Skip to main content

Full text of "A study of American intelligence"

See other formats


I 



A STUDY OF AMERICAN 
INTELLIGENCE 



A 

STUDY OF AMERICAN 

INTELLIGENCE 

By carl c! BRIGHAM, Ph.D. 

ASSISTANT PROFESSOR OF PSYCHOLOGY 
IN PRINCETON UNIVERSITY 

A FOREWORD 
By Robert M. Yerkes, Ph.D. 

CHAIRMAN RESEARCH INFORMATION 
service: NATIONAL RESEARCH COUNCIL 



PRINCETON 

Princeton University Press 

LONDON: HUMPHREY MILFORD 
OXFORD UNIVERSITY PRESS 

1923 






Copyright 1922 by Carl C. Brigham 

Second Printing 



PRINTED AT THE PRINCETON UNIVERSITY PRESS, PRINCETON, U.S.A. 




FOREWORD 



Two extraordinarily important tasks confront our nation : 
the protection and improvement of the moral, mental and 
physical quality of its people and the re-shaping of its in- 
dustrial system so that it shall promote justice and encour- 
age creative and productive workmanship. I have been 
asked to write this Foreword because of my official con- 
nection, as chief of the Division of Psychology, Office of 
the Surgeon General of the Army, with psychological ex- 
amining during the war, but I have consented to write it 
because of my intense interest in the practical problems 
of immigration and my conviction that the psychological 
data obtained in the army have important bearing on some 
of them. 

When in April, 1917, 1 visited Canada to learn what use 
our neighbors were making of psychological principles and 
methods in their military activities, I found Mr. Carl C. 
Brigham attached as psychologist to the Military Hospitals 
Commission. With him as my guide, I spent several hours 
in interviewing military and civil officers and in discussing 
our mutual problems and needs. The valuable information 
which Mr. Brigham helped me to secure and his advice 
contributed substantially to the report which I later pre- 
sented to my professional colleagues at home, and to rep- 
resentatives of the United States army. 



vi AMERICAN INTELLIGENCE 

In October, 1917, our friend, eager for larger opportuni- 
ties for professional service than the Canadian army prom- 
ised, accepted appointment in the Sanitary Corps of the 
United States army for psychological service. He aided 
efficiently in the trials of methods of examining at Camp 
Dix, New Jersey, and he was then ordered to the office of 
the Surgeon. General in Washington to help with the revis- 
ion of tests and the preparation of new methods. Thus he 
became thoroughly familiar with the procedures and results 
of psychological examining in the army, while at the same 
time contributing generously of ideas, labor and enthu- 
siasm. With deep satisfaction I use this opportunity to men- 
tion Mr. Brigham's national service and his exceptional 
fitness to study and to discuss the relations of army meas- 
urements of intelligence to nativity and residence. 

It appears that Mr. Charles W. Gould, a clear, vigorous, 
fearless thinker on problems of race characteristics, amal- 
gamation of peoples and immigration, raised perplexing 
questions which drove Mr. Brigham to his careful and 
critical re-examination, analysis, and discussion of army 
data concerning the relations of intelligence to nativity 
and length of residence in the United States. In a recently 
published book, America, A Family Matter, to which this 
little book is a companion volume, Mr. Gould has pointed 
the lessons of history for our nation and has argued strongly 
for pure-bred races. 

For the observational data which Mr. Brigham used in 
preparing this book we are indebted to the competent and 
devoted company of psychologists which during the war 



FOREWORD vii 

labored in camp and laboratory on the preparation of meth- 
ods, the conduct of examinations, and the application of re- 
sults. But the fruits of the labors of these many psychol- 
ogists might have been lost to the world had it not been for 
the insight, zeal, and industry of Carl R. Brown, Mark A. 
May and Edwin G. Boring, who evolved methods of statis- 
tical treatment, applied them and prepared the resulting 
materials for publication. 

Mr. Brigham has rendered a notable service to psychol- 
ogy, to sociology, and above all to our law-makers by 
carefully re-examining and re-presenting with illuminating 
discussion the data relative to intelligence and nativity 
first published in the official report of psychological exam- 
ining in the United States army. Far from belittling or 
casting doubt on the general reliability of the results con- 
tained in the report, he has essentially confirmed the major 
findings in the field of his special inquiry and has adduced 
new evidences of the trustworthiness and scientific value 
of the statistical methods used by military psychologists. 
His task has been arduous and difficult, involving an im- 
mense amount of tedious labor for mathematical calcula- 
tions and critical study of results. The volume which is the 
outcome of Mr. Brigham's inquiry, and which I now have 
the responsibility and satisfaction of recommending, is sub- 
stantial as to fact and important in its practical implica- 
tions. It is not light or easy reading but it is better worth 
re-reading and reflective pondering than any explicit dis- 
cussion of immigration which I happen to know. The 
author presents not theories or opinions but facts. It be- 



viii AMERICAN INTELLIGENCE 

hooves us to consider their rehabihty and their meaning, 
for no one of us as a citizen can afford to ignore the menace 
of race deterioration or the evident relations of immigra- 
tion to national progress and welfare. 

Robert M. Yerkes 

Washington, D. C. 
June 1922 



CONTENTS 

PAGE 

Foreword v 

Introduction xix 

Part I : The Army Tests 

Section 1 examination alpha 3 

Section 2 examination beta 32 

Section 3 the individual examinations 54 

Section 4 reliability of the measures 59 

Part II : Statistical Analysis of the Army Test 
Results 
Section 1 the principal sample 75 

Section 2 analysis of the main groups of the 

PRINCIPAL sample 77 

Section 3 analysis of the white draft into 

FOREIGN AND NATIVE BORN 84 

Section 4 analysis of the foreign born white 

DRAFT into YEARS OF RESIDENCE 
GROUPS 88 

Section 5 analysis of immigration to the 

UNITED states 112 

Section 6 analysis of the foreign born white 

DRAFT BY COUNTRY OF BIRTH 118 

Section 7 reliability of the results 154 

Section 8 the race hypothesis 157 

Section 9 re-examination of previous con- 
clusions IN THE LIGHT OF THE RACE 
HYPOTHESIS 177 

Section 10 comparison of our results with the 
conclusions of other writers on 
the subject 182 

Conclusions 197 



IX 



PLATES 

PLATE PAGE 

I Alpha test 1 : oral directions 5 

II Alpha test 2 : arithmetical reasoning 9 

III Alpha test 3 : practical judgment 13 

IV Alpha test 4 : synonym-antonym 18 
V Alpha test 5 : disarranged sentences 21 

VI Alpha test 6 : number series completion 24 

VII Alpha test 7 : analogies 26 

VIII Alpha test 8 : information 29 

IX Beta test 1 : maze 35 

X Beta test 2 : cube analysis 38 

XI Beta test 3 : x-o series 41 

XII Beta test 4 : digit-symbol 44 

XIII Beta test 5 : number checking 47 

XIV Beta test 6 : picture completion 50 
XV Beta test 7 : geometrical construction 53 



FIGURES 

FIGURE PAGE 

1. Distribution of scores of the oral directions test 6 

2. The Gaussian normal distribution 7 

3. Distribution of scores of the arithmetical reason- 

ing test 10 

4. Distribution of scores of the practical judgment 

test 14 

5. Distribution of scores of the synonym-antonym 

test 17 

6. Distribution of scores of the disarranged sentence 

test 20 

7. Distribution of scores of the number series com- 

pletion test 23 

8. Distribution of scores of the analogies test 25 

9. Distribution of scores of the information test 28 

10. Black-board chart for demonstrating the maze 

test 33 

11. Distribution of scores of the maze test 34 

12. Black-board chart for demonstrating the cube 

analysis test 36 

13. Distribution of scores of the cube analysis test 37 

14. Black-board chart for demonstrating the X-0 

series test 39 

15. Distribution of scores of the X-0 series test 40 

16. Black-board chart for demonstrating the digit- 

symbol test 42 

xi 



xii FIGURES 

FIGURE PAGE 

17. Distribution of scores of the digit-symbol test 43 

18. Black-board chart for demonstrating the number 

checking test 45 

19. Distribution of scores of the number checking test 46 

20. Black-board chart for demonstrating the picture 

completion test 48 

21. Distribution of scores of the picture completion 

test 49 

22. Black-board chart for demonstrating the geomet- 

rical construction test 51 

23. Distribution of scores of the geometrical con- 

struction test 52 

24. The normal distribution curve 59 

25. A skewed distribution curve 60 

26. Examination alpha as independent of education 65 

27. Distribution of intelligence scores according to 

rank 66 

28. Success in Officers' Training Camps as predicted 

by examination alpha 67 

29. Comparison of army test records with various 

independent criteria 69 

30. Success in civil occupations compared with army 

test records 70 

31. Distributions of scores of the white officers, white 

draft, and negro draft on the combined scale 81 

32. Distributions of scores of the native born and for- 

eign born white draft on the combined scale 87 

33. Apparently increasing average intelligence with 

increasing years of residence 94 

34. Distributions of the alpha scores of three groups 108 

35. Analysis of immigration by countries 114 

36. Relative standing of the nativity groups accord- 

ing to their average intelHgence 124 



FIGURES xiii 

FIGURE PAGE 

37. Relative standing of the nativity groups in the 

proportions of A and B men, and D, D and 

E men 146 

38. The proportion of each nativity group obtaining 

scores at or above the average of the white 
ojQficers 149 

39. The proportion of each nativity group at or below 

the average of the negro draft 151 

40. The proportion of each nativity group testing be- 

low the approximate "mental age" of eight 153 

41. Analysis of immigration to the United States ac- 

cording to the estimated amount of Nordic, 
Mediterranean and Alpine blood 164 

42. Volume of immigration by decades 166 

43. The distributions of the intelligence scores of the 

Nordic, Mediterranean and Alpine groups 170 

44. The distributions of the intelligence scores of the 

English speaking Nordic and the non-English 
speaking Nordic groups 173 

45. The distributions of the intelligence scores of the 

non-English speaking Nordic group and the 
combined Mediterranean and Alpine groups 175 

46. The decline of intelligence with each succeeding 

period of immigration 198 

47. The constituent elements of American intelhgence 200 



TABLES 



PAGE 

1. Distribution of the intelligence scores of the main 

groups of the principal sample on the combined 
scale 80 

2. Analysis of the white draft into foreign born and 

native born groups 86 

3. Analysis of the foreign born white draft by years 

of residence in the United States 90 

4. Comparison of the average scores on the combined 

scale of the five years of residence groups of the 
foreign born white draft 91 

5. Comparison of the average scores on the combined 

scale of the native born white draft with the five 
years of residence groups of the foreign born 
white draft 92 

6 . Per cen t . that emigration was of immigration for 1 5 

countries since 1908 98-99 

7. Distribution of alpha scores of five groups 106 

8. Per cent, of total immigration coming from var- 

ious countries during periods roughly correspond- 
ing to the five years of residence groups 113 

9. Analysis of the foreign born white draft by coun- 

try of birth : actual distributions 120-121 

10. Analysis of the foreign born white draft by coun- 

try of birth: percentage distributions 122-123 

11. Differences between England and other countries 126 

12. Differences between Scotland and other countries 127 

13. Differences between Holland and other countries 128 

14. Differences between Germany and other countries 129 

15. Differences between the United States and other 

countries 130 

16. Differences between Denmark and other countries 131 

17. Differences between Canada and other countries 132 



xiv 



TABLES XV 

PAGE 

18. Differences between Sweden and other countries 133 

19. Differences between Norway and other countries 134 

20. Differences between Belgium and other countries 135 

21. Differences between Ireland and other countries 136 

22. Differences between Austria and other countries 137 

23. Differences between Turkey and other countries 138 

24. Differences between Greece and other countries 139 

25. Differences between Russia and other countries 140 

26. Differences between Italy and other countries 141 

27. Differences between Poland and other countries 142 

28. Per cent, of each nativity group in the A and B 

groups 144 

29. Per cent, of each nativity group in the D, D 

and E groups 145 

30. Per cent, of each nativity group at or above the 

average of the white officers 148 

31. Per cent, of each nativity group at or below the 

average of the negro draft 150 

32. Per cent, of each nativity group below the approxi- 

mate "mental age" of eight 152 

33. Tentative estimates of the proportion of Nordic, 

Alpine and Mediterranean blood in each of the 
European countries 159 

34. Arrivals of alien passengers and immigrants, 1820 

to 1920 160-161 

35. Estimate of the amount of Nordic, Mediterranean 

and Alpine blood coming to this country from 
Europe in each decade since 1840 163 

36. Analysis of the foreign born white draft by races 169 

37. Analysis of the Nordic sample into an English 

speaking Nordic group and a non-English 
speaking Nordic group 172 

38. Population of the United States in 1920 203 



ACKNOWLEDGMENTS 



This study is a continuation of the work of the small 
group of psychologists who carried on the difficult task of 
analyzing the data from the army psychological examina- 
tions in the office of the Surgeon General of the Army. My 
presentation contains nothing new in methodology and is 
merely an extension of lines of investigation suggested by 
this group of workers. It rests on the foundations which 
they built. 

I wish to make especial acknowledgment to Colonel 
Robert M. Yerkes, who has read the manuscript several 
times in its various stages of preparation and has given 
many helpful suggestions. Professor Carl R. Brown of the 
University of Michigan, formerly of the Surgeon General's 
staff, assisted me when I first began to use the combined 
scale, and subsequently read Sections 1 to 7 of Part II in 
manuscript. Professor E. G. Boring of Harvard University 
read Sections 1 to 7 of Part II, and gave me invaluable 
assistance, especially in the treatment of Section 4. Pro- 
fessor Mark A. May of Syracuse University, Professors 
Edwin G. Conklin and Howard C. Warren of Princeton 
University read Sections 1 to 7 of Part II, and suggested 
many important changes. Without the assistance of all of 
these gentlemen I could not have carried through the task. 

