enter a store. Advertising,
therefore, was held to be an important factor in forming cus-
tomer decisions. To this end, efforts were made to keep the
store "in the paper" almost daily, to make advertisements dis-
tinctive, to stress quality and style of merchandise, and to avoid
bargain appeals.
FINDINGS
A random sample of store customers was selected and per-
sonal interviews were conducted. Data from store records also
were gathered. Some of these data are contained in the exhibits
which follow. These exhibits are shown just as they came from
original tabulation without processing, and are therefore not
necessarily adapted to the best analysis possible.
ACTION NEEDED
1. How should these data be processed or manipulated for
analysis?
2. What other cross tabulations of these data should be made?
280 MARKETING RESEARCH
EXHIBIT 1
CUSTOMER INCOMES, BY MARITAL STATUS AND AGE
\/f_ 'l. rt l
Income
Marital
status
and age
Below
$7,000
$7,000
to
$10,999
Below
$11,000
(Sub-
total)
$11,000
and
over
Don't
know
Total
Single men
21 to 35
42
4
46
14
o
60
36 to 50
8
14
22
5
n
2
Over 50
10
18
28
24
4
56
Don't know . . .
3
3
2
5
Total
Married men
21 to 35
63
76
36
46
99
122
44
36
6
5
149
164
36 to 50
80
96
176
226
6
408
Over 50
Don't know . . .
72
1
84
1
156
2
270
12
28
4
454
18
Total ....
All men
21 to 35
229
118
227
50
456
168
544
50
44
6
1,044
224
36 to 50
88
110
198
232
6
436
Over 50
Don't know . . .
82
4
102
1
184
5
294
12
32
6
510
23
Total
292
263
555
588
50
1,193
NOTE: Further breakdowns were considered undesirable. Incomes had been
gathered by intervals of $1,000. For all customers the data were: Below $3,000,
24 men; then in sequence by 1,000's: 30; 66; 88; 84; 80; 62; 51; 70; 50; and
40; then $1 3,000 or more, 498; and Income unknown, 50. Less than 2 per cent
of the customers were under 21 years of age.
CASES IN ANALYSIS AND INTERPRETATION
EXHIBIT 2
PLACES AT WHICH CUSTOMERS MADE LAST SUIT PURCHASES,
BY INCOME
281
Income
Last purchase
made at
Below
$7,000
$7,000
to
$10,999
Below
$11,000
(Sub-
total)
$11,000
and
over
Don't
know
Total
Brown Bros
Tailor Shop
90
30
74
32
164
62
196
116
18
4
378
182
Other store
168
154
322
251
26
599
Don't know
4
3
7
25
2
34
Total . .
292
263
555
588
50
1,193
EXHIBIT 3
PLACES AT WHICH CUSTOMERS MADE WINTER SUIT PURCHASES
DURING PAST THREE YEARS, BY INCOME
Income
rurcnases
in past
3 years at
Below
$7,000
$7,000
to
$10,999
Below
$11,000
(Sub-
total)
$11,000
and
over
Don't
know
Total
Brown Bros
136
146
282
283
23
588
Store U
21
29
50
33
5
88
Store V . . .
Stores W, X, Y, Z
Dept. stores
74
21
26
64
30
23
138
51
49
138
49
14
11
4
287
100
67
1 other store
2 other stores ....
3 other stores ....
Tailors
83
16
1
39
102
11
36
185
27
1
75
183
42
6
145
16
2
6
384
71
7
226
Don't know
8
1
9
17
26
Total ....
425
442
867
910
67
1,844
NOTE: Because of purchases made by individuals at more than one store,
totals exceed the number in the sample.
282
MARKETING RESEARCH
EXHIBIT 4
PRICES PAID BY CUSTOMERS FOR ALL WINTER SUITS PURCHASED DURING
PAST THREE YEARS, BY INCOME
Prices
paid
Income
Below
$7,000
$7,000
to
$10,999
Below
$11,000
(Sub-
total)
$11,000
and
over
Don't
know
Total
Under $35
14
27
73
106
190
108
42
26
148
14
11
1
6
4
34
10
2
16
38
68
100
116
73
44
214
26
10
6
6
12
44
7
16
43
111
174
290
224
115
70
362
40
21
7
12
16
78
17
4
18
56
104
150
144
154
138
314
93
64
56
74
121
312
6
6
14
20
13
11
3
4
3
3
2
6
12
20
61
173
292
460
381
280
211
680
136
85
66
88
143
402
23
$35444
$45-$54
$55464
Low price . . .
$65-$74 . . ..
$75-$84
$85-$94
Medium price ....
$954104 ....
$105-$114 . ..
$1154124 ....
$1254134 ....
$135 or more . .
High price
Don't know ....
Total . .
382
365
747
782
36
1,565
NOTE: Because some individuals made purchases in a number of price ranges,
the subtotals are less than the sum of the persons buying each price line but
totals exceed the sample number for the same reason.
CASES IN ANALYSIS AND INTERPRETATION
EXHIBIT 5
283
DISTRIBUTION OF CUSTOMERS BY NUMBER AND TYPE OF WINTER SUITS
PURCHASED DURING PAST THREE YEARS, BY INCOME
Type of suits
bought and
number:
Income
Below
$7,000
$7,000
to
$10,999
Below
$11,000
(Sub-
total)
$11,000
and
over
Don't
know
Total
Ready-made only
1 suit
48
52
66
28
13
8
6 A
221
10
6
6
5
2
29
11
4
1
2
6
24
18
292
17
55
56
29
12
12
H B
195
1
4
7
1
1
1
2
17
6
13
7
6
10
8 H
50
1
263
65
107
122
57
25
20
20
416
11
10
13
6
1
3
2
46
17
17
8
8
10
14
74
19
555
4
57
94
62
39
32
30
318
8
18
25
24
13
19
12*
119
6
35
18
15
20
30 1
124
27
588
9
10
9
1
4
6 D
39
3
2
2
1
8
2
1
3
50
78
174
225
120
68
52
56
773
22
30
40
31
14
22
14
173
23
52
28
24
30
44
201
46
1,193
2 suits
3 suits
4 suits
5 suits
6 suits
7 or more . . .
Ready-made
total . . .
Custom only
1 suit
2 suits . . .
3 suits
4 suits
5 suits
6 suits
7 or more
Custom
total
Both types
2 suits
3 suits
4 suits
5 suits . .
6 suits
7 or more
Both types
total . .
N A.
Grand total
NOTE: Number of suits in open-end classes: A-46; B-118; C-268; D-54; E-16;
F-106; G-48; H-62, and 1-284. Breakdowns for purchases of both types of suits:
Below $7,00042 Ready-made and 48 Custom, total 90. $7,000-$ 10,999
126 Ready-made and 98 Custom, total 224. $11,000 and over 356 Ready-
made and 304 Custom, total 660.
284 MARKETING RESEARCH
3. In view of the problem of an unexpectedly large open-end
class, is the income breakdown used the most feasible?
4. How consistent are, and should be, the findings as to the
share of their customers' total suits purchases which Brown
Brothers secures as based on last purchase and 3-year-pur-
chase data?
5. What statistical tests of significance might be used on these
data? In what way and to what degree are the assumptions
made by management about the store's market and the
actions of customers proved or disproved?
6. What inferences may be drawn from these data regarding
the effectiveness of the firm's merchandising policies?
7. What recommendations might be based on these data?
Part VIM
PRESENTATION AND FOLLOW-UP
The Nature of the Tasks
The bitter truth is that a great deal of marketing research
results in little or no action. The reason is that between tech*
nical findings and business actions lies a chasm difficult to
bridge. How difficult is suggested by the fact that today many
research staffs and agencies employ specialists for writing re-
ports. But even an effective report, alone or with an oral pres-
entation, may fail unless presentation is followed up to insure
that action is taken or that failure to act is specifically justified.
The major task of presentation is to convey the findings (al-
most always statistical in nature and often complex) in the
simplest, shortest, most persuasive, and most practical manner
possible. The language of statistics and research has to be
translated into operable terms.
The task of the follow-up is to avoid losing the practical
advantages of research because of the pressure of day-to-day ac-
tivities of busy executives.
The researcher does not bear sole responsibility for insuring
that his studies become actions; but his is the principal task.
Working Methods
Presentation. In his interpretation of results the researcher
already has considered the problems of the company and the
personalities and functions of the executives. In preparing his
presentation he, as does every writer, gives special attention to
his "audience." The form of his report, as noted below, will
depend on this. More directly, he may consult with executives
as to findings and possible actions. By formulating possible
285
286 MARKETING RESEARCH
alternatives, he can get executives to express their viewpoints
as to usefulness of findings and the form in which they are most
valuable. This may secure a far more enthusiastic acceptance
of findings and recommendations. An alternative is to with-
hold all findings in order that the final report may be a dramatic
surprise.
Most commonly the research findings will be of interest to
both technicians and nontechnicians. The marketing research
specialist will want to know about sampling methods, the
questionnaire used, the supervision of the field staff, statistical
manipulations, and other technical details. Moreover, the
specialist may prefer to read his findings in technical language.
The technical report will be addressed to this group as a separate
report, or such materials will be included in an appendix to a
report.
Research intended to be read by company officials will con-
tain the fewest possible technical details. It will be designed
and written to arouse reader interest (without compromising
accuracy) by headline-style writing, by liberal use of themes,
catchwords, and analogies, and by much graphic illustration.
Such a report is called a popular report; and research reports in-
tended to be read by the general public are nearly always of this
simplified and dramatized type.
Compromise reports, however, are common. These are re-
ports which combine the elements of both technical and pop-
ular reports. For example, a summary and recommendation
may be popularized, a major findings section may be some-
what technical, and appendices may contain highly technical
details. Sometimes the research staff will present only general
findings and conclusions to company executives, further de-
tails and technical aspects being held ready should any execu-
tive request additional data or proof.
The translation of statistical data into the language of
business is a major challenge. The first and most basic step
usually is to present statistics in the form of charts and graphs
of various sorts. Map and illustrations (particularly picto-
graphs) also aid comprehension by nontechnicians, and add
life to a report.
PRESENTATION AND FOLLOW-UP 287
No mechanical devices or tricks, nonetheless, can substitute
for clear expression of ideas in words arid an appropriate sales
or business slant to conclusions. Here, near the end of re-
search, is when a carefully planned beginning pays dividends.
The most useful tool at this stage is an outline. It acts as a
guide to the writing and a test of its adequacy, and it offers the
readers an index to all materials.
Follow-Up. After the written report has been submitted,
often at the time of an oral presentation that highlights findings
and utilizes visual methods to a great extent, it is necessary to
follow up the report.
Most commonly this is the function of the research director.
He performs it often by talking informally with executives con-
cerned in the days and weeks following submission of the re-
port. He relies on his knowledge of the company and his close
friendship with executives to get action. Sometimes a special
assistant of the company president will be charged with respon-
sibility for securing adequate consideration of a report by exec-
utives. Occasionally, as in new product development, the
research staff may be given direct responsibility to put their
recommendations into action. A research agency sometimes
is asked, also, to put recommendations into action; but usually
reliance is put on effective presentation plus informal follow-up
procedures. A few companies set deadlines, dates on which
executives must act or give reasons for not acting.
Special Difficulties
The unsolved problem, both in theory and application, of
the proper borderline between research as a staff function and
the line functions of management creates many difficulties.
Not many years ago most marketing research was expected to
stop short at fact finding. Repeated failures of executives
(either because of unwillingness, inability, or lack of time) to
interpret findings and formulate business actions gradually has
extended the research function. Almost universally (for all
but the simplest of studies), researchers are now expected to
reach statistical decisions and to draw conclusions. In many,
if not most, cases they are expected to formulate specific recom-
288 MARKETING RESEARCH
mendations for action. (As has been noted, they even may be
asked to supervise the execution of their recommendations.)
It is clear that marketing research is a service or staff function.
Yet it is generally believed that the researcher capable of
formulating a business problem for study also is capable of
understanding operating problems. He should not order or
command; but he is capable of formulating realistic alterna-
tive lines of action and of recommending ideas and actions.
What should be done in an individual case depends on the
researcher's abilities, his relations with and understanding of
the company, and the executive personalities concerned.
Another difficulty is that the technical training needed and
the literary skill required may not be found in one man. A
report, therefore, may have to represent the combined efforts of
several persons.
A further difficulty is that researchers often fail to leave
enough time for this report-writing and presentation phase.
All too often careful fact finding is followed by hasty report
preparation to meet a deadline.
The Cases in this Part
Case 1, ''Preparation of Chart Material," raises the problem
of translating common forms of statistical data into more effec-
tive presentation forms.
Case 2, "Preparation of a Retail Trade Area Map," is a more
specialized problem which involves far-reaching conclusions
and preparing both a map and a brief report from the data
given.
Case 3, "Consumer Reaction to Self-Service Meat Market-
ing," poses the problem of rewriting certain introductory and
technical materials into a report form so that it will be suited
to other, nontechnical readers.
Case 4, "Report on Branch Store Markets and Policies," is
offered primarily for evaluation, the criticism being made of
many possible aspects of report writing. Opportunity also is
offered for consideration of graphic presentation and the prep
aration of an oral presentation.
PRESENTATION AND FOLLOW-UP 289
Case 5, "Follow-Up on Product Performance Studies/' re-
views the situation faced by a company's research staff in
securing action and poses problems of evaluation or perform-
ance for many aspects of follow-ups.
AMERICAN MARKETING ASSOCIATION. The Technique of Marketing Research.
New York: McGraw-Hill Book Co., Inc., 1937. Chapters 17 and 18 con-
tain a fairly conventional treatment of the presentation and follow-up on
research results.
BRADFORD, E. S. Marketing Research. New York: McGraw-Hill Book Co.,
Inc., 1951. There is an especially good section on graphic presentation in
Chapter 13.
BROWN, L. O. Marketing and Distribution Research. New York: The Ronald
Press Co., 1949. Chapter 25 gives an excellent discussion of the primary
principles of reporting on research work.
BUREAU OF THE CENSUS. Manual of Tabular Presentation. Washington, D.C.:
U. S. Government Printing Office, 1949. An outline of theory and practice
in the presentation of statistical data for publication in tables.
COUTANT, F. R. "Market Research Must Hold Executive Attention." Journal
of Marketing, X (January, 1946), pp. 288-89. The author shows the im-
portance of research to top management.
FROTHINGHAM, R. S., et al. "Preparation and Presentation of the Research
Report." Journal of Marketing, XIII (July, 1948), pp. 62-72. A detailed
treatment of the ways in which reports should be prepared for various audi-
ences.
HALL, R. O. Handbook of Tabular Presentation. New York: The Ronald
Press Co., 1946. This is considered the standard work related to the prepa-
ration of tables.
HEIDINGSFIELD, M. S., and A. B. BLANKENSHIP. Market and Marketing Analy-
sis. New York: Henry Holt & Co., 1947. The latter half of Chapter 11 is
devoted to a brief discussion of the form of the report, its presentation, and
the follow-up that is necessary.
LUCK, D. J., and H. G. WALES. Marketing Research. New York: Prentice-
Hall, Inc., 1952. Chapters 15 and 16 give details of the outline of report
writing as well as a variety of ways in which the findings can be put to work.
LORIE, J. H., and H. V. ROBERTS, Basic Methods of Marketing Research. New
York: McGraw-Hill Book Co., Inc., 1951. Chapter 19 shows how market-
ing research can be coordinated with the needs of management.
SAUNDERS, A. G., and C. R. ANDERSON. Business Reports. New York: Mc-
Graw-Hill Book Co., Inc., 1941. A good standard reference on the prepara-
tion of business reports of all types.
SESSIONS, R. E. "A Management Audit of Marketing Research." Journal of
Marketing, XIV (January, 1950), pp. 563-71. Discussion of the ways in
which marketing research keeps pace with company needs, the administra-
tion of research that can be made to set the pace for operating efficiency, and
the ways in which the research staff can keep in touch with the realm of
ideas which determine sales growth and future market position for the com-
pany.
Part VIII Cases
CASE I PREPARATION OF CHART MATERIAL
BACKGROUND
A large publishing house had collected some income and
population data through its Division of Marketing and Re-
search. These data are shown in table form in Exhibits 1-5.
EXHIBIT 1
INCOME DISTRIBUTION OF U. S. FAMILIES, 1939 AND 1947
Income group
January 1
,1939
January 1
,1947
Number
Per cent
Number
Per cent
A $5 000 & over
1 360 000
4.0
3 721 000
96
B. $3,000-$4,999 . . .
2,168,000
7.7
7*056*000
18.3
C. $2,000-$2,999
6,324,000
18.6
11,815,000
30.6
D. $1,000-$1,999
12,410,000
36.5
10,904000
28.2
E. Under $1,000
11,288,000
33.2
5,113,000
13.3
Total
34,000,000*
100.0
38,609,000*
100.0
Inc.
* U. S. Bureau of the Census estimates
Source: Division of Marketing and Research of Macfadden Publications,
EXHIBIT 2
INCOME DISTRIBUTION OF URBAN FAMILIES, BY OCCUPATION, 1936 AND 1946
(In percentages)
Income
1936
1946
Wage earner
White collar
Wage earner
White collar
A. $5,000 & over.
B. $3,000-$4,999
C. $2,000-$2,999
D. $1,000-$1,999 .
E. Under $1,000
0.1
1.6
6.2
24.4
20.5
3.6
5.9
11.1
18.8
7.8
1.5
10.8
22.1
16.1
7.9
8.4
8.5
11.5
10.3
2.9
Source: National Survey of Liquid Assets, Bureau of Agricultural Economics.
Division of Marketing and Research of Macfadden Publications, Inc.
290
CASES IN PRESENTATION AND FOLLOW-UP 291
EXHIBIT 3
INCOME DISTRIBUTION OF U. S. FAMILIES, FOR SELECTED YEARS
(Total families = 100% )
Year
Upper
(over
$5,000)
Lower
(under
$2,000)
Middle
($2,000-
$4,999)
1896...
1.9%
96.3%
1.8%
1910...
1.7
92.1
6.2
1 out of 10
families had more than
$40 a week
1921...
2.5
86.4
11.1
1929 ..
8.2
59.5
32.3
4 out of 10
families had more than
$40 a week
1933 ..
3.3
82.0
14.7
2 out of 10
families had more than
$40 a week
1/1/39
4.0
69.7
26.3
1/1/40
4.1
68.8
27.1
1/1/41
4.4
66.9
28.7
1/1/42
5.7
58.5
35.8
1/1/43
7.6
48.7
43.7
1/1/44
8.3
43.6
48.1
1/1/45
8.8
42.7
48.5
1/1/46.
8.9
42.7
48.4
1/1/47
9.6
41.5
48.9
6 out of 10
families had more than
$40 a week
Sources: Charles Spahr, The Present Distribution of Wealth; W. I. King,
The Wealth and Income of the People of the United States; National Industrial
Conference Board, unpublished studies; Brookings Institution, America's Ca-
pacity to Consume; Financial Survey of Urban Housing; National Resources
Planning Committee; Macfadden Publications, Inc., Division of Marketing and
Research.
THE PROBLEM
The Marketing and Research Division of the publishing
house wanted to prepare a popular report that would include
tables, charts, and graphs. The objective was to keep the writ-
ten copy to a bare minimum and keep reader interest high by
using a number of unique and pleasing charts and graphs.
ACTION NEEDED
From the tables given in this Case, select those suitable for
preparing charts and graphs. Indicate your reasons for this
selection. Then prepare the charts and graphs, and write out
a statement giving the reasons for your choice of the various
types of charts and graphs you have prepared.
292 MARKETING RESEARCH
EXHIBIT 4
NATIONAL INCOME PAYMENTS, BY DISTRIBUTIVE SHARES, FOR
SELECTED YEARS
(Dollar amounts in billions)
Year
Total
income
payments
Salaries
and wages
Entrepre-
neurial
income
Dividends,
interest, and
net rents
Other
income
payments
1910
$ 31.1
$ 148
*
$15 2
#
1929
100.0%
$ 82.6
47.7%
$ 52 5
$136
52.1%
$15 3
$1 2
1939
100.0%
$ 70.8
63.5%
$ 42 2
16.5%
$11 2
18.5%
$11 2
1.5%
$62
1940 . ...
100.0%
$ 76 3
59.6%
$ 46 5
15.8%
$12
15.8%
$11 5
8.8%
$6 3
1941
100.0%
$ 92.8
60.9%
$ 58.2
15.7%
$15 8
15.1%
$12 5
8.3%
$62
1942
100.0%
$117.3
62.7%
$ 77.6
17.0%
$206
13.5%
$13 1
6.7%
$60
1943
100.0%
$143.1
66.1%
$ 99.2
17.6%
$23 5
11.2%
$140
5.1%
$64
1944
100.0%
$1566'
69.3%
$1089
16.4%
$24 1
9.8%
$15 1
4.5%
$8 5
1945
100.0%
$160.8
69.6%
$110.2
15.4%
$256
9.6%
$16 3
5.4%
$87
1946
100.0%
$1650
68.6%
$1052
15.9%
$302
10.1%
$18 1
5.4%
$11 5
100.0%
63.7%
18.3%
11.0%
7-0%
* Information not available in comparable form.
Sources: U. S. Department of Commerce; W. I. King, The Wealth and In-
come of the People of the United States.
EXHIBIT 5
CIVILIAN LABOR FORCE, 1940-47
Year
Civilian
labor force
(im'llions)
Number
employed
(millions)
Per cent of
labor force
employed
March, 1940
530
45 1
85 1%
March, 1941
52.0
460
88 5
March, 1942
53.5
502
93 8
March, 1943
52.3
51 2
979
March, 1944
51.4
50 5
982
March, 1945
55.7
53.0
95.2
March, 1946
55.2
52.5
95.1
March, 1947
58.4
56.1
96.1
Source: Bureau of the Census, U. S, Dept. of Commerce.
CASES IN PRESENTATION AND FOLLOW-UP 293
CASE 2 PREPARATION OF A RETAIL TRADE
AREA MAP
BACKGROUND
For many years researchers in the retail field have attempted
to develop formulas, or 'laws/' to account for the division of
trade between market areas. The first "law of retail gravita-
tion" was proposed by William J. Reilly in 1931 and purported
to divide trade from an intermediate town between two neigh-
boring cities according to the following formula: 1
B
where
B is the proportion of trade gained by city A from the intermedi-
ate town,
B 6 is the proportion gained by city B ;
P is the population of city A,
P & is the population of city B,
D ft is the distance between city B and the intermediate town, and
D a is the distance between city A and the intermediate town.
The boundary of a city's trade area in relation to another city
was determined from the following formula, which was devel-
oped at a later date:
Miles between cities A and B
Boundary for city B in miles = , :=: ,
where B is the smaller city.
*W. J. Reilly. The Law of Retail Gravitation. (New York: William J.
Reilly, 1931).
294 MARKETING RESEARCH
A formula for estimating the loss of trade of one town to an-
other was developed by P. D. Converse in 1949, as follows: 2
where
B tt is the proportion of trade gained by town A, the outside town,
B & is the proportion of trade retained by town B, the home town,
P a and ?& are as before, and
d= distance between towns A and B.
The integer, 4, in the above formula is denoted by Converse
as the "inertia-distance factor of trading at home" and was
derived on the basis of a series of studies of towns in Illinois with
a population over one thousand.
It should be noted that all these "laws" were formulated pri-
marily with fashion, or shopping, goods in mind; that is, apparel,
household furniture and furnishings, and similar items.
THE PROBLEM
In 1951 the merchants of Fairville, a large town in the Middle
West, called in a researcher to aid them in analyzing their trade
position in relation to other towns in the area. What the
merchants wanted were some benchmarks against which they
could evaluate the effectiveness of their merchandising meth-
ods. They planned to evaluate these methods by means of a
series of retail trade surveys. In effect, they wanted a map
delineating the "normal" boundaries of their trade area and
giving some idea of the expected division of trade between them
and each of the neighboring towns.
DATA AVAILABLE
A map of the towns in the vicinity of Fairville is shown on
page 295, together with the distances in road miles between
8 P. D. Converse. "New Laws of Retail Gravitation/' Journal of Marketing,
XIV (October, 1949), pp. 379-84.