Mr. Charles W. Gould suggested this continuation of 
the army investigations in the first instance, has sponsored 
the work throughout, has read and re-read all of the manu- 
script at every stage of its preparation, and is mainly 
responsible for the whole work. In my treatment of the 
race hypothesis I have relied on his judgment and on two 
books, Mr. Madison Grant's Passing of the Great Race, and 

xvii 



xviii AMERICAN INTELLIGENCE 

Professor William Z. Ripley's Races of Europe. These 
three gentlemen cannot be held responsible however for my 
percentage analysis of the present racial constitution of 
European countries, an analysis which I made as a novice 
in the field of anthropology, and for which I offer further 
apologies in the text of Section 8. 

Mr. David M. Maynard and Mr. Charles H. HeUiwell, 
two of my ujidergraduate students, assisted me faithfully 
in carrying through the laborious statistical calculations 
involved in using the combined scale. 

Carl C. Brigham 

Princeton, N. J. 
September 1922 



INTRODUCTION 



The question of the differences that may exist between 
the various races of man, or between various sub-species 
of the same race, or between pohtical aggregations of men 
in nationahty groups may easily become the subject of 
the most acrimonious discussion. The anthropologists of 
France and Germany, shortly after the close of the Franco- 
Prussian war, fought another national war on a small 
scale. It is difficult to keep racial hatreds and antipathies 
out of the most scholarly investigations in this field. The 
debate becomes especially bitter when mental traits are 
discussed. No one can become very indignant on finding 
his race classified by its skull dimensions, stature, or hair 
color, but let a person discover the statement that his race 
is unintelligent or emotionally unstable, and he is immedi- 
ately ready to do battle. 

Until recent years we have had no methods available 
for measuring mental traits scientifically, so that the lit- 
erature on race differences consists largely of opinions of 
students who are very apt to become biased, when, leaving 
the solid realm of physical measurements, they enter the 
more intangible field of estimating mental capacity. 

Gradually, however, various investigators using more or 
less refined psychological measurements commenced to as- 
semble a body of data that will some day reach respectable 
proportions. The status of the psychological investigations 
of race differences up to 1910 has been admirably sum- 
marized by Woodworth.i Since 1910, we have witnessed 

IR. S. Woodworth. Racial Differences in Mental Traiis, Science, New Series, Vol. 
31, pp. 171-186. 



XIX 



XX AMERICAN INTELLIGENCE 

in this country a remarkable development in methods of 
intelligence testing, and these methods have been applied 
to the study of race differences. Scattered investigations 
report and compare the intelligence scores of children of 
white, negro, or Indian parentage, and sometimes the 
scores of various nationality or nativity groups. The re- 
sults of these investigations are, however, almost impos- 
sible to correlate, for they have been made by different 
methods, by different measuring scales, on children of a 
wide variety of chronological ages, and above all, on com- 
paratively small groups of subjects, so that conclusions 
based on the studies have no high degree of reliability. 

For our purposes in this country, the army mental tests 
give us an opportunity for a national inventory of our own 
mental capacity, and the mental capacity of those we have 
invited to live with us. We find reported in Memoir XV of 
the National Academy of Sciences i the intelligence scores 
of about 81,000 native born Americans, 12,000 foreign 
born individuals, and 23,000 negroes. From the standpoint 
of the numbers examined, we have here an investigation 
which, of course, surpasses in rehability all preceding in- 
vestigations, assembled and correlated, a hundred fold. 
These army data constitute the first really significant con- 
tribution to the study of race differences in mental traits. 
They give us a scientific basis for our conclusions. 

When we consider the history of man during the half 
million years which have probably elapsed since the time 
of the erect primate, Pithecanthropus, the temporary polit- 
ical organizations, such as Greece, Rome, and our modern 
national groups, become of minor importance compared 
with the movements of races and peoples that have oc- 
curred. The tremendous expansion of the Alpine race at 
the end of the Neolithic and the beginning of the Bronze 

^^Psychological Examining in the United States Army. Edited by Robert M. 
Yerkes. Washington: Government Printing Office, 1921, Pp. 890. 



INTRODUCTION xxi 

Period, the submergence of this race by the Nordics in the 
2000 years preceding the Christian era, and the subsequent 
peaceful re-conquest of Eastern Europe by the Alpine Slavs 
from the Dark Ages on, represent an historical movement 
in comparison with which the Great World War of 1914 
resembles a petty family squabble. 

If the history of the United States could be written in 
terms of the movements of European peoples to this con- 
tinent, the first stage represents a Nordic immigration, for 
New England in Colonial times was populated by an almost 
pure Nordic type. There followed then a period of Nordic 
expansion. The next great movement consisted of the mi- 
grations of Western European Mediterraneans and Alpines 
from Ireland and Germany, a movement which started 
about 1840, and which had practically stopped by 1890. 
Since there is a considerable proportion of Nordic blood in 
Ireland and Germany, we should not regard the original 
Nordic immigration as a movement which stopped sud- 
denly, but merely as having dwindled to two-fifths or one- 
half of the total racial stock coming here between 1840 and 
1890. The third and last great movement consisted of mi- 
grations of the Alpine Slav and the Southern European 
Mediterraneans to this continent, a movement that started 
about 1890, and which has not yet ceased. Running parallel 
with the movements of these European peoples, we have 
the most sinister development in the history of this con- 
tinent, the importation of the negro. 

The army mental tests enable us to analyze the elements 
entering into American intelligence. The intelligence test 
records of the native born, the foreign born, and the negro 
are at our disposal. The records deserve the most serious 
study. But before considering the results of the army tests, 
a person should be well informed concerning the nature of 
the tests, and the manner in which they were constructed. 



xxii AMERICAN INTELLIGENCE 

The army psychological tests included three types of 
examination : 

(1) Group examination alpha, which included eight dif- 
ferent sorts of tests, most of which involved the ability to 
read English. 

(2) Group examination beta, which included seven dif- 
ferent sorts of tests, none of which involved the ability 
either to read English or to imderstand spoken Enghsh, 
the tests consisting of pictures, designs, etc., and being 
given by instructions in pantomime. 

(3) Individual examinations of two types: 

(a) Those involving the use of English, the Stanford 
revision of the Binet-Simon scale and the point 
scale, and 

(6) Those involving no English, consisting of con- 
struction puzzles, etc., the instructions being 
given by gestures, the "performance scale." 

When a detachment reported for psychological examina- 
tion, the first step was that of separating the English 
speaking and literate from the non-English speaking or 
iUiterate. Those who were both English speaking and liter- 
ate were given examination alpha. All others were sent to 
beta. At the close of examination alpha, all men who had 
made low scores were sent to beta. After examination beta 
had been given, the examiners tried to recall for individual 
examinations all men who had made a low score in beta. 
In the rush of examining it was impossible to recall all men 
for individual examinations who should have been given 
special examinations, and some men were graded on alpha 
who should have been graded on beta, and vice versa, but 



INTRODUCTION xxiii 

most men were properly graded by the rough methods in 
use. In each one of the examinations the range of scores 
was so great that most men had an opportunity to score. 

The great contribution of the committee that first de- 
vised the army examining methods and of the men who 
subsequently developed additional methods in the army 
consisted of creating and standardizing group examinations 
alpha and beta. The methods of individual examining were 
already in existence, the Stanford-Binet scale being an elab- 
oration of Binet's "mental age" scale, and the tests of the 
performance scale having been more or less completely 
worked out by other investigators. The task of examining 
men in large groups was first carried through successfully 
in the army. Before the war, many psychologists would 
have scoffed at the notion of examining two or three hun- 
dred men at once by giving them booklets containing dif- 
ferent sorts of tests, but the large group examinations be- 
came matters of daily routine. Group tests have subsequent- 
ly been tried out in schools and industries with excellent 
results from the standpoint of test administration. Indeed, 
when the army alpha examination was given at Ohio State 
University in October, 1919, practically the entire student 
body, 6000 in number, was tested by five examiners in 
eight hours. In the service, it was found that one examiner 
could control a group of 200 men with ease. The alpha in- 
structions were read by the examiner, and the men ordered 
to start and stop at the proper time. Examination beta was 
more difficult to administer, and was given to smaller 
groups. 

The statistical methods of treating the results of the 
army tests used in this study are rather intricate, but the 
principles involved are easily understood. At the outset we 
must frankly admit that there were minor errors in the 
three types of examinations given. We can not correct the 



xxiv AMERICAN INTELLIGENCE 

type of tests that were used, but we can correct the method 
of scoring them. Most of the difficulties of scoring arise 
from the fact that different types of measuring scales were 
used. During the war, the different scales were converted 
into one general scale of letter grades (A, B, C+, C, C , 
D and D ). This method was rough, and although it an- 
swered the purposes of the army at the time, it can not be 
used in any scientific interpretation of the results. 

Examination alpha was scored by finding the score on 
each of the eight tests, adding to get a total, and then con- 
verting the total into a letter grade. Beta was similarly 
scored. It is apparent that some tests in alpha might be 
more difficult than others, that some tests in beta might be 
easier than any test in alpha, and that variations might 
have occurred which it was impossible to predict at the 
time the examinations were made. Recognizing these facts, 
then, the army statisticians worked out another method of 
scoring the results, which eliminates all of these sources 
of error. This method is known as the combined scale, a 
theoretical intelligence scale running from to 25, into 
which the alpha, beta and individual examination scores 
may be converted, so that we finally have one measure- 
ment instead of three. 

Psychological measurements involve much more than 
creating tests and giving tests. After all the results are in, 
we still have the problem of interpreting the results, and 
this interpretation is largely a statistical problem. Too 
much credit can not be given to the staff of the Psycholog- 
ical Division of the Surgeon General's Office, who con- 
tinued in the service long after the war was over, patiently 
studying and analyzing the results. The combined scale 
was very largely the work of two young psychologists, Carl 
R. Brown and Mark A. May, and their work on this prob- 
lem, reported in Chapter 2, Part 3 of Memoir XV, is with- 



INTRODUCTION xxv 

out doubt the greatest contribution that has yet been made 
to the statistical phases of the science of mental measure- 
ment. 

The theory underlying the combined scale is simply that 
of regarding each test of alpha and beta as a separate 
measuring scale. One group of individuals including 1047 
men born in English speaking countries, was examined on 
alpha, re-examined on beta, and if possible, examined again 
on the Stanford-Binet scale. This group of 1047 cases con- 
stituted the basis on which a method of combining the sep- 
arate tests into a combined scale was empirically evolved. 

From now on in the course of our study of the army test 
records, we must regard alpha and beta as two booklets 
containing, in all, fifteen different measuring scales of in- 
telligence. The first step in the study consists of under- 
standing the nature of each of the fifteen scales. In Part I, 
the fifteen tests have been reproduced (Plates I to XV), 
and the actual records of the 1047 men shown in each in- 
stance, so that the reader may see exactly how the tests 
worked. 



PARTI 
THE ARMY TESTS 



SECTION I 

EXAMINATION ALPHA 

Alpha Test 1. Oral Directions 

The first test in alpha consisted of a series of commands 
or directions which were to be executed quickly. The in- 
structions, with the incidental commands about stopping 
and starting eliminated, are reproduced below. One may 
read the instructions for each item to himself slowly and 
turn the page to Plate I to test his own ability to execute 
the commands. 

Instructions : Oral Directions (Form 8) 

Item 1. Time limit : 5 seconds. 

"Make a figure 2 in the second circle and also a 
cross in the third circle." 

Item 2. Time limit : 5 seconds. 

"Draw a line from circle 1 to circle 4 that will 
pass below circle 2 and above circle 3." 

Item 3. Time limit : 10 seconds. 

"Make a figure 1 in the space which is in the 
square but not in the triangle, and also make a 
cross in the space which is in the triangle and in 
the square." 

Item 4. Time limit : 10 seconds. 

"Make a figure 2 in the space which is in the 
circle but not in the triangle or square, and also 
make a figure 3 in the space which is in the tri- 
angle and circle, but not in the square." 

Item 5. Time limit : 10 seconds. 

"If taps sound in the evening, then put a cross 

3 



4 AMERICAN INTELLIGENCE 

in the first circle; if not, draw a line under the 
word NO." 

Item 6. Time limit : 10 seconds. 

"Put in the first circle the right answer to the 
question: *How many months has a year.?' In 
the second circle do nothing, but in the fifth cir- 
cle put any number that is a wrong answer to 
the question that you have just answered cor- 
rectly." 

Item 7. Time limit : 10 seconds. 

''Cross out the letter just after F arid also draw 
a line under the second letter after I." 

Item 8. Time limit : 10 seconds. 

"Make in the first circle the last letter of the^r^^ 
word; in the second circle the middle letter of the 
second word and in the third circle the first letter 
of the third word." 

Item 9. Time limit : 15 seconds. 

''Cross out each number that is more than 50 but 
less than 60." 

Item 10. Time limit : 15 seconds. 

"Put a 4 or a 5 in each of the two largest parts 
and any number between 6 and 9 in the part next 
in size to the smallest part." 

Item 11. Time limit : 25 seconds. 

"Draw a line through every odd number that is 
not in a square, and also through every odd 
number that is in a square with a letter." 

Item 12. Time limit : 10 seconds. 

"If 4 is more than 2, then cross out the number 3 
unless 3 is more than 5, in which case draw a 
line under the number 4." 



1.O00OO 

2 t^i@li)0 




4 

5 O O O ^'es /Vo 

e OOOOO 

7 ABGDEFGHIJKLMNOP 

8 O O O MILITARY GUN CAMP 

9 34-79-56.87-68-25-82-47.27-31-64.93-71-41-52-99 

10 



11 HAOA [I]A[I1 

12 123456789 



PZa^e I. Alpha Test 1 : Oral Directions. 



6 AMERICAN INTELLIGENCE 

In scoring the papers one point was given for each cor- 
rect response. The group of 1047 individuals born in Eng- 
Hsh speaking countries obtained the following scores: 

Total score of test 12 3 4 5 6 78 9 10 11 12 

Number who made each score ... 73 78 93 116 100 121 131 94 82 67 52 28 12 

These scores are shown graphically in Figure 1, the 
horizontal direction indicating the total score from the 
lowest possible (0) to the highest possible (12), while 



TEST 1, 
ALPHA 



100- 




Figure 1. Distribution of scores of the Oral Directions test. (From 
p.624<,MemoirXV.) 



the vertical scale represents the number of cases getting 
each score, 72 at 0, 78 at 1, etc. 

For our purposes, we do not want a test that everyone 
can pass, for if everyone passed, no one would be graded. 
An ideal test would be one in which practically everyone 
could obtain some score and which very few could finish. 



AMERICAN INTELLIGENCE 7 

Then all people would be measured. An ideal test would 
also show a distribution of responses grouped symmetric- 
ally about the average, for, as a general rule, all measures 
of individual differences in mental traits show a distribution 
similar to the normal probability or chance distribution. 
An ideal test would give the type of distribution shown in 
Figure 2. 



200" 


^ 


150- 








lOG- 










J L 


50- 

0- 
( 


n M .1 ...... n 



Figure 2. The Gaussian normal distribution. 



Examining Figure 1, one may see that in general the oral 
directions test gives a type of distribution which is approx- 
imately similar to the Gaussian normal distribution shown 
in Figure 2. Our distribution is limited slightly at the lower 
end, and it is easy to imagine that the introduction of two 



8 AMERICAN INTELLIGENCE 

or three items easier than any in the present test would 
give us a step-down at the zero end of the scale similar to 
the one at the upper end. 

The oral directions test really gives an excellent score 
distribution. It is not a "speed" test in the popular sense, 
for the time limits for each item, while short, give ample 
time for following the directions. If a person understands 
the directions he can execute them easily in the time al- 
lowed. If the directions are not understood, an hour to 
execute them is no more generous than five seconds. In 
practice this test was useful in "acclimating" the recruit to 
the conditions of the examination. It was probably one of 
the poorest tests in alpha as a genuine test of intelligence, 
but it served its purpose as a "warming up" test. It is an 
adaptation of a type of test that has been used in psycho- 
logical laboratories for many years with rather mediocre 
results. 

Alpha Test 2. ^Arithmetical Problems 
Time limit : 5 minutes 

Test 2 is more of a reasoning test than a measure of 
proficiency in the fundamental arithmetical operations. 
The first items really constitute a literacy test, for if a 
person can read, he can answer the questions correctly. 
The distribution of scores in this test is shown in Figure 3. 
The zero scores (66 in number) are probably due to the 
inclusion of illiterates in the group of 1047 cases. Disregard- 
ing the zero scores, the distribution is regular. This test 
illustrates admirably the principle of fixing a time limit 
such that very few people can answer all the items cor- 
rectly. The approximate rule adopted in fixing the time 
limit in the first instance was that this limit should be such 
that not more than five per cent, of an unselected group 



Get the answers to these examples as quickly as you can. 
Use the side of this page to figure on if you need to. 

{1 How many are 5 men and 10 men? Answer ( 15 
2 If you walk 4 miles an hour for 3 hours, how far 
do you walk? Answer ( 12 

1 How many are 60 guns and 5 guns? , Answer ( 

2 If you save $9 a month for 3 months, how much will you 
save? Answer ( 

3 If 48 men are divided into squads of 8, how many squads will 
there be? Answer ( 

4 Mike had 11 cigars. He bought 2 more and then smoked 7. 
How many cigars did he have left?. Answer ( 

5 A company advanced 8 miles and retreated 2 miles. How far 
was it then from its first position? Answer ( 

6 How many hours will it take a truck to go 42 miles at the rate 

of 3 miles an hour? > .Answer ( 

7 How many pencils can you buy for 60 cents at the rate of 2 

for 5 cents? Answer ( 

8 A regiment marched 40 miles in five days. The first day they 
marclied 9 miles, the second day 6 miles, the third 10 miles, the 
fourth 6 miles. How many miles did they march the last 
day? , Answer ( 

9 If you buy 2 packages of tobacco at 8 cents each and a pipe for 
65 cents, how much change should you get from a two-dollar 
bill? , , . , Answer { 

10 If it takes 4 men 3 days to dig a 120-foot drain, how many men 

are needed to dig it in half a day? . .Answer ( 

11 A dealer bought some mules for $2,000. He sold them for 
$2,400, making $50 on each mule. How many mules were 
there? Answer ( 

12 A rectangular bin holds ^00 cubic feet of lime. If the bin is 

10 feet long and 5 feet wide, how deep is it? Answer ( 

13 A recruit spent one-eighth of his spare change for post cards 
and twice as mugh for a box of letter paper, and then had $1 .00 
left. How much money did he have at first Answer ( 

14 If 3H tons of clover cost $14, what will 6J^ tons cost?. .Answer \ 

15 A ship has provisions to last her crew of 700 men 2 months. 
How long would it last 400 men? Answer ( 

16 If an aeroplane goes 250 yards in 10 seconds, how many feet 
does it ^o in a fifth of a second? , Answer ( 

17 A U-boat makes 8 miles an hour under water and 20 miles on 
the surface. How long will it take to cross a 100-mile channel, 

if it has to go two-fifths of the way under water?. Answer ( 

18 If 134 squads of men are to dig 3,618 yards of trench, how 
many yards must be dug by each squad? .Answer ( 

19 A certain division contains 5,000 artillery, 15,000 infantry, and 
1,000 cavalry. If each branch is expanded proportionately 
until there are in all 23,100 men, how many will be added to the 
artillery? .Answer ( 

20 A commission house which had already supplied 1 ,897 barrels 
of apples to a cantonment delivered the remainder of its stock 
to 37 mess halls. Of this remainder each mess hall received 54 
barrels . What was the total number of barrels supplied? . . Answer C 

Plate II. Alpha Test 2 : Arithmetical Reasoning (Form 8). 

9 



10 



AMERICAN INTELLIGENCE 



would complete all the items in a test. In our group of 1047 
cases, 5 persons answered 18 problems correctly in the five 
minutes allowed, but no one answered more than 18 prob- 
lems correctly. Of course no one was expected to answer 
them all. If a person passed Q5% of each test in alpha he 
was graded "A"; perfection was not required. 



TEST 2, ALPHA 



100 




Figure 3. Distribution of scores of the Arithmetical Reasoning 
test. (From p. 624, Memoir XV.) 



One often hears the statement that the army tests were 
"speed" tests, and penalized the slow but accurate indi- 
vidual. Experiments were made to determine how the re- 
sults would change with extended time. A group of 475 
men examined showed in Test 2 an improvement from an 
average of 8.00 to 9.16 with double time. In five minutes 
they solved on an average 8 problems correctly, in ten min- 
utes 9.16. The relationship between single time and double 
time scores may be measured by the statistical value known 
as the coefficient of correlation. Two measures that stand 
in a perfect one to one correspondence have a coefficient 



AMERICAN INTELLIGENCE 11 

of correlation of 1.00. Two measures that stand in a perfect 
chance relationship have a correlation coefficient of 0. In 
practice it is found that a correlation of 0.90 is so high that 
one might substitute one series of measures for the other 
without seriously changing the results. The correlation be- 
tween the single time and double time scores was 0.937, a 
value so high that it indicates that there were very few 
changes in the relative position of the members of the 
group, and that such changes as occurred were small. 

The experiments that were conducted on time limits with 
the various tests all pointed to the conclusion that the re- 
sults would not be changed with the more extended time 
limits. Of course the absolute scores would be higher with 
the extended time, but the relative position of the mem- 
bers of the group would be about the same. In the experi- 
ment on double time referred to above, all the tests from 2 
to 8 in alpha showed coefficients of correlation between 
single time and double time above 0.90 except Test 3 
(0.879), and the correlation of the two total scores ob- 
tained under single and double time was 0.967. The army 
experimenters after considering all the evidence concluded 
that "doubling the time does not result in any demonstrat- 
able improvement in alpha as a whole." (p. 417). It is prob- 
ably true that very high scores depend on "speed," but 
inasmuch as a person only needed to answer correctly Q5% 
of the items to be rated "A" and 50% of the items to be 
rated "B," it can not be considered that "speed" is a factor 
that would affect the results seriously. 

The army findings of a correlation of 0.967 between the 
single time and double time trials of alpha, and the general 
conclusion that the results would not have been changed 
appreciably with more liberal time allowances, definitely 
controvert the popular belief that anything which is per- 
formed with a time limit handicaps the "slow but sure" 



12 AMERICAN INTELLIGENCE 

individual. Popular judgment classifies the population into 
two groups, the "slow but sure" and the "quick and inac- 
curate," and would have us believe that the quick type 
must of necessity be inaccurate, and that the sluggish indi- 
vidual is infallible. Science shows us that we really rate 
individuals on two scales, a scale of speed and a scale of 
accuracy, and that we find people who are both quick and 
accurate as well as people who are slow and inaccurate. 
Science would elaborate the popular classification by add- 
ing these two types. The popular "slow but sure" charac- 
terization is more apt to be an apology for dullness than 
a scientific diagnosis. At least in our consideration of the 
army test results we may definitely discard the opinion 
that we are testing "speed" rather than intelligence. The 
arithmetical reasoning test in alpha actually proved to be 
one of the best tests in the series. 



Alpha Test 3. Practical Judgment 

Time limit : Ij^ minutes 

The practical judgment test is one of the most interest- 
ing tests in alpha from many standpoints. There is no other 
test in alpha which contains, in all of the five forms used, so 
many individual items that may be criticised by a person 
who actually inquires into the logical validity of the an- 
swers accepted as correct. Item 12, for instance, might 
profitably be taken as the subject of an intercollegiate de- 
bate, as it has been the subject of many debates in the 
history of penology. The critics of the army tests are all 
too apt to consider the whole scale invalid if they can dis- 
cover a single incorrect item, for they fail to realize that a 
person could fall down on 35% of the individual items and 
still be rated "A." 



Tbis is a test of common sense. Bel6w are sixteen questions. Three answers are given to each 
question. You are to look at the answers carefully; then make a cross in the square before the best 
answer to each question, as in the sample: 

/ Why do we use stoves? 

sample! 



Because 
thfey look well 
they keep us warm 
they are black 

Here the second answer is the best one and is marked with a cross 
on until time is called. 



^ a 



Begin with No. 1 and keep 



1 It is wiser to put some money aside and not 
spend it all, so that you may 

D prepare for old age or sidcness 

D collect all the different kinds of money 

n gamble when you wish 

2 Shoes are made of leather, because 
D it is tanned 

P it is tough, pb'able and warm 
in it cari be blackened 

3 "Why do soldiers wear wrist watches rather 
than pocket watches? Because 

D they keep better time 
D they are harder to break 
Q they are handier 

4 The mam reason why stone is used for building 
purposes is because 

D it makes a good appearance 
n it is. strong and lasting 
Q it is heavy 

5 Why is beef' better food than cabbage? 



D it tastes betta* 

n it is mo^ nourishing 

it is hairdo to obtain 

If some one does you a favor, what dudd you 

do? 

D try to forget it 

n steal for him if he asks yoti to 

n return the favor 

If you do not get a letter from home, which you 

know was written, it may be because 

D it was lost in the mails 

D you forgot to tell your people to write 

D the postal service has been discontinued 

The main thing the. farmers do i& to 

Q supply luxiu-ies 

n make work for the unemployed 

Q feed the nation 

^"00 to. No. 9 above 



9 If a man who can't swim should fall into a 
river, he should 

O yell for help and try to scramble out 
D dive to the bottom and crawl out 
O lie on his back and float 

10 Glass insulators are used to fasten telegraph 
wires because 

O the glass keeps the pole from being burned 
D the glass keeps the current from escaping 
n the glass is cheap and attractive 

11 If your load of coal gets stuck in the mud, 
what should you do? 

D leave it there 

n get more horses or men to pull it out 

D thro^ off the load 

12 Why are criminals locked up? 
O to protect society 

n to get even with them 
D to make them work 

13 Why should a married man have: his life in 
snud? Because 

n death may come at any time 

n insurance companies are usually honest 

D his family will notihen suffer iJE he dies 

14 In Leap Year February has 29 days because 
D February is a short month 

D some people are bom on February 29th 
n otherwise the calendar would not come out 
right 

15 If you are held up and robbed in a strange city, 
you should 

D apply to the police for help 

P ask the first man you meet for money ta 

get home 
D borrow some money at a bank 

16 Why should we have Congressmen? Because 
n the people must be ruled 

D it insures truly representative government 
D the people are too many to meet and make 
their laws 



Plate III. Alpha Test 3: Practical Judgment (Form 8). 



13 



14 AMERICAN INTELLIGENCE 

The distribution of the scores made in Test 3 is shown 
in Figure 4. 



TEST 3, ALPHA 



150- 



100- 




Figure 4. Distribution of scores of the Practical Judgment test. 
(From p. 624, Memoir XV.) 



Disregarding the large number of zero scores (163), which 
are probably the result of illiteracy plus failure to under- 
stand instructions, and also recognizing the fact that a few 
low positive scores may be due to chance, we may look at 
the distribution as entirely satisfactory. 

Many persons object to the short time limit {V/^ min- 
utes), but the test was undoubtedly more effective with this 
short limit than it would have been with the longer time. 



AMERICAN INTELLIGENCE 15 

The average score improves from 6.32 to 9.85 with double 
time, and the correlation between Ij^ and 3 minutes work 
on the test is 0.879. There are decided indications that 
double time would not be useful in improving the record 
of those whose score was high in the first 1)^ minutes. 

A very excellent criterion of the efficiency of a test is 
its value in differentiating between officers and men. In 
general, a sample of officers would contain a larger per- 
centage of men of high intelligence than a sample of en- 
listed men. The amount that a test differentiates the groups 
would indicate the value of the test. This test of practical 
judgment was the worst test in the whole series in differ- 
entiating officers from men. If we used this criterion alone 
there would be no possible excuse for retaining the test in 
the series. In differentiating officers from men, it was about 
twice as bad as the next to the poorest test (oral direc- 
tions). 