CASES IN PRESENTATION AND FOLLOW-UP 295
.HIGH CITr
VICTORIA
SPRINGTON
CLAYTON
RICHLANO
neighboring towns. The population of each of these towns in
1950 was as follows:
Fairville
Richland
Clayton
17,500
6,900
9,200
Springton
High City
Victoria
3,000
64,000
6,200
The materials by Reilly and Converse cited above are highly
pertinent to this problem.
ACTION NEEDED
Three things are needed:
1. A retail trade area map for Fairville outlining the boundaries
of its trade area as based on the formulas of retail gravitation.
296 MARKETING RESEARCH
2. Estimates of the division of trade between Fairville and each
of its neighboring towns, other things being equal.
3. A brief memorandum to the merchants of Fairville outlin-
ing the uses and limitations of these data and of the retail
trade area map.
CASE 3 CONSUMER REACTION TO
SELF-SERVICE MEAT MARKETING 1
BACKGROUND
The following are excerpts from a study that was made to
ascertain the reaction of consumers to self-service meat market-
ing. The sections of this study that have been reproduced here
are taken from a thesis entitled "Economic Analysis of Self-
Service Meat Marketing/' by William R. Bennett.
HISTORY OF SELF-SERVICE RETAILING OF MEAT
The first attempt to prepackage meat seems to have been that of
the Hudson's Bay Company in Winnipeg, Manitoba, Canada, in 1923. 2
Meats were wrapped in cellophane and sold by service clerks. Five years
later Mr. Donaldson, who had conducted the above operation, con-
vinced Mr. Frank Parsloe of the H. C. Bohack Company, Brooklyn, New
York, of the feasibility of the plan, and by 1929 the Bohack Company
was serving prewrapped meats to fifty stores from a central plant. The
project was abandoned because of improper refrigeration and display
equipment, lack of meat knowledge by clerks, inadequate wrapping
materials, and a high percentage of returns to the central plant. 3
Around 1925 the first transparent flexible film package of meat made
its appearance. It was not suitable for self-service selling because it was
not moistureproof and because of the high humidity of self-service cases.
The wrap did give packer identification, however. 4
In the early thirties, the Atlantic and Pacific Tea Company conducted
1 Based upon a study prepared by W. R. Bennett as a doctoral dissertation,
on file at the University of Illinois library.
* Meat Merchandising Self-Service Meat Manual. St. Louis: Meat Mer-
chandising, Inc., 1949, p. 9.
* Ibid., p. 10.
*N. Allen. "The Latest 'Know-How* in Prepackaging of Meats," Super
Market Merchandising, XV (June, 1950), pp. 127-31.
CASES IN PRESENTATION AND FOLLOW-UP 297
experiments with prepackaging but discontinued them because of the
above difficulties. 5
The Hy-Grade Food Products Corporation conducted experiments in
1933 by putting precut meat into cardboard bread trays and overwrap-
ping them with cellophane on a bread-wrapping machine. The project
was also abandoned because of technical and managerial difficulties. 8
In 1935 the Loblaw chain in Canada tried a fixture with a display top
and stock drawers below, but the refrigeration was inadequate. 7 By
1938 the Sanitary Grocery Company of Washington, D.C., had devel-
oped a rolled roast wrapped in cellophane and the idea was copied by
other chains and independents. Legs of lamb were next wrapped in
cellophane and placed on top of the meat cases. In 1940, chickens were
prepacked and sold from a service case. 8
The first self-service case (an old fish and delicatessen case) went into
operation at the A & P store at 467 Center Street, Jamaica Plains, Massa-
chusetts, on February 4, 1941, and 1,400 packages were sold from the
case in the first week of operation. A 30 per cent increase in volume
was experienced without additional labor cost. 9 In 1941 Empire
Markets opened a self-service meat department in Schenectady, New
York; in 1942 Caler's opened a market in Los Angeles, and by the end
of 1944 there were ten 100 per cent self-service meat departments in
operation in the United States. 10 By April 1 of 1949, there were in the
United States 878 self-service meat departments in operation. The
A. C. Nielsen Company reported 7,754 stores on a partial self-service
basis as of the same date. 11
It should be mentioned here that before self-service meat marketing
could develop, two technological improvements had to be made, inno-
vations themselves: the development of a satisfactory wrapping material
for fresh meat, and the development of a satisfactory meat case for self-
service. It is interesting to note that refrigeration men did not believe
that such an improvement was possible at the time they were assigned
the problem. 12
5 Meat Merchandising Self -Service Meat Manual, p. 10.
9 Ibid.
7 Ibid.
8 Ibid, p. 11.
9 Ibid.
11 Prepackaged Self -Service Meats 1949 Report. Chicago: Armour & Co.,
1949, p. 3.
u Meat Merchandising Self -Service Meat Manual, p. 11.
298 MARKETING RESEARCH
THE CHAMPAIGN-URBANA SURVEY
Limitations of Surveys and Reasons for Present Study
Analysis of the degree of difference in attributes of households ac-
cepting the innovation from those not accepting it is not found in the
above studies. 13 Knowledge of attributes associated with acceptance
(and rejection) of the innovation should be useful to retailers who are
faced with the problem of whether or not to convert to self-service, to
students of marketing who are interested in the study of innovations by
marketing institutions, and to students of economics interested in the
theory of innovation.
The above surveys are concerned primarily with whether or not con-
sumers approve the innovation, and what they like or dislike about it.
Inferences regarding the population cannot be made on a probability
setting on the basis of any of the above surveys because the universes
were not defined and the samples were not random.
A number of possibilities were open to the author in a study of con-
sumer reaction to the innovation. One approach would have been to
study households before and after the installation of the innovation in
the community to detect changes in buying patterns and consumption.
The investigator would have been compelled to find a community in
which a meat department was being converted to self-service, survey a
random sample in the store's trade area several weeks before the intro-
duction of the innovation, several weeks after conversion, and again
several months later without respondents' knowing they would be sur-
veyed again. Unless one could find a neighborhood store being con-
verted, the bounding of the trade area might be a major problem. For
example, one large food store in Champaign-Urbana has a very large
country trade and comparatively little town business. Such a study
would probably necessitate extensive recording of meat purchases by
households and would extend over a rather long time. Because of these
problems, the approach did not appear to be feasible. A study of at-
tributes required the completion of only one questionnaire, did not
require either memory or recording of meat purchases, and could be
done with the resources and facilities at the disposal of the author.
18 For purposes of this study, a household is assumed to have accepted the in-
novation if half or more of the meat purchased by the household is from a self-
service meat department. A household is a spending unit (for purposes of food
buying) that has cooking facilities. Households with over three boarders were
not included in the sample.
CASES IN PRESENTATION AND FOLLOW-UP 299
Definitions
For purposes of clarity it is necessary to define terms that will be used
throughout the account of the results of the Champaign-Urbana meat
survey.
A self-service meat department is one in which all meat is cut,
wrapped, heat-sealed, weighed, labeled, priced, and sold from an open
display case. The term "self-service store" is used interchangeably with
"self-service department."
A service or conventional meat department is one in which each cus-
tomer is served by a clerk who weighs, prices, and wraps each order.
A partial self-service meat department is one in which fresh meats are
sold by the conventional method, other meats being sold self-service.
The partial self-service meat department is grouped with the conven-
tional departments for purposes of this study, since the main interest of
the study is in the sale of fresh meats self-service.
The term town or city refers to the area within the city limits of both
Champaign and Urbana.
A call was a visit to a household.
An interview was a conversation with a representative of a household
resulting in the completion of Question 1 on the schedule.
A nonusable schedule was a schedule discarded because of incon-
sistencies.
A refusal was a call by an interviewer resulting in the interviewee
refusing to answer Question 1 on the schedule.
A final refusal was a call by the author resulting in the interviewee's
refusal to answer Question 1 on the schedule.
A callback was the term used to designate an interview made on the
third or subsequent call.
An at home was the term used to designate an interview made on one
of the first two calls.
Fresh meat refers to fresh beef, pork, lamb, and veaL Cured or
processed meats are not included.
Most means over 50 per cent.
Community Chosen for Survey
The community chosen for the survey, Champaign-Urbana, Illinois,
twin cities of 62,404 population (1950), 14 is located in east central
Illinois and is the home of the University of Illinois. Five self-service
meat departments, operated by two chains, are located in the town, but
u Champaign-Urbana News-Gazette, June 15, 1950, p. 3.
300 MARKETING RESEARCH
they are grouped in such a way that most of the city (74 per cent of the
total number of blocks) is more than five blocks from a self-service store,
making it possible to study the effect of distance from the self-service
meat department upon acceptance of the innovation. A good distri-
bution of households among geographic areas was obtained through
random selection from the population.
Since the stores are somewhat different in layout and location, one
can study criticisms and good points of the innovation mentioned by the
customers of the stores to determine if these differences influence con-
sumer attitudes. Conventional meat departments of comparable size
and handling meats of comparable quality are located within three
blocks of three of the stores and small neighborhood stores are near the
other two. Champaign-Urbana was also desirable because of the use of
student interviewers.
Four of the self-service departments in the town were converted in a
six-month period in 1949. This limits the study of the effect of time on
acceptance of the innovation. Dates of conversion of the Champaign-
Urbana self-service meat departments are shown in Exhibit 1.
EXHIBIT 1
DATES OF CONVERSION OF SELF-SERVICE MEAT DEPARTMENTS,
CHAMPAIGN-URBANA
Store
Location
Date of Conversion
A
Champaign
May, 1949
B .
Champaign
December 1947
C
Champaign
October, 1949
D
Urbana
October, 1949
E
Urbana
November, 1949
Techniques Used
In order to be able to state conclusions in probability terms, a random
sample of 447 households in Champaign-Urbana was interviewed. The
sample was designed so that the standard error would not exceed 2.5
per cent. A detailed account of the construction of the questionnaire,
the use of a pilot survey, the design and selection of the sample, the
reliability of results, the treatment of nonresponses, the time and dura-
tion of the survey, the techniques of interviewing, and the handling of
tabulation is included as a part of the Appendix to the original study.
CASES IN PRESENTATION AND FOLLOW-UP 301
ACTION TO BE TAKEN
1. How would such materials be presented in an introduction
to an article for The Journal of Marketing? Write this in-
troduction.
2. How would such materials be incorporated into a report
designed for business executives of a chain store organiza-
tion?
3. How would such materials be handled in a report intended
for the general public?
CASE 4 REPORT ON BRANCH STORE MARKETS
AND POLICIES
BACKGROUND
The findings of a study of branch store markets seemed most
important in terms of the basic policies pursued by the depart-
ment store for operating its branches. The basic theme of the
report, therefore, was the adaptation of these policies to the
number, types, and behavior of the people who comprised
the markets for the branches.
Consultations with the director of the store's research staff
during the drafting of the report encouraged the research group
to believe that even a lengthy report would get a full reading.
It was decided, however, to withhold findings until an oral pres-
entation could be made. This would be done at a meeting of
the operating committee (including all top executives of the
store), which the president and the director of research also
would attend. At this meeting, executives would be given a
copy of the written report.
EXCERPTS FROM A FINAL DRAFT OF THE WRITTEN REPORT
Shown below are excerpts from the final draft: the table of
contents (excluding Tables 16 to 37), purposes, summary and
recommendations (excluding the sections on lower-price lines,
special study for Branch B, and advertising policy), and a sam-
ple page from the major findings and from the statistical tables
sections.
302 MARKETING RESEARCH
EXHIBIT 1
TABLE OF CONTENTS
PAGES
I Purposes 3
II Summary and Recommendations 5
The basic policy of uniformity ..... 5
Exception to basic policy branch sales events . . 6
Exception to basic policy lower price lines . . .6
Special study of the market for Branch B . .7
Advertising policy .8
Future branch policy .9
Use of credit records for branch store research . . .10
III Major Findings - 11
How many customers? .11
How "independent" or "interdependent" are the markets
of branches and the main store? 12
Do branches create plus business? 13
How well are branch store price-lines adapted to their
markets? H
Does the lack of special branch sales fit the market? . .15
What are the trading areas of the branches and how great
is their attraction power in these areas? . . .17
IV Appendix 20
A. The Methods Used 20
B. The Sample 22
C. The Questionnaire Used 23
D. Statistical Tables 24
( 1 ) Division of customer business . . . .24
(2) Proportion of branch customers patronizing the
main store 24
(3) Number of store credit customers patronizing
branches for credit or cash . . . .25
( 4 ) Family income of customers by stores patronized 2 5
(5) Size of customer families by stores patronized 26
(6) Proportion of total branch business by size of
branch accounts 26
(7) Size of accounts by stores patronized . . 26
(8) Division of expenditures of branch patrons by
stores patronized 27
CASES IN PRESENTATION AND FOLLOW-UP 303
(9) Division of expenditures by all store patrons re-
siding in branch areas . '. . . .27
(10) Customer age groups by stores patronized . 28
(11) Families that include preschool children, by
stores patronized . . . . . .28
(12) Total size of credit accounts (branch plus main
store) by income groups . . . .28
(13) Size of basement store credit account by stores
patronized .29
(14) Basement store accounts by income groups . 29
(15) Customers desiring basement store merchandise
in branches by stores patronized . . .30
I PURPOSES
When first discussed, in September, the general purpose of this study
was "to analyze such customer actions and purchase patterns as are re-
yealable in our credit records in order to secure information of value in
merchandising and promotional activities." (The possibility of extend-
ing any study by a survey also was noted.) Six specific store problems
ivere listed under this general purpose.
A statement of proposals was submitted to each member of the oper-
iting committee. Each voted that Problem 6 was "both interesting and
valuable." The wording voted on was:
Geographical Distribution of Business. To analyze customer patron-
age at the branches and Philadelphia store to discover interrelation-
ships of business by types of customers on purchases. Objectives: use
in credit solicitation, advertising coverage in branches, special promo-
tional or merchandising problems.
The discussions held later with store executives, plus published data
ind interviews conducted with housewives who patronized branch
tores, revealed these problems:
. Almost no factual data on branch customers exist. It is not even
possible to determine their numbers without a laborious search of
microfilm records over many months. (This not only greatly in-
creased the difficulties of sampling the records but also forced the
gathering of much background data before specific issues even could
be isolated.)
. A number of branch policies depend primarily on internal operating
economies, but depend also on important, untested assumptions
304 MARKETING RESEARCH
about customers and their actions. For example (the characteriza-
tion here is highly simplified) :
A. Centralization of buying; of receiving, checking, and marketing;
of promotion; of credit operations, and so forth, provides large
operating economies. The resulting uniformity of branch and
parent offerings, it is assumed, is not an important restriction on
sales because: (1) the markets are homogeneous, (2) most
branch patrons are also patrons of the parent store, ( 3 ) customers
expect branches to carry "everything," even if in limited assort-
ments, and (4) for furs, furniture, and other shopping lines most
branch patrons will come to the parent store in the downtown
shopping center.
B. An outstanding exception to uniformity, the absence of basement
operations in the branches, is necessary because of lack of space
and because (assumption) the higher-income suburban patrons
have little interest in these price lines.
C. Another exception to uniform offerings, the almost total lack of
special store-wide promotions in the branches, is necessary be-
cause of the high cost of processing small lots and transporting
sale merchandise to and from the branches. It is justified also
(assumption) because most branch patrons are less salesminded
and because traffic at the parent store is more necessary and more
difficult to develop.
D. Newspaper advertising by the present store (with by-lines and
other devices to show availability of merchandise at the branches)
also promotes branches. Little individualized branch advertising
in branch areas is needed.
The final purposes, therefore, were established as being:
1 . To gather facts on the shopping actions of branch patrons relevant to
the problems and assumptions noted.
2. To assess branch operations in an effort to uncover opportunities for
increased volume (without losing operating economies) through bet-
ter adaptation to market realities.
3. To provide a firmer basis for planning future branch store expansions.
4. To test the feasibility of using credit records primarily, or even ex-
clusively, for research of branch problems.
Special Note: No attempt has been made to analyze transaction or
sales data as usually gathered, whether by departments or otherwise.
Instead, an effort has been made to bring new facts to light. The
temptation to pour the new wine of facts about customer actions into
the old bottles of sales analysis has been resisted. The primary value of
CASES IN PRESENTATION AND FOLLOW-UP 305
customer data is the new insights provided. Although costs of operation
and sales volume by departments and lines have not been ignored, the
emphasis of this study lies on people and their shopping actions.
II SUMMARY AND RECOMMENDATIONS
That central city stores, although increasing in dollar sales, are getting
a smaller and smaller share of total business has been established many
times. Indeed, this fact has been a strong impetus to the branch store
movement. It is highly probable that this relative decline of downtown
department stores will continue.
In short, the great bulk of future increases in sales must come from
branch stores.
Markets also are people. The relation of this simple fact to branch
store operations may have been neglected. After all, little factual in-
formation on suburban purchasers as people has been available and
relatively little attention is given to it.
Store records, for example, indicate that the people who patronize the
two branches studied are responsible for about 35 per cent of total busi-
ness. With the addition of a third branch the proportion certainly will
rise to about 50 per cent. Not at the branches. No. But markets are
people.
The basic underlying trends are for these people to prefer the suburbs
over downtown Philadelphia for shopping and to increase their suburban
purchases. (They already spend more in the branches than they do in
the main store downtown.) Thus, these people represent exceptional
opportunities for increased branch business and for cultivation of their
business through their relations with the branches. The implications of
markets as people, in fact, run far ahead of current dollar sales figures.
The Basic Policy of Uniformity
The findings of this study fully support the basic policy of branch
operations.
Not that branch patrons are entirely like nonb ranch patrons. Op-
portunities to differentiate, or individualize, children's wear, sportswear,
Dr other branch departments, or to sell proportionately more of the same
joods in some price lines and categories unquestionably exist. And such
problems as parking are far different. Fundamentally, however, the
narkets are highly similar.
In fact, the great majority of all branch patrons also are patrons of the
nain store (and patrons in large part for the same things offered them in
306 MARKETING RESEARCH
branches) . Based on the facts of the market found, there seems every
reason to continue the policy of making branches duplicates, or counter-
parts, in all essential ways of the main store.
The only danger suggested by this policy investigation is that the main
store may dominate policies sometimes for its own good but not for the
good of the entire operation. Adjustments of operations should be ad-
justments to the total market. Uniformity which lacks this emphasis
may result in failure to realize branch potentials.
Exception to Basic Policy Branch Sales Events
In fact, the danger just noted seems present now in relation to branch
sales events. Present policy may be holding present costs in line, and
protecting main store traffic somewhat, but it is at the cost of neglecting
some genuine opportunities to increase branch business. The assump-
tion now made that branch patrons (and potential branch patrons) are
not as interested in sales as other people is erroneous. The self-
generation of branch traffic because of convenience also can be doubted.
The majority of branch customers visit the branches less than once a
month.
Store records show that, on a credit basis, branch patrons attend as
many main store sales events as do other customers. They also report
attending sales at other downtown department stores, though propor-
tionately less often than nonbranch customers.
Approximately two thirds of branch patrons report that they do attend
sales events at suburban stores. By their answers the great majority of
branch customers and potential branch customers desire branch sales
events.
During one event, lasting fifteen days, the one time when customers
had a choice of branch or downtown sales, they showed a great prefer-
ence for the branch. Store records show that during this period 81 per
cent of Branch A patrons patronized (for credit) only the branch, and
83 per cent made more than half their dollar expenditures there. This
factual test clearly indicates high potential of branch sales.
Nor is this all. The evidence indicates that not only do branches de-
velop plus business but also extra plus business, some of it from regular
credit customers, but much of it from otherwise inactive credit cus-
tomers or from people without credit accounts. (Twenty-seven per cent
of persons whom the records show made no purchases report attendance
at sales events.)
Moreover, expecting customers to come to the main store when
the branches do not have sales is risky. Although no absolute proof is
CASES IN PRESENTATION AND FOLLOW-UP 307
possible with the data gathered, there are many indications that com-
peting downtown stores siphon off much of this business. (For ex-
ample, as many patrons of Branch A report attending sales events during
1952 at sales events of competing department stores downtown as report
having gone to the main store sales events. The same is true for patrons
of Branch B.)
We conclude from the market data that branch patrons, and potential
branch patrons, very definitely desire branch sales events.
We recommend that more special sales events for branches be
planned. Despite the handicap of added expense, it is probable that
such events can be made profitable. In addition, such sales should do
much to increase the number of patrons and the amount of patronage
during regular selling days. Traffic for branch sales events also is the
easiest to develop of all store traffic.
Future Branch Policy
Branch A, in particular, has won for itself a considerable measure of
"market independence." Nineteen per cent of its customers do not
patronize (for credit) the parent store. It gets 61 per cent of the busi-
ness it divides with the parent. It gets more than half of all store busi-
ness in its area. Its customers as a group certainly do not patronize it
merely for convenience goods or fill-in items. And if the mail, tele-
phone, and basement sales of the parent are eliminated, it "competes"
: ar more strongly with the downtown store than even these figures indi-
cate. Moreover, much branch business is plus business volume that
.he parent store itself could never secure.
These facts indicate the desirability of a policy sufficiently flexible
o allow a branch to differentiate its offerings to realize sales opportuni-
ies in its "own" market. These facts also suggest the merit of a policy
yhich would provide for a considerable amount of separate branch pro-
notion, a direct cultivation of what is in one true sense its own, separate
narket.
The fact remains, nevertheless, that branch and parent customer
roups are interlocked or overlapping. Exclusive customers of either
re not as valuable as "mixed" patrons. The success of the enterprise
ssts on a successful joint cultivation and servicing of customers. To be
harply remembered, however, is the fact that branches generate most
f their own business and do not steal it from their parent. Market facts
Iso prove that you cannot successfully play battledore and shuttlecock
rith customer habits. If not offered certain merchandise or services in
ne location, customers will not flock automatically to the other.
308 MARKETING RESEARCH
These facts and considerations suggest that a basic policy of "chain
store" or centralized operations is essential. Acute as some organiza-
tional problems may become in the future, branch autonomy would
seem a poor remedy indeed. The danger for policy probably lies in the
dominance of branch activities by executives most of whom are also and
predominantly concerned with the welfare of the parent operations.
The lack of basement operations, or lower-price lines, in the two
branches studied apparently is causing some loss of potential volume.
Some remedy, we believe, is needed for Branch B. The market facts
gathered, however, may be most important in planning future expansion.
Both branch areas, and especially that of Branch A, are unusual. We
are convinced that in many areas in which the store might open a branch,
genuine opportunities for true basement operations exist. The opening
soon of a competitor's branch with a basement should be watched
closely. The expansion of suburban areas and the dilution of incomes
in them makes less and less the opportunity of "aristocratic" branches
to realize true market potentials.
Use of Credit Records for Branch Store Research
The facts about customer actions which are buried in the credit rec-
ords of the store are such as big manufacturers pay many thousands of
dollars for each year through consumer panels, store audits, and other
research. That these data can be valuable in assessing branch opera-
tions and policy is, we trust, demonstrated by the small example of re-
search possibilities covered by this study.
We recommend establishing a continuous branch study based on such
credit records. We do not recommend, however, that the methods used
here be repeated.
First, because it was impossible to identify branch patrons, our sample
included much deadwood and included far too many nonbranch patrons.
Samples built up from branch sales slips or current billings would result
in a great reduction in these wastes in time and money.
Second, the search of microfilms for data is not recommended. This
digging backward into the past is exceedingly laborious, time consuming,
and therefore relatively costly. Once a sample has been chosen
(with means of keeping it up to date), methods can be established for
pulling out data at little expense during the regular processing. This has
proved feasible under regular billing methods and, undoubtedly, could
be adapted to cycle-billing methods.
Third, the objectives of the research done should be pinpointed more
than has been possible in this exploratory study. The executives who
CASES IN PRESENTATION AND FOLLOW-UP 309
voted that the problem of "geographical distribution of customers" was
"both valuable and interesting" found it difficult, understandably, to
cite specific facts wanted or specific decisions which depended on spe-
cific market facts. Considerably more attention to the "working as-
sumptions" underlying branch operations and policies and their depend-
ence on market situations would greatly increase the value of the research
done.
These problems being solved, we would heartily recommend consider-
ation of establishing a continuing "panel" based on the regular flow of
data on customer purchases. Such a project need not be costly; but it
would present to executives basic facts about customers of value in assess-
ing current operations and in planning future moves.
Facts Gathered
Facts as facts are presented in the statistical appendix to this report.
Our interpretation of them is presented partly in this appendix, but
largely in this summary and in the section on major findings which fol-
lows.