On the other hand, we need tests in alpha which are 
effective at the lower end of the scale, and we can set up 
as our criterion here the value of the test in differentiating 
between feeble-minded individuals and enlisted men. The 
alpha tests were given to the high grade feeble-minded 
population of two custodial institutions, and the results 
compared with a group of 300 English speaking enlisted 
men. Test 3 proved to be very much superior to any other 
test in the series in differentiating between feeble-minded 
individuals and enlisted men. This fact more than justifies 
the inclusion of Test 3 in the scale. 

All of these facts are difficult to interpret. My own inter- 
pretation is that the sixteen items did not measure or grade 
"practical judgment" in any sense, but that the inclusion 
of at least one very obviously false and really quite silly 
alternative in each item acted as an effective pitfall for the 



16 AMERICAN INTELLIGENCE 

feeble-minded. At least we are sure that the actual experi- 
mental results are conclusive enough to dispose of any and 
all arm-chair criticisms. 



Alpha Test 4. Synonym-Antonym 

Time limit : Ij^ minutes 

If one will review the experimental literature of intelli- 
gence testing, he will find the synonym-antonym or "op- 
posites" test, used sometimes as a test of controlled asso- 
ciation, sometimes as a test of vocabulary, sometimes as a 
test of intelligence, but uniformly with excellent results. 
Given a group with a knowledge of English and sufficient 
intelligence to understand the nature of the problem, 
the synonym-antonym test will give as good a differentia- 
tion between the bright and dull members of the group, 
rated by an outside criterion, as any other standard test 
now available. 

The distribution of the scores in Test 4 is shown in 
Figure 5. 

The most striking feature of the distribution is the large 
number of scores that were either zero or one (393). This 
large number of zero scores is due to three causes. First, 
the illiterate group could not attempt it. Second, the stupid 
and literate could not understand the instructions and 
could not make the kind of judgment demanded. Third, in 
the long run chance or random responses would give scores 
around zero, for in scoring all tests that were a 50-50 
guess, the total score was the number of right responses 
minus the number of wrong responses. If a person under- 
lined ^'same^' for every item, his score would be 20 right 
minus 20 wrong, or zero. If he merely guessed, he would, 



400- 



350- 



300- 



250 



200- 



150- 



100- 



50- 



TEST 4, ALPHA 




Figure 5. Distribution of scores of the Synonym-Antonym Test. 
(From p. 625, Memoir XV.) 



17 



If the two words of a pair mean the same or nearly the same, draw a 
line under same. If they mean the opposite, or nearly the opposite, draw a 
line under opposite. If you cannot be sure,. guess. The two samples are 
already marked as they should be. 



SAMPLES i ^^ ^^^ same opposite 

V little small same opposite 

1 no ^yes same opposite 1 

2 day night same opposite 2 

3 go leave same opposite 3 

4 begin commence same opposite 4 

6 bitter sweet same opposite 5 

6 assume suppose same opposite 6 

7 command obey same opposite 7 

8 tease plague same opposite 8 

9 diligent industrious same opposite 9 

10 corrupt honest same opposite 10 

11 toward from- same opposite 11 

12 mascuUne feminine same opposite 12 

13 complex simple same opposite 13 

14 sacred hallowed same opposite 14 

15 often seldom same opposite 15 

16 ancient modem same opjwsite 16 

17 enormous gigantic same opposite 17 

18 confer grant same opposite 18 

19 acquire lose same opposite 19 

20 compute calculate .same opposite 20 

21 defile purify . . same opposite 21 

22 apprehensive ^fearful same opposite 22 

23 sterile fertile same opposite 23 

24 chasm abyss same opposite 24 

25 somber ^gloomy same opposite 25 

26 vestige trace.. same opposite 26 

27 vilify praise same opposite 27 

28 finite limited same opposite 28 

29 contradict corroborate same opposite 29 

30 immune susceptible same opposite 30 

31 credit debit same opposite 31 

32 assiduous diligent same opposite 32 

33 transient ^permanent same opposite 33 

34 paUiate mitigate same opposite 34 

35 execrate revile ..same opposite 35 

36 extinct extant same oppositfe 36 

37 pertinent relevant same opposite 37 

38 synchronous simultaneous.. . .same opposite 

39 supercilious disdainful same opposite 



39 



40 abstruse recondite same opposite 40 

Plate IV. Alpha Test 4: Synonym-Antonym (Form 8). 

18 



AMERICAN INTELLIGENCE 19 

in the long run, guess half of the responses right and half 
wrong. Chance scores would then be zero (which includes 
all minus scores) and 1 or 2 on the positive side. 

In general we may interpret Test 4 as a "high grade" 
test. It is too difficult to give any differentiation between 
low grade individuals, but it effectively grades the higher 
orders of intelligence. The time limit is not too short, for 
doubling the time only raises the average score from 10.50 
to 12.60, and the correlation between regular and extended 
time is 0.940. It is one of the most effective tests in the scale 
for differentiating officers from enlisted men, and for dif- 
ferentiating feeble-minded from enlisted men. The only 
criticism is that it was too hard for a large number of 
people examined. Figure 5 really gives only about half of 
the normal distribution. If the test were so easy that the 
lower end of the scale could be extended to about 20, the 
distribution would become normal. 



Alpha Test 5. ^Disarranged Sentences 
Time limit : 2 minutes 

This test is an adaptation of a type of test which gives 
excellent results in the Binet-Simon scale. As it stands in 
alpha it is not a particularly good test. The distribution of 
scores shown in Figure 6 indicates a pile-up of zero scores 
due probably to the same three causes described as op- 
erating in Test 4. The test is fairly good in differentiating 
between officers and enlisted men, but for some reason or 
other it is the very worst test in the whole series in dif- 
ferentiating between feeble-minded and enlisted men. On 
the whole it is one of the poorest tests in our measuring 
scale. 



250 n 



200 



150- 



100- 



50- 



TEST 5, ALPHA 




Figure 6. Distribution of scores of the Disarranged Sentence Test. 
(From p, Q2Q, Memoir XV.) 



20 



The words A EATS COW GRASS m that order are mLxed up and 
don't make a sentence; but they would make a sentence if put m the- 
right order: A COW EATS GRASS, and this statement is true. 

Again, the words HORSES FEATHERS BAYE. ALL would make 
a sentence if put in the order ALL HORSES HA\'E FEATHERS, 
but this statement is false. 

Below are twenty-four mixed-up sentences. Some of them are true 
and some are false. When I say "go," take these sentences one at a 
tkne. Think what each would say if the words were straightened out, 
but don't write them yourself. Then, if what it would say is true, draw 
a line under the word "true"; if what it would say is false, draw a line. 
vmder the word "false." If you can not be sure, guess. The two 
samples are already marked as they should be. Begin with No. 1 
and work right down the page until time is called. 



( a eats cow grass true, .false 

SAMPLES^ ^ 

V n 



feathers have all true'. : false 

1 oranges yellow are true . .false 1 

2 hear are with to ears. true, .false 2 

3 noise cannon never make a true, .false 3 

4 trees in nests build birds tpie. .false 4 

5 oil water not and will mix , true, .fake 5 

6 bad are shots soldiers all true.. false 6 

7 fuel wood are coal and for used true, .false 7 

8 ; moon earth the only from feet twenty the is true . . false 8 

9 to life w:ater is necessary true, .false 9 

10 are clothes all made cotton of true .. .false 10 

11 horses automobile an arfe than slower true, .false 11 / 

12 tropics is in the produced rubber true . . false 12 

13 leaves the trees in lose their fall true . . false 13 

14 place pole is north comfortable a the true . . false 14 

15 sand of made bread powder and is true . .false 15 

16 sails is steamboat usually by propelled a true . . false 16 

17 is the salty in water all lakes true, .false 17 

18 usually judge can we actions man his by a true . . false 18 

19 men misfortune have good never true . .false 19 . 

20 tools valuable is for sharp making steel true, .false 20 

21 due sometimes calamities, are accident to true . . false 21 

22 forget trifling friends grievances never true, .false 22 

23 feeling is of painful esaltation the true . . false 23 

24 begin a and apple acorn ant words with the true . . falsa 24 

Plate V. Alpha Test 5: Disarranged Sentences (Form 8), 

31 



2 AMERICAN INTELLIGENCE 

Alpha Test 6. ^Number Series Completion 

Time limit : 3 minutes 

This test is the only one in alpha demanding a high order 
of intelligence almost entirely independently of the use of 
language. The greatest difficulty was experienced with the 
instructions for this test, when the first trial was made at 
the army camps. The preliminary forms contained only two 
rows of samples, and the instructions included the rather 
involved statement: "In the lines below, each number is 
gotten in a certain way from the numbers coming before it. 
Study out what this way is in each line and then write in 
the space left for it the number that should come next. 
The first two lines are already filled in as they should be." 
In the final alpha revision, four samples were included, and 
the instructions were simplified verbally and read very 
slowly. The instructions were given as follows: "Look at 
the first sample row of figures at the top of the page : 2, 4, 
6, 8, 10, 12; the two numbers that should come next are, 
of course, 14, 16," etc., for each sample. Long pauses fol- 
lowed the reading of each sample. 

The distribution of responses given in Figure 7 shows 
that the simplified instructions gave very good results, for 
although there were many zero scores in our experimental 
group of 1047 cases (244), there were probably no more 
zero scores than might have been expected when we con- 
sider that the mere understanding of what was wanted re- 
quired considerable intelligence. On the whole the number 
series completion test proved to be entirely satisfactory. 



TEST 6, ALPHA 



150- 



100- 




50- 



Figure 7. Distribution of scores of the Number Series Completion 
Test. (From p. 624, Memoir XV.) 



23 



14 


16 


3 


2 


5 


5 


4 


7 



f 4 6 8 10 12 

9 8 7 6 5 4 

SAMPLES ^2233 4 4 

17 2 7 3 7, 

Look at each row of numbers below, and on the two dotted lines 
write the two numbers that should come next. 



3 4 5 6 7 8 

8 7 6 5 4 3 

10 15 20 25 30 35 

9 9 7 7 5 5 

3 6 9 12 15 18 

8 16 14 1 

5 9 13 17 21 25 

8 9 12 13 16 17 

27 27 23 23 19 19 

1 2 4 8 16 32 

19 16 14 11 9 6 

11 13 12 14 13 15 

2 3 5 8 12 17 

18 14 17 13 16 12 

29 28 26 23 19 14 

20 17 15 14 11 9 

81 27 9 3 1 Yz 

1 4 9 16 25 36 

16 17 15 18 14 19 

3 6 8 16 18 36 

Flate VI. Alpha Test 6: Number Series Completion (Form 8). 

24 



AMERICAN INTELLIGENCE 25 

Alpha Test 7. ^Analogies 

Time limit : 3 minutes 

The analogies test gave results on a par with the syn- 
onym-antonym test. The distribution of scores given in 
Figure 8 shows a large number of zero scores (284), but this 
number is not larger than might have been expected, con- 
sidering the amount of intelligence necessary even to 
understand the nature of the task to be performed. This 
test again shows only a partial distribution. If it had been 
easier, it would probably have shown a symmetrical dis- 
tribution. 



- 


"TEST 7. ALPHA 


250 - 




200- 




150- 


- 


100- 






50- 
0-- 




1 1 1 1 [ 1 1 1 1 1 1 1 1 1 1 



Figure 8. Distribution of scores of the Analogies Test. (From p. 
625, Memoir XV.) 



( sky blue : : grass table green warm Ijig 

SAMPLES < fish swims:; man paper time walks girl 

( day night : : white red black clear pure 

In each of the liiies below* the first two words are related to each other in some way. What 
you are to do in each line is to see what the relation is between the first two Words, and under- 
Kne the word in heavy type that is related in the same way to the third word. Begin with 
No. 1 and mark as many sets as you can before time is called. 

1 shoe foot :: hat kitten head knife penny 1 

2 pup dog :: lamb red door sheep book... ,..,. 2 

3 spring summer :: autumn winter warm harvest rise 3 

4 devil angel : : bad mean disobedient defamed good. ,. , 4 

5 finger hand : : toe body foot skin nail 5 

6 legs frog : : wings eat swim bird nest 6 

7 chew teeth : : smell^ sweet stink odor nose 7 

8 lion roar:: dog drive pony bark harness 8 

9 cat tiger:: dog wolf bark bite snap 9 

10 good^ad : : long tall big snake short... 10 

11 giant large : : dwarf jungle small beard ugly. 11 

12 winter season : : January February day month Christmas . . 12 

13 skating winter :: swimming diving floating hole summer 13 

14 blonde light :: brunette dark hair brilliant blonde 14 

15 love friend : : hate malice saint enemy dislike 15 

16 egg bird : : seed grow plant crack germinate 16 

17 dig trench :: build run house spade bullet 17 

18 agree quarrel i : friend comrade need mother enemy 18 

19 palace king : : hut peasant cottage farm city 19 

20 cloud-burst shower ; : cyclone bath breeze destroy West. . . 20 

21 Washington ^Adams:: first president second last Bryan .. 21 

22 parents command : : children men shall women obey 22 

23 diamond rare:: iron common silver ore steel l . 23 

24 yes affirmative :: no think knowledge yes negative 24 

25 hour day:: day night week hour noon 25 

26 eye head : : window key floor room door 26 

27 clothes man : : hair horse comb beard hat 27 

28 draw picture :.: make destroy table break hard. 28 

29 automobile wagon : : motorcycle ride speed bicycle car 29 

30 granary wheat :: library- read books paper chairs 30 

31 Caucasian English : : Mongolian Chinese Indian negro yellow. . 31 

32 Indiana United States : : part hair China Ohio whole 32 

33 esteem despise :: friends Quakers enemies lovers men 33 

34 abide stay : : depart come hence leave late 34 

35 abundant scarce :: cheap buy costly bargain nasty 35 

36 whale large : : thundei loud rain lightnmg kill 36 

37 reward- hero : : punish God everlasting pain traitor .... 37 

38 music soothing :: noise hear distrdcting sound report. ...... 38 

39 book writer :: statue sculptor liberty picture state 39 

40 wound pain : : health sickness disease exhilaration doctor. . 40 

Plate VII. Alpha Test 7: Analogies (Form 8). 
96 



AMERICAN INTELLIGENCE 27 

In the construction of this test it was dehberately plan- 
ned to make many items very difficult by introducing as a 
wrong alternative a word that was frequently associated 
with the key word. For instance, we find in the last 20 items 
the following easily associated pairs which would be wrong : 
(21) first-last, (24) no-yes, (25) day-night, (26) window- 
door, (27) hair-comb, (28) make-break, (29) motor-cycle- 
ride, (30) library-read, (32) part-hair, (35) cheap-buy, (36) 
thunder-Kghtning, (38) noise-hear, (39) statue-Kberty, (40) 
health-sickness. The test therefore involves not only the 
selection of the right word, but the refusal to accept as the 
solution a word that is exceedingly attractive owing to fre- 
quent associations. 