FROM MAJOR FINDINGS: How "independent" or "interdependent" are
the markets of branches and the main store?
Department stores have huge trading areas. In Philadelphia they get
about half of all business from customers living outside the city and
county of Philadelphia. Their branches, as in this case, are located well
within the trading area of the parent.
A basic problem of branch operation, therefore, is the separation of
branch customer groups from parent store customers and the ways in
which customers patronizing both divide their patronage. The question
underlies policies of centralized and uniform operations and decisions
as to the promotional and merchandising emphasis given to branches.
Store records reveal its branches had relatively few credit patrons who
are not also credit patrons of the main store:
Exclusively Branch Patrons
Branch A 19%
Branch B 13%
Store records reveal, however, that of all credit expenditures by pa-
trons of the branches, the branches get the major share:
Percentage of Branch Patrons' Credit Expenditures at Branches
Branch A 61%
Branch B 55%
310 MARKETING RESEARCH
Although a strong showing for the branches, these figures actually
understate the true "competitive" position of branches and the main
store, for the parent gets much business from branch patrons not avail-
able to the branches themselves.
This extra business includes mail and telephone purchases, purchases
of basement store merchandise, and purchases of sales merchandise.
(During the ten-month period analyzed, there were twenty-two store*
wide special sales events at the parents store plus ten additional special
basement store sales. When branch business is compared to personal
purchases in upstairs departments on regular days, then exclusively
branch patrons jump from 16 per cent to 24 per cent of the total (an
increase of 50 per cent) and the proportions making more than half of
all purchases at the branches jumps from 55 per cent to 66 per cent (an
increase of 20 per cent).
Conclusion: Much more important than these facts may be a point of
view. In any problem of branch and parent, or branch vs. parent, the
tendency certainly is to consider them as about 20 per cent of total busi-
ness. But branch customers as people are responsible for 35 per cent or
more of total business. The addition of a third branch unquestionably
will mean that customers patronizing only the parent store will account
for less than half of all business.
This raises no necessary question as to centralized operations. It does
suggest a serious danger that policies and activities good for operations
of the main store may sometimes be unwisely applied to the branches or
forced upon them. Both the facts and the point of view about the im-
portance of branch customers will be helpful in discussing later some
apparent opportunities for increased business.
Most noteworthy, perhaps, is that exclusive patrons of branches are
seldom more frequent patrons than those who also purchase at the main
store in downtown Philadelphia, For the latter as a group, branch store
purchasing represents extra shopping, not a substitute for downtown
shopping trips.
ACTION NEEDED
1. What graphical methods might be employed for this report?
2. What improvements can be made in the report: format,
style, themes, and so forth?
3. How should the oral presentation, to occupy about thirty
minutes, be made?
CASES IN PRESENTATION AND FOLLOW-UP 311
FROM STATISTICAL TABLES SECTION:
NUMBER OF DIFFERENT DAYS CREDIT CUSTOMERS PURCHASED AT BRANCHES
BY STORES PATRONIZED *
(In percentages)
Number of different
days purchased
Stores patronized
Branch A
Branch B
Main store
only
Inactives
1 to 3
42
25
26
7
33
32
24
11
43
17
33
7
48
28
15
9
4 to 9
10 to 21
22 or more
Total .
100
100
100
100
* From store records of 617 customers, including, however, only thirty cus-
tomers of Branch B only. Note that, despite a "vocal minority" of branch
patrons who when interviewed claimed they were "always running over" to a
suburban center, about two thirds of all credit customers bought less than ten
times in ten months, or less than once a month. Even if all cash purchases of
these credit customers were made on separate days, still less than half would buy
oftener than once a month (except for the category of Branch B and mam store
patrons estimated at 64 per cent). Also, even if those patronizing branches
only one to three times are omitted, to avoid distortion of percentages by occa-
sional buyers, still one half of patrons buy less than once a month.
NUMBER OF DIFFERENT DAYS CUSTOMERS PURCHASED AT MAIN STORE
BY STORES PATRONIZED *
(In percentages)
Number of different
days purchased
Stores patronized
Branch A
and
main store
Branch B
and
main store
Main store
only
1 to 3
47
36
14
3
38
40
20
2
39
40
18
3
4 to 9
10 to 21
22 or more
Total
100
100
100
* Based on store credit records for all types of purchases. Since mail and
telephone orders are counted, this does not reflect personal visits entirely. It
will be noted that more than three fourths of customers purchase less than once
a month. Even if all customers patronizing main store only one to three times
are omitted, still two thirds are patrons less than once a month. Except for one
category (patrons of Branch B and main store), it is noteworthy that branch
visits are moree frequent than visits to downtown stores.
312 MARKETING RESEARCH
CASE 5 FOLLOW-UP ON PRODUCT
PERFORMANCE STUDIES
BACKGROUND
Tabor, Inc., a manufacturer of pharmaceuticals and food
products, had by intensive advertising secured a large volume of
sales for a personal-use product.
The product was packed in boxes of twenty-four and forty-
eight units. Early in the 1940's the company had introduced
a machine for dispensing single units for consumption at foun-
tains and lunch counters. By 1947 these dispensers were in
use in 40 per cent of the stores handling the product and in
70 to 80 per cent of all stores with fountains and lunch counters.
The dispensers, however, had proved troublesome. Espe-
cially in humid weather the apparatus clogged, resulting in
unsightly, unsanitary, or inoperable dispensers. Soon the
company felt it necessary to provide each salesman with a repair
kit. On each call the salesman dismantled, cleaned, and reas-
sembled the dispenser.
After the war, the sales department became seriously con-
cerned over the amount of time spent by salesmen on this non-
selling activity. No one in the company believed that sales
through dispensers should be discontinued, even though they
were a small part of total sales. In addition to being excellent
advertising, the satisfaction of one trial of the product, the com-
pany believed, often developed a steady customer for home use
of the product.
As a solution for the problem, Tabor, Inc., introduced early
in 1947 a new type of dispenser utilizing a new, special package.
By the middle of 1948, 36 per cent of all outlets carrying the
product had the new-type dispenser (13 per cent still retained
the old-type dispenser).
Before long new difficulties became apparent. The 49 per
cent of stores with dispensers reached by mid-1948 proved to
be a peak. By late 1950 only 28 per cent had the new-style dis-
penser, of which one in five was out-of-stock on the special pack-
CASES IN PRESENTATION AND FOLLOW-UP 31?
age designed for use in the new dispenser, and nine per cent still
had the old-style dispenser.
FOLLOW-UP PROBLEMS
As the situation developed, the research staff of Tabor became
cumulatively more convinced that another dispenser was neces-
sary; but management remained unpersuaded. Well over a
year after the first formal report had been circulated among all
department heads and other top executives, and despite a total
of three special reports on the dispenser problem, no action had
been taken.
The research director and staff felt that the methods of pres-
entation and follow-up might have been faulty. Perhaps they
had been overconfident about the conclusiveness of the facts.
The first report, being comprehensive, may have failed to em-
phasize the main propositions. Perhaps they had been too
aggressive, or not aggressive enough, in relaying their findings
and conclusions. Because the first change in dispensers had
fared so poorly, it could be expected that management would
be cautious about further changes. In summary, the research
staff members were reviewing their past actions in order to con-
sider how they might effectively present the facts gathered.
REPORTS SUBMITTED ON DISPENSERS
The first report was submitted in March, 1951. Excerpts
and condensations from this report follow.
The Basic Problem is twofold: 1. Our dispensers are in about half
the retail outlets with fountains and lunch counters, well below the pre-
war level of 70 to 80 per cent. 2. We are showing continued distribution
losses, having dropped steadily from a peak of 36 per cent in midsummer
of 1948 to our present 25 per cent.
[A table showed the percentage of outlets using old- and new-style dis-
pensers, and totals, since the introduction of the new-style dispenser
early in 1947.]
Moreover, a national survey late in 1950 showed that one store in five
did not have the special dispensing package in stock. Our records (in-
complete) show 27,419 dispensers placed in stores.
Thus, the new dispenser and new package are losing ground. What
314 MARKETING RESEARCH
are some of the possible reasons for this? (1) Out-of -stock occurs be-
cause the new package is an incidental extra. When "out-of-stock"
occurs, retailers use a regular package. (2) Some retailers object to carry-
ing the extra item. (3) It is very difficult to tell how many units are in
the dispenser. Reorders do not get in the "want book" until too late to
prevent running out. (4) The retailer makes more profit by using a reg-
ular package, the special package listing for $1 .00 and the regular package
for 87 cents. (5) Some retailers sell the special package to preferred
customers, which, since most retailers order in quarter or half dozen lots,
reduces stock rapidly and causes "outs."
In conclusion it would seem that: 1. Perhaps a new type of dispenser
using regular packages from stock, on which we net a larger return, could
be developed. 2, As a compromise it would be helpful to change the
special package dispenser so that the amount remaining in it would be
visible at all times. 3. Steps should be taken continuously to remind
retailers to keep an adequate stock of the special package at all times.
This report secured the wholehearted support of a regional
office; and the letter received in April from this regional office
was reproduced and distributed without comment to all those
receiving the report.
Early in June, 1952, a second report on dispensers was cir-
culated. Excerpts and condensations follow.
Current data show we now have dispensers in thirty-three per cent
of the stores carrying the product (9 per cent with old-style and 24 per
cent with new-style dispensers), or about 17,000 stores. For the fourth
straight year dispenser usage has shown a loss. (A table presented the
annual figures for both styles; another table showed percentage usage by
regions and percentage changes for each of the five years beginning just
prior to the introduction of the new-style dispenser.)
Only in the Southeast (+140 per cent) and in the Southwest (+133
per cent) has dispenser usage increased. In the five other regions usage
is down from 8 to 41 per cent, usage in the United States, as a total, be-
ing down 21 per cent. The percentage of sales of the special package is
steadily decreasing, from 19.0 per cent in 1949 to 15.3 per cent in 1950
to 12. 5 per cent in 1951.
Why is the new dispenser not more successful? There are several
reasons: 1. The special package is an item of minor importance to re-
tailers. 2. Retailers make a greater profit on sales at the fountain or
lunch counter by using regular stock. Because of larger quantity orders
CASES IN PRESENTATION AND FOLLOW-UP 315
for regular packages, the average discounts run almost 50 per cent higher
than for special packages. The net price per unit is 1 .60 cents for the
regular and 1 .94 cents for the special. 3. When out-of-stock occurs, sub-
stitution is easily made with the regular packages. 4. Despite the small
size of the average order (one quarter dozen), turnover of the special
package is very slow. 5. Inability of fountain help to tell when stock is
running low.
How then can dispenser business be regained? By having a new dis-
penser especially designed for use of regular stock merchandise. This
would prevent out-of-stock and re-establish a valuable advertising mes-
sage at many points. If sold through our own salesmen, it could help
build volume orders.
Following this report one executive suggested that the com-
pany design a plastic package. The research staff was asked to
comment on this suggestion and late in June submitted a mem-
orandum from which excerpts and condensations follow:
The proposed plastic package has merit provided we can substitute it
for the present special package. We do not favor adding another pack-
age type, unless it is proved by testing in the market that another pack-
age would be profitable saleswise. The markup on the proposed package
would give us a margin about half as large as that on regular packages.
A substantial saving in production, storage, and handling costs can be
realized by eliminating one package. In the past five years, exclusive of
storage and handling costs at warehouses, we have invested nearly
$300,000 in dispensers. The new dispenser has not been too successful.
Its use has declined for the past four years and only 24 per cent of the
stores carrying our product now have the new-style dispenser.
Some of the reasons why the new dispenser has not been successful
are: 1. Retailers make a greater profit on sale of regular packages; 2.
Retailers object to carrying an extra package; 3. Turnover is very slow;
and 4. Fountain and counter help are unable to determine when stock
is running low.
It is our opinion that we need a new dispenser for regular-package
stock. Our recommendation for adopting the new package was made
after considering the following: Advantages: utility and simplicity; self-
service feature; would occupy front-counter position; visible supply at
all times; new display card each time a package is opened; the possibility
of a substantial saving by eliminating one package or discontinuing dis-
pensers. Disadvantages: pilferage. Also, does the proposed package
have equal display value?
316 MARKETING RESEARCH
ACTION NEEDED
In view of the failure of management to act on these reports,
an evaluation of their effectiveness seemed to be called for.
This evaluation required consideration of the following ques-
tions:
1. Should the reports have included discussions on patent and
legal problems, threatened raw material shortages, and in-
ability to replace worn-out dispensers during the war?
2. Should the relative importance of dispensers to other pro-
motional efforts have been assessed in the reports?
3. Had the reports failed to emphasize the main issues?
4. Were the facts available conclusive or had the analysis been
faulty?
5. How should the reports have been presented and what other
action might have been taken by the research staff?
6. Should further efforts be made now, and, if so, what efforts
and why?
Parf IX
ORGANIZATION FOR RESEARCH
Two different organizational problems are covered under this
heading. One problem is the organization of a staff to work on
a particular research project, at the termination of which the
staff is disbanded. The second problem is the formation of a
presumably permanent research department within a firm
which would handle a variety of research problems. In partic-
ular, the latter problem raises the question of the place of re-
search within the company as a line function, a staff function,
or a combination of both.
Organization for a Particular Project
With either problem, the task of hiring a staff and setting up
a going organization is much the same. In hiring people for
a particular research project, however, more emphasis can be
placed on capability, experience, and knowledge of specialized
tasks, and less on permanency and potentialities for advance-
ment. Even in the case of the director of the project, more
weight can be placed on technical competence and less on
ability to sell one's self and the value of research. Specializa-
tion can sometimes be achieved even on a small project by con-
tracting out some of the work to research agencies concentrating
on certain operations; for example, tabulating concerns.
Budgeting for a research project is a fine art learned largely
through experience. One general rule that can be offered, how-
ever, is that it is better to overestimate the costs than to under-
estimate them, as it can prove very awkward to request
additional funds. At the same time, few things are likely to
please a client or sponsor as much as to be informed that costs
will be less than anticipated and that a refund is in order.
317
318 MARKETING RESEARCH
Organization of a Research Department
Irrespective of the place or form of setup, there are, in gen-
eral, two broad groups of considerations that affect its organiza-
tional structure. First, there are external considerations, which
in an industrial business or firm focus attention upon the rela-
tionship of marketing research activities to those of other opera-
tions of a business, such as finance, production, and sales. In
a research firm, these considerations pertain to the relationship
of the firm to its clients. The second consideration relates to
the internal organization of marketing research activities, either
in a research agency or in the marketing research department of
a business.
Line vs. staff form of organization. A line organization
makes it possible to place marketing research under the vice-
president in charge of all research activities or under the vice-
president in charge of some operating division, usually sales.
Such an arrangement maintains direct lines of responsibility
from top executives to individual workers. The danger of this
form of organization is that the marketing research activities
might be slanted toward research in favor of that particular
division. This is especially true because the head of a given
division may tend to develop his work independently, and to
emphasize his own work without proper consideration of other
activities which are also necessary functions of the entire enter-
prise. It also provides some temptation to use research to vindi-
cate present activities rather than to find out what is actually
taking place. Thus, an essential quality of good research work,
an impartial outlook, may be lost.
The staff organization, also known as the functional organ-
ization, enables a high degree of specialization in marketing
research activities to be achieved by other divisions of the busi-
ness. By this means, special knowledge and skills can be ex-
ploited within each division as well as outside of it. Thus,
when the production department runs into special marketing
problems in connection with product development, the market-
ing research specialists attached to that division can be called
upon to help.
ORGANIZATION FOR RESEARCH 319
A drawback to the staff type of operation is that coordination
of activities between marketing researdh functions and those of
other departments may be difficult, and much time may have to
be spent in liaison activities. Sometimes lines of authority be-
come confused, and, as a consequence, it may not be clear just
who is responsible for research decisions. Since there are more
persons involved in planning operations, the time consumed in
arriving at a decision may be greater than in the line organiza-
tion. Finally, one of the by-products of staff organization is
the development of specialists to such an extent that an over-all,
broad point of view may be lacking. This is an ever-present
danger as the marketing research activities of a business become
more specialized.
The "line and staff" organization represents a compromise
between the two previous forms, seeking to combine the ben-
efits of both. In effect, the line executives are supported by the
advice of staff personnel who are added to the main-line activ-
ities. Hence, marketing research operations relating to mer-
chandising are placed, for example, as a staff activity under the
vice-president in charge of merchandising. The function of
the marketing research man and his assistants in this division
would be to aid the merchandising executive in formulating
policy or in arriving at operating decisions. The actual deci-
sions, however, rest with the merchandising officer. The mar-
keting researcher merely advises; he has no authority beyond
that.
The advantages of this type of organization are its flexibility,
greater opportunity for making effective use of the abilities and
desires of personnel (because of the greater variety of jobs), and
better coordination of activities by the line officers.
External considerations for a research agency. The external
considerations facing a research agency revolve around the
maintenance of goodwill of potential customers and, in par-
ticular, of its clients. Building up a reputation for competent
work, supplemented by speeches and articles by executives, is
about the only way of gaining the goodwill of users of market-
ing research. Such activities also greatly facilitate the "care
and feeding of clients/' The latter, however, is a much more
320 MARKETING RESEARCH
specialized task. Contact with the client is needed throughout
the course of a study to insure avoidance of misinterpretations
and the meeting of client specifications.
The person in the research agency maintaining this contact
varies with the agency's size and scope. In large agencies cer-
tain individuals, so-called "client-service men/' are assigned to
this work on a full-time basis. Theirs is the responsibility of
coordinating the clients' wishes with what the agency's research
personnel are doing. On large-scale or highly technical studies,
however, the research man in charge may also work with the
client, since the client-service men are not selected primarily
for their technical abilities.
Internal Considerations
Internal organization problems for marketing research are
concerned with such matters as the size and scope of organiza-
tion, allocation of personnel, and space considerations.
The factors determining the size and scope of a marketing
research department are numerous and varied. The adjudged
importance to the company of marketing research work, the
type of company, its financial situation, and the personalities
involved all play dominant roles. In the past, companies in
consumer goods fields have tended to do more extensive market-
ing research than those producing and selling industrial goods.
Advertising agencies are generally more active in research than
other firms of similar sizes, providing it in part as an additional
service to their clients. The size of research departments tends
to expand rapidly when a company is prosperous and contract
sharply when business falls off the very time research is needed
most.
The organization of independent research firms depends
largely on the whims of the owners and the type of business
available. Some firms attempt to develop a general research
agency, accepting all types of business, though often subletting
individual stages of a study to specialized agencies. Other firms
prefer to remain small and establish a reputation in one par-
ticular type of research, either in certain stages of a research
operation, such as editing and tabulating, or in certain research
ORGANIZATION FOR RESEARCH 321
fields, such as product-testing, readership surveys, and so on.
Perhaps the outstanding development in recent years among
research firms has been the rise of these specialized agencies.
The Cases in This Part
The four cases in this Part are divided equally between organ-
ization for a particular project and organization of a research
department within a firm. Of the two cases on organization
for a particular project, Case 1 relates to a contemplated three-
year project on consumer habits of buying life insurance, and
Case 2 to a longer-term consumer panel operation within a
research firm. In effect, the first case deals with organization
for large-scale research for a limited period, whereas the prob-
lem in the second case is organization of one particular research
program on a presumably permanent basis.
Research for a department store and for a pharmaceutical
company are the subject of the two cases (3 and 4) on organ-
ization of a research department. These cases are somewhat
similar in that both raise questions of function, place, and or-
ganization of a research department within a firm. Case 4,
however, poses a problem of a somewhat broader nature, since
the objective is to integrate economic research with marketing
research within the same department.
BIBLIOGRAPHY
AMERICAN MARKETING ASSOCIATION. The Technique of Marketing Research.
New York: McGraw-Hill Book Co., Inc., 1937. Contains much material
on internal organization, including selection of a survey staff (pp. 47-53), se-
lection of field personnel (pp. 149-63), and organization and operation of
field work (Chapters 9 and 10).
BRADFORD, E. S. Marketing Research. New York: McGraw-Hill Book Co.,
Inc., 1951. Pages 298-300 are devoted to a code of professional practices.
BROWN, L. O. Marketing and Distribution Research. New York: The Ronald
Press Co., 1949. Chapter 2 presents a general discussion of the different
users of marketing research with some treatment of the functions and or-
ganizational structure of marketing research departments.
CAPT, J. C. "Market Research without a Market Research Department."
Marketing Series No. 75. New York: American Management Association,
1949. Discusses ways and means by which smaller companies not having
regular research departments can undertake research activities.
HOBART, D. M. (ed.) . Marketing Research Practice. New York: The Ronald
322 MARKETING RESEARCH
Press Co., 1950. A good discussion of the type of organizations engaged in
marketing research (Chapter 2) and of the different forms of research organ-
izations (Chapter 3).
LORIE, J. H., and H. V. ROBERTS. Basic Methods of Marketing Research. New
York: McGraw-Hill Book Co., Inc., 1951. A discussion of what research
functions should be performed internally for a company (pp. 320-25), includ-
ing some highly useful suggestions on the use of an outside research agency.
LUCK, D. J., and H. G. WALES. Marketing Research. New York: Prentice-
Hall, Inc., 1952. Chapter 2 discusses organization of marketing research,
with emphasis on the diverse functions of marketing research departments
(pp. 27-38) and with a list of considerations entering into the use by a com-
pany of outside research firms and consultants (pp. 44-46).
NATIONAL INDUSTRIAL CONFERENCE BOARD. Organization for Market Re-
search. Part I, Studies in Business Policy No. 12, 1945; Part II, Studies in
Business Policy No. 19, 1946. Contains very useful information on the or-
ganization of marketing research departments based on actual research de-
partments.
Part IX Cases
CASE IADMINISTERING A LIFE INSURANCE
STUDY
BACKGROUND
In 1952, a study was planned to ascertain consumer buying
habits and buying motives for life insurance as well as consumer
attitudes toward life insurance. Although a great deal of atten-
tion has been given to the selling strategy that life insurance
companies develop for their sales representatives, little atten-
tion has been given to the basic factors that influence consumers
in their purchase of life insurance. If such information could
be obtained, it would greatly assist in developing a sound basis
for the marketing of life insurance both from the standpoint of
personnel selling efforts and from that of advertising programs.
It was proposed that consumer buying habits with regard to
life insurance be explored from the viewpoint of units pur-
chased, from whom purchased, and frequency of purchase.
Buying motives would be studied in relationship to such factors
as old age, disability, and death benefits, tax and estate prob-
lems, or reciprocity arrangements between buyers and sellers.
While data were collected on these subjects, it was also thought
possible to gather data regarding the understanding that con-
sumers have about the types and uses of life insurance as well
as their expectations regarding agents' services, the major rea-
sons why people do not buy life insurance, and why they have
to be influenced by extreme emotional approaches into buying
insurance.
The planners of the research project felt that the selling and
advertising policies of life insurance companies were based
primarily on the most intelligent guess possible regarding con-
323
324 MARKETING RESEARCH
sumer behavior toward life insurance purchasing. For this rea-
son, data relating to consumer buying habits and motives would
be useful in formulating the policies of these companies toward
their advertising and selling programs, thereby permitting a
somewhat more objective approach to be taken to these prob-
lems than was currently being followed.
THE PROBLEM
The following statement of objectives provides a more de-
tailed idea of the aims of this project:
1 . To study the specific buying habits of consumers in their purchases
of life insurance by covering solicited and unsolicited sales of life insur-
ance and relating this to (1) unit size and kind of purchases, (2) type of
agent (part-time or full-time, age, business handled, and so on), and (3)
period of time between purchases of life insurance and time over which
purchases extend; or frequency of purchase.
2. To investigate the buying motives that are given for the purchase
of life insurance, such as (1) conservation of personal assets; (2) program
aspects of death, old age, and disability, (3) credit, (4) debt liquidation,
(5) taxes, and (6) reciprocity between buyer and seller.
3. To determine the relationship between the above buying habits
and buying motives with regard to age, occupation, income, marital
status, sex, and war service.
4. To study the relationship between consumers* knowledge of types
and uses of life insurance with regard to the factors that seem to influ-
ence them in their purchase of life insurance. It is of especial interest
to determine whether or not agents are giving consumers the information
they desire about life insurance and whether or not consumers have con-
fidence in agents.