The analogies test is the most effective test in the entire 
series in differentiating officers from men. For some reason, 
not understood, it does not rank high in differentiating 
feeble-minded from enlisted men. The scores show a con- 
siderable average improvement with extension of time limit 
(8.60 to 12.46) but a correlation of 0.920 between three and 
six minutes work. On the whole it is an excellent test. 



28 AMERICAN INTELLIGENCE 

Alpha Test 8. Information 

Time limit : 4 minutes 

The army information test has been criticised more than 
any other test, and it has undoubtedly received much more 
abuse than it deserves. From the standpoint of test con- 
struction it is satisfactory, for, aside from the zero scores 
probably due to illiteracy, the distribution as shown in 
Figure 9 is rather good. 



150-1 



100- 



TEiST 8, ALPHA 




{ 1 I 1 I M I I i I- I 1 I i I i M I r 
10 20 30 40 

Figure 9. Distribution of scores of the Information Test. (From 
p. 626, Memoir XV.) 

The most frequent charge made against the test is that 
a person could fail in certain items and still be intelligent. 
This is certainly true, and the criticism would be valid if 
anyone were expected to answer all the items, or if he were 
considered unintelligent if he failed. The average person 
answered less than 15 items correctly. 



Notice the sample sentence: 

People hear with the eyes ears nose mouth 

The correct word Is ears, because it makes the truest sentence. 

In each of the sentences belo^ you have four choices for the last word. Only one of thenx is cor- 
rect. In each sentence draw a line imder the one of these four words which makes the truest sentence. 
If you can not be sure, guess. The two samples are already marked as they should be. 



'I 



People hear with the eyes ears nose mouth 
SMIPLES^ "^ 

France is b Europe Asia Africa Australia 

1 The apple grows on a shrub vine bush tree 1 

2 Five hundred is played with rackets pins cards dice ,.... 2 

3 The Percheron is a kind of goat horse cow sheep , 3 

4 The most prominent industry of Gloucester is fishing packing brewing automobiles.. 4 

5 Sapphires are usually blue red green yellow ..., 5 

6 The Rhode Island Red is a kind of horse granite cattle, .fowl 6 

7 Christie Mathewson is famous as a writer artist baseball player comedian 7 

8 Revolvers are made by Swift Sc Co. Smith & Wesson W. L. Douglas B. T. Babbitt. 8 

9 Carrie Nation is known as a singer temperance agitator sufffftgist nurse 9 

10 "There's a reason" is an "ad" for a drinJt revolver flour cleanser 10 

^11 Artichoke is a kind of hay corn vegetable fodder H 

12 Chard is a fish lizard vegetable snake -,..... li 

13 Cornell Uriversity is at Ithaca Cambridge AnnapoUs New Haven. .., 13 

14 Buenos Aires is a city of Spam Brazil Portugal Argentina 14 

15 Ivory is obtained from elephants mines oysters reefs ^^ 15 

16 Alfred Noyes is f anlous as a painter poet musician sculptor 16 

17 The armadilid is a kind of ornamental shrub animal musical instrument dagger 17 

18 The tendon of Achilles is in the heel head shoulder abdomen 18 

19 Crisco is a patent medicine disinfectant tooth-paste food product 19 

20 An aspen is a machme fabric tree drink 20 

21 The sabre is a kind of musket sword cannon pistol ;.. , 21 

22 The mimeograph is a kind of typewriter copying machine phonograph pencil 22 

23 Maroon is a food fabric drink color 23 

24 The clarionet is used in music stenography ' book-bmdmg lithography , 24 

25 Denim is a dance food fabric drink , 25 

26 The author of "Huckleberry Fmn" is Poe Mark Twain Stevenson Hawthorne 26 

27 Faraday was most famous m Uterature war religion science 27 

28 Air and gasolene are mixed in the accelerator carburetor gear case differential ' 28 

29 The Brooklyn Nationals are called the Giants Orioles Superbas Indians .' 29 

30 . Pasteur is most famous in poUtics literature war science ... . 30 

31 Becky Sharp appears in Vanity Fair Romola The Christmas Carol Henry IV 31 

32 The number of a Kaffir's legs is- two four six- eight , 32 

33 Habeas corpus is a term used in medicine law theology pedagogy 33 

34 Ensilage is a term used in fishing athletics farming hunting , 34 

35 The forward pass is used in tennis hockey football golf 35 

36 General Lee surrendered at Appomattox in 1812 1855 1886 1832 36 

~^ 37 The watt is used in measuring wind power rainfall water power electricity 37 

38 The Pierce Arrow car is made in Buffalo Detroit Toledo Flint ,. 38 

'39 Napoleon defeated the Austrians at Friedland Wagram Waterloo Leipzig 39 

40 An irregular four-sided figure is called a ,5cholium triangle trapezium peirtagon 40 

Plate VIII. Alpha Test 8: Information (Form 8), 



1 ^ 



30 AMERICAN INTELLIGENCE 

The test was devised to sample as many fields of infor- 
mation as it was possible to sample with 40 items. In gen- 
eral the five information tests in the five forms of alpha 
sampled similar fields. For instance, the advertising slogans 
which appear in the five forms are "Hasn't scratched yet" 
(Form 5), "The makings of a nation" (Form 6), "Even- 
tually, why not now?" (Form 7), "There's a reason" (Form 
8), and "The flavor lasts" (Form 9), while the Overland, 
Buick, Rolls-Royce, Pierce Arrow and Packard appear in 
each of the five forms. 

Information tests vary considerably in construction. 
There is a great difference between asking for the date of 
Lee's surrender and asking a person to choose between the 
four dates, 1812, 1865, 1886 and 1832. And again, a person 
is merely asked to elect whether the bassoon, xylophone, 
cymbal, clarionet and piccolo, appearing in each of the ^ve 
forms of alpha, should be used in music, stenography, book 
binding, or lithography. Approximately a third of the times 
test for vocabulary rather than information in the literal 
sense. If a person, for instance, knows what a Zulu, or a 
Korean, or a Hottentot, or a Kaffir or a Papuan is, he very 
obviously knows the number of his legs. 

As a rule women object to the information test more than 
men because the test samples rather heavily the fields of 
sport, mechanical interests, etc. The chances are that this 
test would penalize women rather heavily, but as a general 
rule the results of comparing the two sexes on alpha as a 
whole at various colleges show very slight differences in 
favor of the men. The sex differences found are not large 
enough to be significant. 

At Camp Lee a group of 164 captains and 200 enlisted 
men of the same general intelligence level (i. e. A and B) 
were examined. The test showing the greatest differentia- 
tion between the two groups was the information test. The 



AMERICAN INTELLIGENCE 31 

only other test showing a difference in favor of the captains 
was Test 4 (synonym-antonym test), while the test for arith- 
metical reasoning (Test 2) and practical judgment (Test 3) 
showed differences in favor of the A and B enlisted men. 
The differences were, of course, small, but the greatest dif- 
ference was shown by the information test. In differentiat- 
ing between officers and the general run of enlisted men, 
the information test was fairly effective, and it was very 
nearly as good as the arithmetical reasoning and synonym- 
antonym test in differentiating between feeble-minded in- 
dividuals and enlisted men, in spite of the fact that the 
feeble-minded obtained a somewhat higher percentage of 
their total score from this test than did the enlisted men. 
After weighing all the evidence, it would seem that we 
are justified in ignoring most of the arm-chair criticisms of 
this test and in accepting the experimental evidence tend- 
ing to show that the test was a fairly good one. The assump- 
tion underlying the use of a test of this type is that the 
more intelligent person has a broader range of general in- 
formation than an unintelligent person. Our evidence shows 
that this assumption is, in the main, correct. 



SECTION II 
EXAMINATION BETA 

When we turn to examination beta, we meet an en- 
tirely different problem, that of testing the inteUigence of 
wholly or partially illiterate persons who could not take 
alpha on account of their language handicap, of testing non- 
English speaking persons some of whom knew only the 
simplest commands in English, and low grade individuals 
who did not have sufficient intelligence to make a substan- 
tial score on alpha. At the time of the first try-out of the 
army tests in the fall of 1917 at four cantonments, Devens, 
Dix, Lee and Taylor, examination a, the fore-runner of 
alpha, was in use, and various types of individual examina- 
tions were being tried out, but there was no non-verbal 
group test. To meet this need, a preliminary try-out of fif- 
teen tests was made early in 1918, and a final examination 
composed of seven tests was subsequently published and 
widely used throughout the country. In the following pages 
the seven beta tests are reproduced in Plates 9 to 15, the 
method of administering them is described, and the results 
from the experimental group of 1047 cases are presented. 
Examination beta was given under the most rigid experi- 
mental conditions. The experimenter stood on a platform 
back of which was a large black-board on which small dup- 
licates of the seven tests could be shown one at a time. A 
demonstrator, whose duties were to act out the test prob- 
lems on the black-board, was an essential part of the ex- 
periment. The experimenter in pantomime showed the de- 
monstrator what to do on the black-board, then, after his 



AMERICAN INTELLIGENCE 



33 



performance was completed, orderlies throughout the room 
explained to the men that they were to go ahead with the 
test and do what the demonstrator had done. The orderlies' 
vocabulary was limited to "Yes," "No," "Sure," "Good," 
"Quick," "Hurry up," "How many?" "Same," "Do it," 
"Fix it." The experimenter used just as few words as pos- 
sible, and acted out every spoken sentence by pointing, 
motioning, etc. The demonstrator never spoke. His duties 
consisted of doing before the group just what the^group 
was expected to do with the examination blanks. 



Beta Test 1. Maze 
Time limit : 2 minutes 

The black-board was turned so that two sample mazes 
as shown in Figure 10 appeared. The experimenter traced 



TEST 1 



i-j^j=4rj^- 



1 ^ 


^ 



Figure 10. Black-board chart for demonstrating the Maze Test. 



34 



AMERICAN INTELLIGENCE 



through the first maze on the black-board, and then mo- 
tioned the demonstrator to go ahead. The demonstrator 
traced through the maze with crayon very slowly. The ex- 
perimenter then traced through the second maze and mo- 
tioned the demonstrator to go ahead. The demonstrator 
in tracing this maze made a mistake by crossing the line 
at the end of a blind alley, was corrected by the experi- 
menter with vigorous shakes of the head and "no-no," and 
made to re-trace his path back to where he could start right 
again. The demonstrator then traced through the rest of 



- 


TEST 


1, BETA 








200- 










150- 










: 














100- 




J"^ 




- 


50- 


y 




U "" 

( 


,' ' ' 


1 1 i 

5 


-^ 


-T 




1 






Figure 11. Distribution of scores of the Maze Test. (From p. 627, 
Memoir XV.) 



5 



T r 



U 



u > 



b 



?^ 



lAl 



3 



u 



ra 



5 



4 -^ -. 



B 



n 



u 



~i 



H 



u 



B 



U 



C 



Q 



^1 



u 



PZa^e IX. Beta Test 1 : Maze. 
35 



36 



AMERICAN INTELLIGENCE 



the maze with great semblance of haste, stopping momen- 
tarily at each ambiguous point only. The experimenter then 
motioned to the group to do the same thing on their exam- 
ination blanks. The experimenter and the orderlies walked 
about the room, motioning to the men who were not work- 
ing, and saying, "Do it, do it, hurry up, quick." 

The results of the maze test are shown in Figure 11. The 
difference' between the distribution of scores of this test 
and the alpha tests is remarkable. In the first place, our 
large number of zero scores has disappeared only 19 in 
our group of 1047 failed to make any score. In the second 
place, the test is entirely too easy, for it is apparent that 
the men in the upper end of the scale could have done more 
in the time allowed. The maze test was intentionally made 
easy in order to get everybody started. We have at last 
found a test in which practically everybody can do some- 
thing. Aside from the language involved, every test in alpha 
is harder than this beta maze test, for no alpha test has less 
than 7% zero scores. 



Beta Test 2. Cube Analysis 
Time limit : %}/2 minutes 

The black-board was turned to show a series of cubes 
like that in Figure 12. On a shelf was a real three cube 

TEST 2 



^ 



U 



U 



u 



u 



Figure 12. Blackboard chart for demonstrating the Cube Analy- 
sis Test. 



AMERICAN INTELLIGENCE 



37 



model similar to the first one on the black-board. The ex- 
perimenter pointed to the three cube picture on the black- 
board, then to the model on the shelf, then to the picture 
on the black-board, and asked, "How much?" The experi- 
menter then counted aloud, putting up his fingers while 
counting, and encouraged the men to count with him. The 
experimenter then tapped each cube on the black-board 
and asked the demonstrator, "How much.f^" The demon- 
strator then went to the black-board, counted the cubes by 
pointing, and wrote the number 3 in the space below the 
illustration. A similar performance was enacted for the 
other three problems on the ^black-board, the models being 
shown and elaborately counted. 