TYPE OF INFORMATION SOUGHT
Sex
City size
Age
Occupation
Income
Ownership
Amount
Type of insurance
CASES IN ORGANIZATION FOR RESEARCH 325
Per cent of income for insurance
Form of savings preferred *
Attitude toward buying insurance
Have adequate life insurance
Additional insurance desired
Settlement plans
Purpose of insurance
Marital or family status
Not married and no dependents
Not married with dependents
Wife (or husband) and children
Wife (or husband) and no children
Children not with wife (or husband)
No children not with wife (or husband)
Home ownership
Attitude toward life insurance agents
Bought or been sold
Initiated purchase with agent
Number of agents calling on consumers
Family attitude toward purchase of life insurance
Method of approach to selling of life insurance
Request from consumer by mail
Request agent to call
Attitude toward insurance as an investment
Willingness to give complete information to agent
Own recent economic experiences
Deal with man or woman agent
Are savings features valued?
What information do consumers want about life insurance?
How long had client known agent from whom insurance was purchased?
Reasons for purchase of life insurance
Help agent
On basis of analysis of life insurance made for consumer
Promotional materials
Materials received through mail
Materials from insurance agents
Are materials helpful?
Did materials help select company?
Did agent help select company, or yourself?
Basic consumer knowledge about life insurance
Even though these statements seemed to be workable form-
ulations for the planners of the project, some troublesome prob-
lems faced the research workers. For example, it was pointed
out to them that respondents would be reluctant to report on
326 MARKETING RESEARCH
some of the items set forth in the objectives, such as uses for
debt liquidation, taxes, and establishing a base for credit. Fur-
thermore, there would be no assurance that the reports of
respondents would be accurate unless an actual check could
be made, through life insurance companies, of the policies
actually held by respondents.
Approximately $50,000 a year was available for the work over
a period of three years. From these funds, it would be necessary
to defray the expenses of interviewing and travel expenses con-
nected with the study. Travel would involve a distance of 100
miles a day round trip for the interviewers in reaching the city
where the interviews would be conducted. Additional expense
would be incurred in going from one sample household to the
other. It was estimated that each interview would require about
45 minutes. It was thought necessary to provide a probability
sample.
DATA AVAILABLE
The following past studies in this general area were available
at the time this study was planned :
Life Insurance Ownership among American Families. Ann Arbor: Survey Re-
search Center, University of Michigan. Published annually 1948-1952.
YOUNG, J. R. The Life Insurance Buyer in Baton Rouge. Baton Rouge:
Louisiana State University Business Bulletin, XII, No. 2 (October, 1950).
ACTION NEEDED
1. Outline an organization plan for a project of this type.
2. Prepare a tentative budget for its operation, assuming that
a director for the project is obtainable at a salary of $9,000
per year.
CASE 2 SETTING UP A CONSUMER PANEL
OPERATION
BACKGROUND
In 1950, the Ames Company launched a continuous national
consumer panel operation. The objective of the operation was
CASES IN ORGANIZATION FOR RESEARCH 327
to collect data on all consumer expenditures, on radio and TV
listenership, and on newspapers and magazines entering the
home. With these data, analyses could be carried out on such
varied subjects as the characteristics of consumers purchasing
specific goods and specific brands, the effect of advertising pro-
grams on consumer purchases, and the relative effectiveness of
different advertising media.
The data were to be collected by the consumer diary method
on a weekly basis. In other words, sample members were to be
sent diaries every week containing about twelve or sixteen pages
in which they were to list all their purchases during the follow-
ing week, the purchase of each item being recorded individually
under a specific heading containing a place for the date, size,
amount spent, brand name, and type of store at which the pur-
chase was made. At the end of the week the family would re-
turn that week's diary and begin filling in the new one, which
would have arrived several days before. To obtain the coopera-
tion of these families, it was decided that some sort of premium
plan would have to be instituted, although nobody seemed sure
what sort of plan would best retain the cooperation of the panel
members.
The plans called for a mailing sample of some three thousand
families scattered over the nation. To handle such a large
operation, it was estimated that a staff of between one hundred
and one hundred fifty persons would be required. This staff
was to be concentrated in one city, for reasons of economy.
THE PROBLEM
The task facing the planners of this operation was to prepare
an organization chart for this staff, breaking it up into depart-
ments and outlining the function of each department, and to
set up a procedure to be followed in processing the diaries and
issuing reports. Such reports would be issued on a monthly
basis, it was contemplated, with a separate type of report for
each client. In addition, special analyses and reports would be
prepared when requested by clients.
328 MARKETING RESEARCH
DATA AVAILABLE
At the time this study was planned, the following published
material on panel operations was available:
CHURCHMAN, C. W. (ed.) Measurement of Consumer Interest, Philadelphia:
University of Pennsylvania Press, 1947.
GAUDET, H., and C. DANIEL. Radio Listener Panels. Washington, D. C.:
Federal Radio Education Committee, 1941.
LAZARSFELD, P. F., and M. FISKE. "The 'Panel* as a New Tool for Measuring
Opinion," Public Opinion Quarterly, II (October, 1938), pp. 596-612.
PARTEN, M. Surveys, Polls and Samples. New York: Harper & Bros., 1950.
Chapter 3.
ACTION NEEDED
The following are needed for organizing this operation:
1. An organization chart showing the division of the staff by
departments and other major operational units, as needed.
Administrative positions and responsibilities have to be in-
dicated alongside the chart.
2. A flow chart outlining the direction of flow of the work,
including rough estimates of the timing.
3. Some sort of premium plan for securing maximum coopera-
tion from the panel members.
CASE 3 ORGANIZING A DEPARTMENT STORE
RESEARCH DEPARTMENT
BACKGROUND
A. C. Bertram, Inc., a member store in an Eastern city of one
of the department store chains, was considering late in the
forties the establishment of a Marketing Research Department.
Except for some centralization of buying, the individual stores
in the chain were largely autonomous in operations. No
branch stores were operated at the time, although A. C. Bert-
ram, Inc., was looking forward to branch operations in the
future. The store's volume of sales had been increasing steadily
and was now well past $40 million annually. Profits also were
increasing. The store was organized into six major divisions
CASES IN ORGANIZATION FOR RESEARCH 329
headed by a secretary and treasurer, a controller, a merchandise
manager, a sales promotion manager, a store manager, and a
personnel director.
THE PROBLEM
The formation of a research staff or department had been
considered several times before and rejected. The executive
committee of the store considering the proposal at the present
time seemed to agree, however, that the store now was able to
afford such a move and that it was desirable at least to try it.
Discussion centered primarily on organization issues, partic-
ularly the activities of such a department and the executive to
whom the research director should be made responsible.
INFORMATION AVAILABLE
The origin of department store research in any formal sense
dates back to 1916. Under the sponsorship of the famous
Boston merchant, Lincoln Filene, the forerunner of the present
Research Division of the Associated Merchandising Corpora-
tion (AMC) was then founded. Research departments in
stores affiliated with the AMC gradually followed for exchange
of information with AMC and for independent operations.
Even in the late forties, however, formalized research in the
department store field was notably scant. Except in a few
chains, few department stores had established separate research
staffs with their own directors. The executives of A. C. Bert-
ram, Inc., therefore, had no well-established precedent organ-
izationally, certainly nothing comparable to the divisional plan
as formulated by Paul M. Mazur. 1 A prevalent idea, in fact,
was that research needs of companies varied so widely that no
single organizational plan was feasible.
ACTION RECOMMENDED
In considering the activities, or functions, which might be
performed by the proposed marketing research department, a
number of points of view developed. Some of the old feeling
*P. M. Mazur. Principles of Organization Applied to Modern Retailing,
(New York: Harper & Bros., 1927).
330 MARKETING RESEARCH
persisted that most analysis and research were operating prob-
lems best left to the various executives and their assistants. For
example, the advertising department, under the sales promo-
tion manager, regularly checked sales results of advertised items.
Unless this function was carried on within the division, it was
argued, the full cooperation of the personnel and even of some
executives might be hard to secure. Moreover, this was a daily,
repetitive operation based on a system of analysis already estab-
lished and was not any new or special problem requiring the
services of a research department. The counter argument was
made that in general a researcher could not, of course, be ex-
pected to be wholly unbiased in his fact-finding and recommen-
dations if he was in any way checking the work of a man to
whom he was directly responsible. As to possible failure of an
executive to utilize what a researcher reported to him, it was
argued, any executive was eager for real facts or else he wasn't
"worth his salt/'
Similar points were made concerning the budgeting and fore-
casting, expense control, and merchandise statistics operations
under the controller. During the discussions, the application
of market research to many other aspects of store operations
was noted. The personnel and records of the comparison
shopping bureau, under the merchandising manager, it was
claimed, might be highly useful in checking on customer service
problems (customer services were under the province of the per-
sonnel director), especially in relation to competitor offerings
and their acceptance. A research department also might be
asked to study delivery problems and the perennial problem,
returned goods (under the store superintendent) . Problems in-
volving displays (window, interior, and other), direct mail, store
traffic, departmental layouts, store systems, and selling effective-
ness, among others, were also suggested by some executives as
potential subjects within the province of a research department.
It was even proposed that unit control methods and operations
(of stock control) were a research problem.
The experience of a store which devoted most of its research
staff time to detecting and studying difficulties in weak depart-
ments was related. In this regard, however, the store's own
CASES IN ORGANIZATION FOR RESEARCH 331
experience with customer analysis was brought up as a point
against considering research, particularly the highly statistical
type, as a broad merchandising and promotion function.
Through the urging of a former controller, the store had once
established detailed records for each credit customer on all pur-
chases. This had proved extremely expensive. The principal
use to which these records were put was for direct mailings.
The records showed, for example, which customers had bought
women's shoes, and these were sent mailings at the time of
special shoe promotions. No one believed that the results had
been worth the expense, and one executive expressed serious
doubts that the experiment as conducted was conclusive evi-
dence on the value of customer analysis. Full agreement was
expressed that without frequent consultation some researchers
were likely to produce broad, interesting, and useless facts.
In particular, the general merchandise manager remarked
that all this use of questionnaires and surveys to find out what
customers wanted to buy was nonsense. All you had to do was
put an article on the counter. The tens of thousands of cus-
tomers passing it by within a few days would tell you whether
they liked it or not by what they bought. The answer would
be got quicker, cheaper, and more accurately than by any "high-
falutin research." More generally, he felt that retail research
differed greatly from research for manufacturers, who were far
distant from consumers and interested only in their own brands.
Retail customers were always close at hand. An executive
could observe them any day by going out on the sales floors.
In addition, store records provided all kinds of information
to supplement the numerous outside sources: want slips and
customer suggestions, unit control records on sales and stocks,
unit and dollar transaction data, sales slips, delivery records,
complaints, comparison shopping records, NRDGA reports,
Federal Reserve Board indices, newspaper studies of the trad-
ing area, and so on. There was no need to go out and ask cus-
tomers more questions, unless, of course, researchers had found
a way to answer the only really puzzling question: Why people
bought what they did the way they did.
Despite these and other avowed limitations and weaknesses
332 MARKETING RESEARCH
of marketing research, and despite some opposition to some
aspects of research in general, the fact that good marketing re-
search could be genuinely useful in some ways to each of the
major store divisions was reaffirmed. A proposal was made,
therefore, that a decentralized research department be estab-
lished with operating units in all, or most, of the store divisions.
This seemed to have considerable merit as a long-range objec-
tive. Some day each division might have separate research
bureaus, or there might be one central staff for storewide prob-
lems with subsidiary staffs of specialists in the divisions. At
present, it was agreed, this would be too expensive, and it was
better to learn to walk before trying to run. A staff composed
of a director, his assistant, a secretary, and part-time helpers was
visualized as a beginning.
The feeling was that the research department should report
directly to the president or to one of the major executives. The
secretary-treasurer, in particular, was considered a good possible
choice. This man was highly regarded in the organization, was
known to be extremely favorable to research, and had been
one of the most active proponents of the proposed department.
In addition, it was considered to be a possible advantage that
his division would have probably the least direct need for re-
search studies.
ACTION NEEDED
1. Where should the proposed marketing research department
be located? Why?
2. What might the principal functions of the department be?
How might projects originate?
3. Should the research director participate in policy formation?
If so, how?