The distribution of scores of the cube analysis test is 
shown in Figure 13. Here we find a somewhat larger num- 



100- 



50- 



TEST 2, BETA 




Figure 13. Distribution of scores of the Cube Analysis Test. 
(From p. 627, Memoir XV.) 



ber of zero scores (54) than in the maze test, but a fairly 
good distribution in general. On the whole the test is easy. 



m 



y ^ y A 



I ffl 



y y ./ 







^5f1 










w 





m 



Hi 






Plate X. Beta Test 2 : Cube Analysis. 



AMERICAN INTELLIGENCE 39 

Beta Test 3. X-0 Series 

Time limit : 1% minutes 

The black-board was turned to show the chart repro- 
duced in Figure 14. The experimenter traced with a pointer 



TES T 3 

goioioioi 1 1 1 1 



ixi ixi ixi ixi 1 1 n 



loioixioiohdoioixi 1 1 1 1 



ixixixioixiqxixixlolxl 



Figure 14. Black-board chart for demonstrating the X-0 Series 
Test. 



each "O" in the top chart, and then wrote (with his pointer) 
an imaginary "O" in the four remaining spaces. The dem- 
onstrator then filled in the four "O's" with crayon. The 
experimenter then traced the first "X" by tracing a semi- 
circle above the chart and so on. The demonstrator then 
filled in every other space with an "X" following the ex- 
perimenter's elaborate exercise. The demonstrator then 
worked out the remaining problems with the same ritual, 
following which the men were instructed to go ahead. 

The distribution of scores of the X-0 series test is shown 
in Figure 15. 

The army writers state that: "Beta 3 defies interpreta- 
tion." (p. 638.) We know that the test was devised to dup- 
licate in pictorial form the number series completion test 



40 



AMERICAN INTELLIGENCE 



150- 



100- 



50 



TEST 8, BETA 



Uu 



I I I I i I I I ,1 I I 

5 10 

Figure 15. Distribution of scores of the X-0 Series Test. (From 
p. 627, Memoir XV.) 



in alpha. Aside from knowing the purpose of the test, we 
know very little about it. My own guess would be that the 
first five or six items were entirely too easy, and that if 
they had been disregarded in the scoring, as practice items, 
and six more items added, comparable in difficulty to the 
last six items, the distribution would have been satisfactory. 



1 


E 


E 


F 


F 


Ixlxlxl 


r 


r 


r 


r 


1 










2 


^ 




X 




X 




X 




X 






r 


M 






S 


X 





X 


o 


X 





X 





X 


o 










4 








X 


X 




K 


K 




^ 


X 




X 


X 










1 






5 


X 


o 




X 


o 




X 


o 




X 



















6 


X 


X 





X 


X 





X 


X 





X X 


o 


















7 





o 


Xi 


X 





OXX 


o 


OXX 


1 









8 




x]x 











XX 


o 


H 





X 


X 


O 


o 










9 


X 


ox xox x.x No 


Ai 11 


LG 




xHo|xl<,|x|xloU| 


^ 


i^ 


2^ 


d 


X 


1 


_ 






J 



U |X|0|X|X|0|X|X|X|<>|X|0|X|X|0|X|^|X|C| I I I I I I I I I 



lg |X|X|X|X|o|o|o|x|X|.|x|X|x|X|.|o|o|x|x|o| I I I I I M I I I 
Plate XI. Beta Test 3 : X-O Series. 



41 



42 AMERICAN INTELLIGENCE 

Beta Test 4. Digit-Symbol 
Time limit : 2 minutes 

The black-board was turned to the chart shown in Fig- 
ure 16. The experimenter pointed to each number and then 



TEST 4 




m 

u 





3ll2|3|2l|2|l|3|4|7 5l4lte^ 


III IN 



Figure 16. Black-board chart for demonstrating the Digit-Symbol 
Test. 



to the symbol under it. The experimenter then pointed to 
the number 3 in the sample, then to the space below it, then 
to the number 3 in the key above, then to the symbol for 
3, and finally traced the outline of the symbol for 3 in the 
proper space in the sample. This procedure was then re- 
peated for the first ^yq samples. The demonstrator then 
went to the black-board, and worked through the process 
of filling in the symbols under the figures, touching each 
figure and symbol in the key while he drew the proper sym- 
bol in the sample. The group then proceeded to fill in the 
symbols on the test blank. 

The distribution of scores of the digit-symbol test is 
shown in Figure 17. 



AMERICAN INTELLIGENCE 



43 



100- 



50- 



TEST 



4. ri BETA 



^ M i M 

10 




TT-r 

20 



Figure 17. Distribution of scores of the Digit-Symbol Test. (From 
p. Q'^S, Memoir XV.) 



This type of distribution is the same as that given by the 
eight alpha tests, the zero scores representing failure to 
understand instructions, and the distribution being fairly 
regular. The digit-symbol test is a standard test, and the 
results in beta are entirely satisfactory. 



J^ 




2^ 




3 

2 




4^ 





U 




j8^ 





2 

A 




zl 


J9_ 










3 


1 


2 


1 


3 2 


1 


4 


2 


3 


5 


2 


9 


1 4 




1 






































6 


3 


1 


5 


A S 


\ 7 


6 


3 


8 


7 


2 


9 


5 4 


- 


2 


































6 


3 


7 


2 


8 1 


L 9 


5 


8 


4 


7 


3 


6 


9 5 


- 


3 


































1 


9 


2 


8 


3 7 


' 4 6 


6 


9 


4 


8 


5 


7 e 


- 


4 


































"9^ 


3 


8 


6 


4 : 


I 5 


7 


2 


6 


^ 


4 


8 


ik 


\ 


6 




















L 






1 










4 


9 


5 


1 


7 ( 


> 2 


6 


9 


^ 


7 


8 


4 


1 I 


I 


6 












J_ 




_ 


L 







_ 




^ . 



Plate XII. Beta Test 4: Digit-Symbol. 



44 



AMERICAN INTELLIGENCE 45 

Beta Test 5. Number Checking 

Time limit : 3 minutes 

The black-board was turned to the chart shown in Fig- 
ure 18. The experimenter pointed first to the 6 in the left 

TEST 5 

62 62 

59 56 

327 327 

249 249 

1536 1536 

3745 3745 

45010 45001 

62019 62019 

Figure 18. Black-board chart for demonstrating the Number 
Checking Test. 



hand column, then to the 6 in the right hand column, then 
to the 2 in the left hand column and to the 2 in the right 
hand column, nodded his head, said, "Yes," and made an 
imaginary cross on the dotted line. The demonstrator then 
made an "X" on the line. The experimenter repeated the 
procedure for the second pair, but indicated clearly, by 
shaking his head and saying, "No," that the 9 and the 6 
were not alike. The experimenter then repeated the pro- 
cedure for three more sets, getting the men in the room to 



46 



AMERICAN INTELLIGENCE 



say, "Yes," if the pairs were identical. The demonstrator 
then worked out the remaining items. 

The results of the number checking test shown in Fig- 
ure 19, give the same distribution characteristic of the 



TEST 5, BETA 




Figure 19. Distribution of scores of the Number Checking Test. 
(From p. 628, Memoir XV.) 



alpha tests. The instructions were clear and the test was en- 
tirely satisfactory. This test is an adaptation of a standard 
test in use for many years. 



650 650 

041 044 

2579 2579 

3281 3281 

55190 55102 

39190 39190 

658049 : . . . 650849 

3295017 3290517 

63015991 63019991 

39007106 , 39007106 

69931087 69931087 

251004818 251004418 

299056013 299056013 

36015992 360155992 

3910066482 391006482 

8510273301 8510273301 

263136990 263136996 

451152903 451152903 

3259016275 3295016725 

582039144 582039144 

61558529 61588529 

211915883 219915883 

670413822 670143822 

17198591 , . . . 17198591 

36482991 36482991 



10243586 10243586 

659012534 6590211354 

388172902 381872902 

631027594 631027594 

2499901354 2499901534 

2261059310 2261659310 

2911038227 2911038227 

313377752 313377752 

1012938567 1012938567 

7166220988 7162220988 

3177628449 3177682449 

468672663 468672663 

9104529003 9194529003 

3484657120 3484657210 

8588172556 8581722556 

3120166671 3120166671 

7611348879 76111345879 

26557239164 26557239164 

8819002341 8819002341 

6571018034 6571018034 

38779762514 38779765214 

39008126557 39008126657 

75658100398 . . 75658100398 

41181900726 41181900726 

6543920817 6543920871 



Plate XIII. Beta Test 5 : Number Checking. 



47 



48 AMERICAN INTELLIGENCE 

Beta Test 6. ^Picture Completion 

Time limit : 3 minutes 

The black-board was turned to the chart shown in Fig- 
ure 20. The experimenter pointed to the hand and said, 



TEST 6 



/f 




cy 



^ 



Figure 20. Black-board chart for demonstrating the Picture Com- 
pletion Test. 



"Fix it." The demonstrator looked puzzled. The experi- 
menter pointed to the place where the finger was missing, 
and said, "Fix it; Rx it." The demonstrator then drew the 
finger. The experimenter then pointed to the fish, and the 
place for the eye, and said,"Fix it." After the demonstrator 
had drawn in the eye, the experimenter pointed to each of 
the drawings and said, "Fix them all." After the demon- 
strator had worked out all the remaining drawings the 
group proceeded to complete the drawings in the beta blank. 
The results of the pictorial completion test as given in 
Figure 21 show an excellent distribution of the same general 
type as the distribution of the eight alpha tests. The num- 
ber of zero scores (12) on this test is smaller than that of 
any other test in the entire alpha-beta series. This is an 
excellent test. 



100- 



TEST 6, BETA 




Figure 21. Distribution of scores of the Picture Completion Test. 
(From p. 628, Memoir XV.) 



49 








V.y. 






Plate XIV. Beta Test 6: Picture Completion. 
SO 



AMERICAN INTELLIGENCE 51 

Beta Test 7. Geometrical Construction 

Time limit : %}/2 minutes 

The black-board was turned to the chart shown in Fig- 
ure 22. The experimenter pointed to the square on the 

TEST 7 



X7 



^n 



cQn 



u 






Figure 22. Black-board chart for demonstrating the Geometrical 
Construction Test. 



black-board, and taking two pieces of cardboard the same 
size as the drawings at the left of the square, fitted them 
on the two drawings. He then fitted the two pieces of card- 
board together on the square to show that they would fill 
it, and motioned to the demonstrator who drew, in the 
square, the lines indicating the manner in which the two 
pieces would fit. This procedure was repeated for the next 
two samples. The demonstrator then worked out the last 
sample. 

The responses given in Figure 23 show, aside from the zero 
scores, a peculiar distribution, the form of which may be 
interpreted by assuming that the test was too hard in its 



52 



AMERICAN INTELLIGENCE 



BETA 



100- 




50- 



Figure 23. Distribution of scores of the Geometrical Construction 
Test. (From p. 628, Memoir XV.) 



beginning and too easy at the end. We can picture the dis- 
tribution of Test 7 as shown in Figure 23 as having come 
from the middle range of a more complete test such as 
Test 5, Figure 19. If we cut the distribution of Test 5 from 
10 to 19, we can picture the range in which Test 7 was 
working. We may assume, then, that the inclusion of a few 
easier items and five or ten harder ones might have given 
a distribution similar to that of the alpha tests. The test is 
faulty because of the limitation of range at both ends. 





8 




^17 




D 
D 



6 



^ 




O 



8 



d 
^ 



9 




10 



D 



Plate XV. Beta Test 7 : Geometrical Construction. 
53 



SECTION III 



THE INDIVIDUAL EXAMINATIONS 

The greatest contribution of the army psychologists to 
the development of mental tests was that of creating the 
two group tests, alpha and beta, that have been discussed. 
Methods of individual examining had been in existence for 
several years. The basic series of tests of the individual ex- 
amination was the Stanford Revision of the Binet-Simon 
scale, which had become a standard measurement and 
needs no description here. Persons interested in this method 
should read Terman'si book on the Stanford scale. A 
method of abbreviating the Stanford-Binet scale was 
worked out in the army, and proved satisfactory. 

The distribution of the scores in terms of the "mental 
ages" of the 653 men in the special experimental group of 
1047 cases who took the Stanford-Binet examination was 
as follows: 

"Mental Ages" 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 

No of cases 1 2 122 62 66 69 8169 77 63 54 47 34 5 

A rough inspection of these figures shows that they give 
us the Gaussian normal distribution. The results obtained 
from the Stanford-Binet examination may be taken as en- 
tirely reliable without question. 

One difficulty in the popular interpretation of the results 
on the Stanford-Binet scale, and other scales constructed 
on the same principle, is the unfortunate use of the term 
"mental age," a term first used by Binet and subsequently 

^L. M. Terman. The Measurement of Intelligence. Boston, 1916, Pp. 362. 

54 



AMERICAN INTELLIGENCE 55 

used in this country. The term "mental age" has no signi- 
ficance whatsoever aside from the particular scale from 
which it was derived. A person might have a "mental age" 
of 13 on the Stanford-Binet scale, of 11 on Goddard's 
translation of Binet's 1908 scale, of 12 on Goddard's 1911 
scale, and so forth for every scale in use. The term "mental 
age" really means a score on a particular series of tests. 
Through rather general usage, the Stanford-Binet scale is 
being adopted in this country as a standard. 

The Stanford-Binet scale was constructed out of some 
90 different tests arranged for different age levels, six for 
each age level from 3 to 10, eight for 12, six for 14, six for 
16 or "average adult," six for 18 or "superior adult," and 
sixteen alternative tests interspersed throughout the scale. 
A person obtains his total score or "mental age" by taking 
all the tests in perhaps four or five age level groups. For 
instance, if a person passed all the tests at the 9 year level, 
five out of the six at the 10 year level, four out of the eight 
at the 12 year level, two out of the six at the 14 year level, 
and failed all tests above 14, his total score or "mental age" 
would be llj^. In assigning a given test to any age level, 
all the tests were first tried out, and the positions of the 
tests juggled about so that the ten year old children tested 
had an average score of 10, the eleven year old children an 
average score of 11, etc. 