CASES IN ORGANIZATION FOR RESEARCH 333
CASE 4 A RESEARCH DEPARTMENT FOR
A PHARMACEUTICAL COMPANY
BACKGROUND
The Eli Lilly and Company laboratories were founded in
1876 in Indianapolis, Indiana. By 1952 the organization had
become one of the major producers in the United States of
pharmaceutical and biological products in both the domestic
and foreign markets. In maintaining its position of leadership
in the drug industry, it had continuously engaged in extensive
research work in both the technical aspects of pharmaceutical
production and in the marketing and economic aspects.
The marketing operations of Lilly were extensive, its products
being sold on a worldwide basis. As a consequence, there was
need not only for marketing research, but also for basic eco-
nomic research. For example, operational data had to be col-
lected on the activities of wholesale druggists, retail druggists,
and competitors, for use in studies to determine significant dif-
ferences in each aspect of drug distribution influenced by the
activities of the various types of business just mentioned. In
addition, inventory control studies, evaluation of the effects
of fair trade legislation, and similar studies had to be made in
so far as these factors influenced the operations of retail drug-
gists, wholesalers, competing firms, and the Lilly Company.
Many additional marketing problems derived from the com-
pany's use of a three-pronged method of distribution the
wholesaler, the retailer, and the physician. The company rec-
ognized the wholesaler as the one who acted as a distributor
for Lilly products, the retailer as the one who dispensed the
products, and the physician as the one who prescribed them.
It can be seen, therefore, that the physician was the major link
in the chain of distribution and that he had to be kept informed
at all times of special drug products.
The sales representative who attempted to convince doctors
of the merits of Lilly products was known as a detail man. . His
job was that of selling the merits of the drug products made by
334 MARKETING RESEARCH
Lilly so that the doctor believed in them sufficiently to use
them repeatedly in his daily practice. The detail man not only
had to call on doctors, but also had to keep in close touch with
pharmacists and wholesale houses to make certain that ad-
equate stocks of Lilly products were being carried to fill the
prescriptions that doctors wrote. Whenever possible, a stren-
uous effort was made to bring new products to the attention of
all key individualsdoctors, retail druggists, and wholesalers.
This called for information that would be helpful in doing an
accurate job of forecasting.
Many other forecasts were required for company operations
in addition to those needed for promotional use. The main
type of forecasts and related data required were as follows:
1. Forecasts of Eli Lilly:
a) Total sales.
b) Commodity sales.
c) Individual products.
2. Estimates of market for new products.
3. Estimates of stock requirements (Eli Lilly Company) for new prod-
ucts in order to supply market.
4. Surveys providing comparative price levels of Lilly and competitive
products.
5. Incidence of disease data.
6. Mortality data on given diseases.
7. Public health data on geographic distribution of diseases and pub-
lic health services to treat them.
The Lilly Company also had need for industry studies to pro-
vide the following information (past and present):
1 . Volume of medicinal market:
a) Consumed by a given country.
b) Produced by a given country.
c) Exported and imported by a given country.
2. Channels used by medicinal industry to move products from pro-
ducer to consumer.
3. Market potentials for Lilly products by geographical areas for use
in deploying field manpower and promotional effort effectively.
4. Determination of normal trading areas for wholesale and retail
drug trade in each country.
CASES IN ORGANIZATION FOR RESEARCH 335
5. Studies of field manpower of Lilly and principal competitors, and
adequacy of Lilly field force in a given country.
6. Periodic studies on the adequacy of wholesaler distribution and
support given Lilly line in each geographical area.
7. Development of proper frequency for daily calls and town visits
standards, and methods for territorial scheduling.
Further information was needed on operations of the com-
pany that are of primary concern to management in formulating
its policies, namely:
1. Plant location.
2. Annual studies of company performance in scientific research ef-
forts that lead to new products and to creating additional sales
volume.
3. Management problems of mterfunctional interest as assigned or
approved (excludes sales forecasts and estimates of market); for
example, utilization of production facilities and their expansion.
4. Management problems of into/functional interest when requested
at divisional levels or above (excludes sales forecasts and estimates
of market); for example, continuation of research on a product.
5. Periodic reviews of low sales volume products and package sizes
(Lilly products) for purposes of deletion.
6. Operational data of competitors, wholesale druggists, and retail
druggists for use in comparative studies, pricing studies, inventory
control, and similar factors.
Certain information was looked upon as being exclusively
!n the realm of marketing studies. This information covered
"he following subjects:
1. Studies to measure and to improve the effectiveness of Lilly pro-
motional efforts through such channels as advertising, house or-
gans, direct mail, and personal sales efforts.
2. Studies on physician and druggist attitudes to company products
and policies.
3. Surveys, as needed, on company product need and product accep-
tance.
4. Surveys indicating the sales position of Lilly products in their re-
spective therapeutic fields and geographical areas.
5. Surveys showing total current consumption of products in given
therapeutic fields.
336 MARKETING RESEARCH
6. Surveys providing comparative price levels of Lilly and competitive
products.
THE PROBLEM
Each grouping of the information needed indicated that
there would be some overlapping between the classes of data
sought. The problem which faced the Lilly Company was,
therefore, to work out a plan of internal relationships that would
organize the research activities as indicated by the type of in-
formation sought. Furthermore, there was an external organ-
ization problem that had to be solved so that the economic
and marketing research activities could be planned in proper
working relationship with other divisions of the company, such
as production, technical product research, finance, and sales.
The executive structure of Lilly is found in many types of
business operation; namely, a president, executive vice-presi-
dent, and other officers who perform the customary activities
associated with their positions.
ACTION NEEDED
1. Work out an organization chart to show the relationship
of economic and marketing research activities to other divi-
sions of the company.
2. How would you organize the economic and marketing re-
search activities within the research division? Show the
organizational chart and how groupings of information
could be handled from the standpoint of internal manage-
ment of the division.
INDEX
Accuracy, meaning, in sampling, 115-
16
Advertising, evaluating effectiveness of,
263-71
Analysis and interpretation, 233-41;
problems in, 242-84, 293-95
Arbitrary selection, 116-17; see also
Judgment sample
Area sampling, 118
Bacon study, preceding for, 211-13
Bias, 116
Branch store profitability, problems in,
17-22, 301-11
Budgeting for research, 317; problem
in, 323-26
Charge accounts, problems in, 91-100,
223-31
Charts, use in reports, 286; problems
in, 290-92, 301-11
Cluster sampling, 118-19
Consumer clinic, 31; nature of, 52-54;
problem in, 52-56
Consumer panel, 122; data available
from, 61-62; nature of, 59-61; prob-
lems in, 56-62, 87-91, 143-46, 326-
28
Costs, estimation of, 66; see also Budg-
eting for research
Credit ratings, problem in, 147-52
Data gathering, 31, 64-66, 159-64;
problems in, 70-75, 110-14, 127-30,
165-201
Dehydrated foods, survey of, problem
in, 23-24
Department store research, organiza-
tion for, 328-32
Depth interviewing, 31; problem in,
192-96
Dispenser study, report on, 312-16
Distnbution costs, problem in, 25-28
Double sampling, 120-21
Editing, 203-4; problem in, 208-32
Experimental methods, 64
Exploratory investigations, 31-32; see
also Informal investigations
External investigations, 31-32; see also
Informal investigations
Findex method, 205-6
Follow-up study, 110-14, 147-52; see
also Presentation and follow-up
Forms, 65; design of, problems in, 167-
71, 213-16; see also Questionnaires;
Tables
Foundry industry, problem in, 127-30
Frozen foods, problem in, 208-11
Functional organization; see Staff
function
Data reliability of, 233-35; problems Hypotheses, formulation and testing
in/ 242-45, 278-84 of, 29-32, 63, 235-37
337
338
INDEX
Ice cream stuc|y, 102-5
Informal investigations, 31-32; prob-
lem in, 40-45
Installment credit, problem in, 87-91
Internal investigation, 29-30; problems
in, 46-52, 79-87
Internal records, 1 59-60; problems in,
165-67, 167-71
Interpretation of results, 236-39; pit-
falls in, 236-37; policy recommenda-
tions, 238-39, 271-77, 278-84; prob-
lems in, 245-50, 251-63, 263-71,
293-96
Interviewer training and selection,
problems in, 180-84, 184-92
Investigations; see Preliminary investi-
gations
Jewelry survey, tabulation of, 213-16
Judgment sample, 117-18; problems
in, 136-39, 152-57
Keysort tabulation, 205-6
Laundry operation, collecting data for,
165-67
Life insurance study, problems in,
editing and tabulation, 217-23; in-
terpretation of results, 245-50; in-
terviewer selection, 180-84; organ-
ization for, 323-26
Line function, research as a, 287-88,
318-20
Luggage market, interviewer instruc-
tions, 184-92; preparing question-
naire, 106-10; problem in, 15-17
Magazine merchandising effectiveness,
problem in, 79-87
Magazine readership, problem in, 152-
57
Mail questionnaires, 64, 160-61; prob-
lems in, 100-2, 147-52, 171-72
Mark sensing, 205
Market for men's clothing, problem
in, 278-84
Marketing a product, problem in, 12-
14
Marketing efficiency, problem in, 52-
56
Marketing research, task of, 3
Mechanical data collection methods,
64, 162
Multipurpose survey, 136; problem in,
136-39
Multistage sampling, 121
Observational methods, 64
Organization for research, 317-22;
problem in, 323-36
Paint study, problem in, 35-38
Personal interviews, 64, 161; problem
in, 192-96
Pharmaceutical company, organization
for research, 333-36
Planning, 63-69; costs, 66; forms, 65;
methodology, 64-65; nature of task,
63; problem in, 70-1 14; specific pur-
poses, 63-64; testing, 65-66
Policy recommendations, consider-
ations in, 238-39; problems in, 263-
71, 271-77, 328-32
Preceding, 204; problem in, 211-13
Prefabricated housing, interpretation
of results, 263-71; survey of, 136-39
Preliminary investigations, 28-34; ex-
ternal investigation, 31-32; internal
investigation, 29-30; nature of task,
29; secondary data, 30-31
Presentation and follow-up, 285-89;
problem in, 290-316
Pretesting, 65-66; problems in, 91-100,
173-80
Probability sample, 117-18; see dso
Sample designs
Problem formulation, 3-8; nature of
task, 3-4; problem in, 9-27; rephras-
ing problems, 4-5
INDEX
339
Questionnaires, construction of, 65;
problems in, 100-2, 102-5, 106-10,
173-80, 211-13; testing of, 65-66;
problem in, 91-100
Quota sampling, 120 footnote
Radio listenership, problems in, inter-
pretation of results, 251-63; plan-
ning, 70-75; sample design, 139-43
Random numbers, sample page of,
126; see also Random selection
Random selection, 116-17; problem in,
125-27
Reports, problems in, 296-301, 301-11,
3 12- 16; types of, 285-87
Research department, organization of,
318-21
Respondent cooperation, obtaining,
173-80
Retail trade areas, problem in, 293-96
Sales methods, problems in, 9-10, 38-
39
Sales organization, problem in, 11-12
Sample designs, 118-22; problems in,
130-36, 136-39, 139-43, 143-46,
152-57
Sampling, 1 1 5-24; basic concepts, 115-
18; need for, 1 1 5; objective of, 116;
principal designs, 118-22; problem
in, 125-57
Sampling error, 116; of stratified sam-
ple, 120
Schedules; see Forms; Questionnaires
Secondary data, investigation of, 31-32
Self-service meat marketing, problem
in, 296-301
Sequential analysis, 121
Shopping habits, preliminary investiga-
tion of, 40-45
Situation analysis, 29-30
Staff function, research as a, 287-88,
318-20
Store audit, data available from, 61-62;
nature of, 57-60; problem in, 56-62
Store expansion, problem in, 75-79
Store hours, problem in, 271-77
Stratified sampling, 119
Survey methods, types of, 64; see also
Sampling
Tables, summary, preparation of, 223-
31, 285-87; problem in, 278-84
Tabulation, manual vs. machine,
204-6; problems in, 208-32, 251-63,
278-84
Telephone interviews, 64, 161-62
Television survey, problem m, 171-72
Testing, 65-66
Unrestricted sampling, 118
Validation, 234-35; problem in, 242-
45
Verification of findings, 237-38
Working outlines, 32, 65, 287
Live Music Archive
Librivox Free Audio
Metropolitan Museum
Cleveland Museum of Art
Internet Arcade
Console Living Room
Books to Borrow
Open Library
TV News
Understanding 9/11