When we say that a person has a "mental age" of eight 
on the Stanford-Binet scale, we do not mean that he has 
the mentality of a child of eight, but that he made a total 
score on that scale equal to that of the average eight year 
child tested in the particular group on which the scale was 
standardized. In all, about 1000 children, approximately 
equally distributed in the chronological ages from 5 to 14, 
formed the basis of the Stanford standardization. This 
standardization is a very excellent method of measuring 



56 AMERICAN INTELLIGENCE 

intelligence, and worked very well with our army adult 
group as shown by the score distribution of the 653 men 
in the special experimental group given above, but we 
should always regard the term "mental age" as a score, 
not as a diagnosis. 

By correlating the alpha test with the Stanford-Binet 
scale, we can find the approximately equivalent score, or 
"mental age" for each possible alpha score. The operation 
resembles that of expressing values sterling in dollars. One 
frequently hears the statement that the army tests proved 
that the average citizen of this country has a mental age 
equivalent to that of a child of thirteen. Nothing could be 
more ridiculous. It is true that the average score of a sample 
of 93,955 soldiers representing the entire white draft, when 
translated into the Stanford-Binet scale, is 13.14. This 
means that the approximately equivalent score on the 
Stanford-Binet is 13.14. To say that the average citizen 
has a mentality of the child of 13 is putting the cart before 
the horse, for we are grading 93,955 people, and by infer- 
ence the entire country, on a standard fixed by some 82 
fourteen year old children who happened to be tested in 
California. 

In addition to the 1000 children on whom the Stanford- 
Binet scale was standardized, the tests at the 16 year 
"average adult" level, and the 18 year "superior adult" 
level were standardized on 30 business men, 150 "migrat- 
ing" unemployed men, 150 adolescent delinquents, and 50 
high school students. It is thus seen that the Stanford- 
Binet standardization rests on a number of cases too small 
to upset the army standards based on 93,955 cases. 

The term "mental age" is bad scientific slang for a total 
score. Psychologists are gradually abandoning the age 
standardization of tests. At the same time, publicists in 
various fields, although novices in psychology, are drawing 



AMERICAN INTELLIGENCE 57 

rather vicious conclusions from "mental age" findings. It 
is an unfortunate situation. 

The methods used in creating and standardizing psycho- 
logical tests are entirely empirical, and therefore rather 
hard to explain to the layman, who is familiar only with 
the "school teacher" type of examination. The school 
teacher writes an examination, and lets it stand as an ab- 
solute measure. The psychologist makes an examination, 
tries it out, and judges each individual member of the 
group as compared with the other members of the group. 
As more and more people are examined his standards of 
judgment become more reliable. In other words, his stand- 
ards are those that he gets, not those that he thinks he 
ought to get. Therefore, instead of deploring the fact that 
the average person has a "mental age" of thirteen, we can 
simply say that the conversion of the results of the army 
test into the Stanford-Binet scale shows an average score 
of 13, and that this is the score to be expected from the 
average adult. 

Another very common mis-statement prevalent con- 
cerning the army results is that they proved that 24.9% 
of the drafted men were illiterate. Among the men sent to 
examination beta would be found, first, English speaking 
illiterates, second, non-English speaking individuals, either 
literate or illiterate in their own tongue, third, defectives, 
and fourth, cases accidentally sent to the wrong examina- 
tion. The method of selecting men for beta varied from 
camp to camp, and sometimes from week to week in the 
same camp. There was no established criterion of literacy, 
and no uniform method of selecting illiterates. In a group 
of 1,552,256 men examined, 386,196 or 24.9% were, for 
some reason or other, sent to beta. The army definition of 
literacy as ^'ability to read and understand newspapers and 
write letters home'' can not be identified with the fact of having 



58 AMERICAN INTELLIGENCE 

been sent to beta. The statistics of the army examinations 
give us no accurate figures on the percentage of ilKteracy. 
The individual examination for ilHterates and non- 
Enghsh speaking, the performance scale, was a composite 
scale, the tests of which were drawn from workers in vari- 
ous fields, particularly from H. A. Knox, who had worked 
on non-verbal performance tests at Ellis Island, R. Pintner 
and D. G. Paterson, who had developed a scale of perform- 
ance tests, William Healy, H. H. Goddard, and other in- 
vestigators. The performance examination was given some- 
times as the long scale (8 or 10 tests) and sometimes as the 
short scale (5 tests). The short scale showed a correlation 
of 0.97 with the long scale, so that the reduction of the 
number of tests to save time in examining was entirely jus- 
tified. A short-cut method was also used in giving the 
Stanford-Binet examination which was quite satisfactory 
(correlation 0.91). The short performance scale showed a 
correlation of 0.84 with the Stanford-Binet scale. The 
Yerkes-Bridges point scale which was sometimes used in- 
stead of the Stanford-Binet was also abbreviated, and the 
abbreviated point scale showed a correlation of 0.934 with 
the complete point scale. In general, the methods of indi- 
vidual examining in use were quite reliable, and so closely 
related to the Stanford-Binet scale that the results could 
be converted into Stanford-Binet scores without any ap- 
preciable source of error. In all calculations in this study, 
scores from the point scale and the performance scale ex- 
aminations have been treated by converting into Stanford- 
Binet scores. 



SECTION IV 

RELIABILITY OF THE MEASURES 

The reader, who has followed the discussion of the indi- 
vidual tests through the preceding pages, will be convinced 
that most of the tests used were satisfactory. In general, 
the eight alpha tests, the Stanford-Binet scale, and tests 
4, 5, 6 and 7 of beta gave complete or limited distributions 
which approximated the Gaussian normal curve. Figure 24. 




Figure 24. The normal distribution curve. A type of distribution 
given by all alpha tests^ tests 4, 5, 6 and 7 of beta^ and the Stan- 
ford-Binet scale. 



Beta test 3 gave a distribution which could not be inter- 
preted, while beta tests 1 and 2, being too easy, gave a 
skewed distribution of the approximate form of the curve 
shown in Figure 25. 

It is not necessary here to enter into any lengthy dis- 
cussion of the method of converting the results of these 
sixteen tests into the combined scale. The reader interested 
in the statistical methods used is referred to Chapter 2, 
Part 3 of Memoir XV (pp. 573-657). If the reader is satis- 

59 



60 AMERICAN INTELLIGENCE 

fied that 13 of the 16 tests give distributions conforming 
in a general way to that shown in Figure 24, that is all that 
is necessary. The tests can very obviously be equated, and 
a combined scale constructed. By treating each of the eight 
tests of alpha, each of the seven tests of beta, and the 
Stanford-Binet test as different measuring scales, the com- 
bined scale was evolved, based on the inter-relations of 
these sixteen scales as shown by refined methods of cor- 
relation. 




Figure 25. A skewed distribution curve. A type of distTibution 
given by beta tests 1 and 2. 



The army results are reported in tables showing the 
number of men scoring in certain class intervals, i. e. be- 
tween and 4, 5 and 9, 10 and 14, etc., up to the interval 
205 to 212 on alpha; between and 4, 5 and 9, 10 and 14, 
etc., up to 115-118 on beta. In the same way, the scores in 
other tests are reported in class intervals. The study of the 
1047 cases showed how individuals falling in each of the 
class intervals were distributed on the theoretical com- 
bined scale, i. e., it was possible to find the combinations 
of tests from which individuals in each class interval would 



AMERICAN INTELLIGENCE 61 

obtain their scores. The combined scale was therefore built 
empirically on the results of the 1047 cases. Tables were 
constructed on this basis showing how individuals falling 
in each class interval of each of the three examinations 
should be redistributed on the combined scale. It is then 
possible to take a group which had been examined partly 
by alpha, partly by beta, and partly by the Stanford- 
Binet examination, and determine how that group would 
have scored on the combined scale if all individuals in the 
group had been given all three examinations. The com- 
bined scale is the most accurate method available for treat- 
ing the data derived from the army examinations. 

In this study the data from the principal sample have 
been re-figured on the combined scale by the method de- 
scribed on page 65^ of Memoir XV: 

"In each group the alpha distribution was distributed 
on the combined scale by the use of table 159, the beta dis- 
tributions by table 162, and the Stanford-Binet mental age 
distribution by table 163. The performance scale distribu- 
tions and the point scale distributions were handled in the 
following way: the performance distributions were first 
transformed into Stanford-Binet mental age distributions 
by the use of the regression formula : 

T.yr i 1 A / N 0.50 Performance score + 72 
Mental Age (m years) = 

This formula was derived from the correlation of a sample 
of 350 cases who had both Stanford-Binet mental age rat- 
ings and performance scale ratings. The point scale distri- 
butions were transformed into Stanford-Binet mental age 
distributions by the use of the table in the examiner's guide. 
Part I, pages 195ff. These transformations only approxi- 
mate the truth, but owing to the fact that the performance 



62 AMERICAN INTELLIGENCE 

and point scale cases constitute less than 3 per cent of any 
group handled it would take a considerable error in trans- 
formation seriously to affect the whole." 

The conversion of the data of the principal sample into 
the combined scale reported in Memoir XV contains some 
inaccuracies. The statistical labor involved in the evolution 
of the combined scale was so great that the method was not 
available until the report was practically completed. The 
calculations were made by different individuals working 
under pressure, and errors were unavoidable. It has there- 
fore been considered worth while to repeat these calcula- 
tions at leisure, checking each operation carefully and 
carrying the analysis of some of the groups of the principal 
sample further than that reported in Memoir XV. 

It is now necessary to review very briefly the results of 
checking the army mental tests against outside criteria. 
We might have a measuring scale, all elements of which 
gave perfect score distributions, and which were highly 
inter-correlated, but even then we would need outside cri- 
teria to prove that we were measuring intelligence. Enough 
material is already on hand to prove that the army tests 
were reliable measures of intelligence. In the following dis- 
cussion we will cite several instances. 

The best proof of the validity of the test series comes 
from a study of the relation between the inteUigence rat- 
ings and education. The correlation of the combined scale 
with reported school grade was 0.75 (based on 653 cases 
from the special experimental group of 1047 men). The 
correlation between alpha scores and schooling for this 
group was 0.75, the eight tests of alpha separately com- 
pared with schooling all showing correlations between 0.60 
and 0.74. The correlation with beta total scores and school- 
ing for this same group was 0.67, and that between Stanford- 
Binet scores and schooling 0.65. These correlations show 



AMERICAN INTELLIGENCE 63 

a positive relationship between intelligence as measured by 
the various methods and years of schooling. 

Very few people realize the severity of the elimination 
process that goes on from year to year in our schools and 
colleges. The study of the schooling of the native born white 
draft, as sampled by upward of 80,000 cases, showed the 
following startling facts: of every thousand native born 
recruits who entered the first grade, 970 remained in school 
till grade two, 940 till grade three, 905 till grade four, 
830 till grade five, 735 till grade six, 630 till grade seven, 
and 490 till grade eight; 230 of them entered high school, 
170 kept on till the end of the second year, 120 till the end 
of the third year, and 95 of the original thousand graduated 
from high school; 50 of these entered college, 40 kept on 
till the end of the second year of college, 20 till the end of 
the third year, and 10 graduated. It is, of course, impossible 
to determine how many of those that leave school leave on 
account of lack of pecuniary opportunity, or on account 
of lack of intelligence. It is ridiculous to assume that 1000 
men in 1000 have sufficient intelligence to graduate from 
college, and equally absurd to assume that only 10 in 1000 
have such a high intellectual endowment that they can 
graduate from college. But, inasmuch as we can not deny 
the intellectual elimination, we must expect a very high 
correlation between intelligence and schooling. 

The army tests uniformly show officers superior to en- 
listed men. This is to be expected, for officers were selected 
for ability. Nevertheless, one may perhaps contend that 
the high scores of the officers were due to superior educa- 
tion and not to greater intelligence. Very nearly half of 
the officers were college graduates, and another quarter had 
begun but not completed a college course. The objection 
that the superior scores of officers were due to education 
rather than intelligence may be effectively answered by a 



64 AMERICAN INTELLIGENCE 

crucial test, which was made by the army investigators 
when they compared the alpha scores of 660 officers who 
had never gone beyond the eighth grade in school with the 
alpha scores of 13,943 native born recruits all of whom had 
gone beyond the eighth grade. The results of this comparison 
are reported (p. 779 of Memoir XV) as follows: "Every 
recruit in the recruit group has had more schooling than 
any officer in the officer group; the least educated recruit 
in the group has had a longer education than the best edu- 
cated officer included. And the group of officers neverthe- 
less makes a slightly higher record on examination alpha. 
It is evident then that the examination is measuring other 
qualities, in which officers stand above recruits, to a 
greater extent than it is measuring education." The distri- 
butions of the alpha scores of these two groups are shown 
in Figure 26. In general, the comparison of the army test 
scores with education indicates that the tests are genuine 
measures of intelligence. 

The army investigators were, of course, called upon early 
in the war to prove that the tests they recommended were 
genuine tests of intelligence. For the assistance of army 
examiners and administrative officers having before them 
the problem of the assignment of men, a small pamphlet. 
Army Mental Tests (Washington, D. C, 1918, pp. 24) 
was prepared, presenting in graphic form the results of 
several different methods used in some of the camps for 
establishing the reliability of the tests by checking them 
against outside criteria. In the following pages, some of the 
charts from this booklet have been reproduced, and the 
method of interpreting the charts is described briefly. All 
of the methods reported use the letter grade classffication 
(A, B, C, etc.), which is less accurate than the combined 
scale method used in this study, but they tend to show in 



OFFICERS 

WHITE DRAFT. NATIVE BORN 




Figure 26. Examination alpha as independent of education. Com- 
parison of alpha scores of officers of eighth grade schooling or 
less with alpha scores of native born white recruits of ninth 
grade schooling or more. "Although these groups overlap in 
schooling not at all, the officers make nevertheless slightly higher 
scores on alpha." (Quotation from p. 779, and figure from p. 
778, Memoir XV.) 



65 



66 AMERICAN INTELLIGENCE 

a general way the trend of the results, and that is all that 
is necessary in this ease. 

In the long run, we should expect a small positive cor- 
relation between intelligence and rank. Intelligence is by 
no means the sole determiner of military success, but since 
it is one element in the complex of abilities required, we 
would expect to find a general tendency toward high scores 
with higher ranks. Figure 27, which is reproduced from the 



ENLISTED MEN (18792) Relatively Illiterate 
ENLISTED MEN (82936) Litebate 
CORPORALS (4023) 

SERGEANTS (8393) 

O. T. C. (9240) 

OFFICERS (8819) 




Figure 27. Distribution of intelligence scores according to rank. 
The officers are above the candidates in the Officers' Training' 
Camps (O. T. C), the candidate officers are above the sergeants, 
the sergeants above the corporals, and the corporals above the 
enlisted men. (From p. 8 of pamphlet. Army Mental Tests.) 



booklet referred to, shows the distribution of scores of vari- 
ous ranks on the rough A, B, C scale. The oflScers form a 
group quite distinct from the general run of enlisted men, 
and they are also above the candidates for commissions in 



AMERICAN INTELLIGENCE 



67 



the OflScers' Training Camp (O. T. C.) group. The sergeants 
are above the corporals, and the corporals above the en- 
listed men. 

The Officers' Training Camps give an additional check 
on the intelligence tests. In the schools examined, the can- 
didates were recommended for a period of special training 
for commissions by the regimental organizations. The selec- 
tion of the candidates was very rigid, then, in the first in- 
stance. Figure 28 shows roughly the results of applying the 



A B C+ C C D 



PERCENT 
SUCCESS 



PERCENT 
FAILURE 



O. T. C. 
1375 MEN 



Figure 28. Success in Officers' Training Camps as predicted by 
examination alpha. Each vertical bar represents all (100%) of 
the candidates who tested A, B, C-|-, etc. All men above thehori- 
zontal line eventually received commissions, and all men below 
failed. 91^% of the men above C-\- received commissions. 
58^^ of the men below C-j- failed. (From p. 9 of pamphlet, 
Army Mental Tests.) 



68 AMERICAN INTELLIGENCE 

army tests to the training camp groups. Each solid vertical 
bar represents all the men of a given letter grade. There 
were, of course, more A and B men than C and D men, 
but for purposes of comparison each letter group is treated 
as a whole (100%). All men above the horizontal line re- 
ceived commissions at the close of the Officers' Training 
Camp, and all the men below the line failed to receive com- 
missions. Figure 28 shows clearly that about nine out of 
ten A and B men eventually received commissions, while 
for C and D men the chances were very slight. 

Figure 29 shows in a general way the manner in which 
groups selected by various outside criteria contributed to 
the upper end of the intelligence scale (A and B), and to 
the lower end of the scale (C , D and E). We are already 
familiar with the differences between ranks shown in this 
figure. Sixty company commanders were asked to designate 
their ten "best" and ten "poorest" privates. The results of 
this comparison of the ten "best" and ten "poorest" pri- 
vates also appears in Figure 29. The other two classifica- 
tions, "men of low military value," and "unteachable men" 
represent a type of officers' rating. In general the test re- 
sults check with officers' ratings independently made. 

The results presented, showing the relation between rank 
and intelligence, and officers' ratings and intelligence, indi- 
cate clearly a certain positive relationship between tests 
and military success. Recognizing the fact that intelligence 
is only one factor tending to produce military success, we 
accept the results of checking the tests against military cri- 
teria as additional proof that the tests are genuine measures 
of intelligence. 

A rough but rather interesting check of the army in- 
telligence tests may be made by glancing at the scores of 
men classified by occupations. Figure 30 gives the range of 
the intelligence scores of the middle 50% of various occu- 



D.D . E C+,C, C 



^ AND B 

COMMISSIONED OFFICERS 

8819 
I 1 



O. T. S. STUDENTS 
9240 



SERGEANTS 
3393 



CORPORALS 
4093 



TEN BEST PRIVATES 

606 



WHITE RECRUITS 

77299 



DISCIPLINARY CASES 
491 Camp Bix 



TEN POOREST" PRIVATES 

582 

I I e 



MEN OF LOW MILITARY VALUE'*^ 

*147 Camp Custer 
t I 

UNTEACHABLE MEN 

255 Camp Hancock 

rzZD 



Figure 29. Comparison of army tests records with various inde- 
pendent criteria. The distributions of scores by ranks are shown 
in another way in Figure 27. The men were rated as "ten best," 
"ten poorest," "of low military value," and "unteachable" by 
their officers. The chart shows a close correspondence between 
the brief psychological examinations and officers' judgments 
made after weeks of observation. (From p. 10 of pamphlet, 
Army Mental Tests.) 



> - I 



I c. I 



c-< 



Laborer . . 
Gen. miner 
Teamster . 
Barbe* ... 



C < 



Horseshoer ..... 
Bricklayer 

Cook 

Baker 

Painter 

Gen. blacksmith . 

Gen. carpenter . . 
Butcher 



Gen. machinist 

Hand riveter 

Tel. and tel. lineman , . . 

Gen. pipefitter 

Plumber 

Tool and gauge maker. 

Gunsmith 

Gen. mechanic 



Gen. auto repairman . 
Auto engine mechanic 

Auto assembler 

Ship carpenter 

Telephone operator . . 



C < 



Concrete const, foreman 
Stock-keeper 

Photographer 

Telegrapher 

R.R. clerk 

Filing clerk 

Gen. clerk 

Army nurse 

Bookkeeper 



B < 



Dental officer 

Mechanical draftsman 

Accountant 

Civil engineer 

Medical officer , , 



A ^ Engineer officer 



fD^T 



1 cT-r 



Figure 30. Success in civil occupations compared with army test 
records. The figure shows in a general way the correlation be- 
tween intelligence as measured by the army tests and intelli- 
gence as indicated by position in civil life. (From p. 829, 
Memoir XV.) 

70 



AMERICAN INTELLIGENCE 71 

pational groups, the position of the man half way up or 
down the scale being marked by a short vertical line. In 
some of the occupational groups, the number of cases is 
small, and the classification itself may be at fault in many 
instances, but the chart nevertheless shows a general ten- 
dency of the sort we should expect to find, for a process of 
intellectual selection occurs in industry which is just as- 
rigid as that occurring in our public schools. 

We have briefly inspected the different sorts of evidence 
from independent fields which indicate that the army tests 
were genuine measures of intelligence. Further discussion 
of this point is unnecessary. The army tests were not in- 
fallible, and mistakes in classifying men were undoubtedly 
made, but the tests were satisfactory rough measures. 
When used in comparing groups as the tests are in this 
study, their reliability is increased, for errors in measure- 
ment would tend to equalize in each group. We should ex- 
pect the same percentage of error in classifying recruits 
born in Russia as we should recruits born in Sweden. Thus 
we use the tests as general measures of group tendencies, 
and as group measurements the tests have a suflSciently 
high degree of reliability to make positive conclusions 
possible. 

In the foregoing pages the army tests have been de- 
scribed briefly, the method of treating the results from vari- 
ous examinations by the combined scale reviewed, certain 
misconceptions discussed, and a few bits of supporting 
evidence assembled. Persons interested in a further study 
of the tests should consult a little book by Yoakum and 
Yerkesi, or. Memoir XV. We may now proceed to analyze 
American intelligence by treating the psychological exam- 
inations made in the army as a mental census of the 
population of this country. 

IC. S. Yoakum and R. M. Yerkes. Army Mental Tests. New York, 1920, Pp.203. 



PART II 

STATISTICAL ANALYSIS OF THE 
ARMY TEST RESULTS 



PART 11. SECTION I 
THE PRINCIPAL SAMPLE 

All results from the psychological examinations in the 
camps were sent to Washington. It was impracticable as 
well as undesirable to tabulate the results in the case of 
every man examined. An intelligent selection or sampling 
of cases will give results more nearly typical of the country 
at large, than the entire group tested, which would be un- 
duly weighted for the more populous States, for camps 
giving the greatest number of examinations, particular 
draft quotas, etc. In order to obtain a sample for the white 
draft and the negro draft, cases were "randomly" (or better, 
impartially) selected in accordance with certain definite 
principles. The groups were as follows: 

Group I: White draft, pro-rated, by States. . . . 41,278 
Selected from 15 National Army 
camps, according to the State from 
which drafted, and according to the 
ratio of one recruit per thousand 
white male population. 

Group II: White draft, additional, by States. . . 14,684 
Additional selection of cases in- 
tended to bring the total represen- 
tation from each State up to 1,000 
cases. 

Group III: White draft, additional, by camps. . . 40,392 
Additional selection of cases in- 
tended to bring the entire samp- 
ling of the white draft up to ap- 
proximately 100,000 cases. 

75 



76 AMERICAN INTELLIGENCE 

Group IV: Negro draft, pro-rated, by States. . . . 19,992 
Selected in the same manner as 
Group I. 
Group V: Negro draft, additional, for Northern 

States 5,400 

Chosen to represent the negro draft 
of the north. 
Other groups were selected to meet other problems as 
follows : 

Group VI: White officers 15,528 

Selected proportionately to their 
occurrence in different arms of the 
service, with some additions to sup- 
plement the smaller arms, and the 
Medical Department. 

Group VIII: White established organizations 24,205 

Selected to provide comparison be- 
tween enlisted men of various arms 
of the service. 

Group X: Special experimental Group 1,047 

Randomly selected individuals of 
the white draft born in English 
speaking countries, who were given 
both alpha and beta, and, where 
possible, the Stanford-Binet ex- 
amination. 
These groups selected as representative of the country 
at large were analyzed by the Hollerith system of mechan- 
ical sorting. 

In this study we are concerned with: 
Groups I, II and III representing the white draft. 
Groups IV and V representing the negro draft, 
Group VI representing the white officers, and Group X, 
the special experimental group. 



SECTION II 

ANALYSIS OF THE MAIN GROUPS OF THE 
PRINCIPAL SAMPLE 



The tabulations in Memoir XV showing the distribution 
of scores on each type of examination of the white draft 
(Groups I, II, and III), the negro draft (Groups IV and 
V), and the white officers (Group VI) have been re-figured 
on the combined scale. The following tables were used: 

For the white draft: 

Alpha: Table 183 (p. 666) for men who took 

alpha only 67,254 

Beta: Table 184 (p. 666) for men who took beta 

only, or alpha and beta 23,547 

Individual: 

Table 185 (p. 667) for men who took Stanford- 
Binet examination only, or follow- 
ing alpha, following beta, or follow- 
ing alpha and beta 1,246 

Table 186 (p. 667) for men who took point 
scale examination only, or following 
alpha, following beta, or following 
alpha and beta 689 

Table 187 (p. 668) for men who took perform- 
ance scale examination only, or fol- 
lowing alpha, following beta, or fol- 
lowing alpha and beta 1,237 

Total Cases 93,973 

77 



78 AMERICAN INTELLIGENCE 

For the negro draft: 

Alpha: Table 239 (p. 716) for men who took 

alpha only 8,429 

Beta: Table 241 (p. 717) for men who took beta 

only, or alpha and beta 14,350 

Individual : 

Table 242 (p. 717) for men who took Stanford- 
Binet examination only, or follow- 
ing alpha, following beta, or follow- 
ing alpha and beta 403 

Table 229 (p. 711) for men who took point 
scale examination only, or following 
alpha, following beta, or following 

alpha and beta 390 

Table 228 (p. 710) for men who took perform- 
ance scale examination only, or fol- 
lowing alpha, following beta, or fol- 
lowing alpha and beta 32 

Total Cases 23,604 

For White officers : 

Alpha: Table 182 (p. 665) for all officers who 

took examination alpha only. 15,544 

It will be remembered that examination beta was given 
to all men who had been selected as illiterate or non-Eng- 
lish speaking before examination alpha was given, and also 
to those who took alpha and failed to make a reliable score. 
In the same way individual examinations were given to the 
lowest scoring cases in beta. Consequently, in figuring the 
results, if a man has taken both alpha and beta, his alpha 
score is disregarded and his beta score taken. In the same 
way, if he took both alpha and beta, and was then given an 



AMERICAN INTELLIGENCE 79 

individual examination, his alpha and beta scores are dis- 
regarded, and his score on the individual examination is 
taken as expressing the best measure of his intelligence. In 
other words we use alpha as our measure if alpha only was 
given, beta as our measure if beta alone, or alpha and beta 
were given, and the individual examination as our measure 
if it was given at all, on the assumption that the most re- 
liable test of a man was the last one given. 

The distribution of the scores of the white officers, white 
draft and negro draft is shown in Table 1. The first three 
columns are read in this manner: six officers and one re- 
cruit measured between 24.0 and 24.9 on the combined 
scale, one hundred and six officers and eighteen recruits be- 
tween 23.0 and 23.9, etc. The first three columns show the 
actual distributions, i. e. six officers out of 15,543, one re- 
cruit out of 93,955, etc. The last three columns show each 
of these distributions arranged according to the number in 
each ten thousand who scored at each class interval of the 
combined scale. The last three columns read in this man- 
ner: four officers in ten thousand test between 24.0 and 
24.9 on the combined scale; sixty-eight officers and two re- 
cruits in ten thousand score between 23.0 and 23.9 on the 
combined scale, etc. The "proportion in each ten thousand" 
may be read as a percentage by pointing off two decimal 
places. 

Table 1 also shows the average score of the white officers, 
white draft and colored draft on the combined scale to be 
18.84, 13.54 and 10.41 respectively. The standard devia- 
tion (S. D.) is also shown. An average has little significance 
without reference to a measure of variability of the series 
of measurements on which it is based. The conventionally 
accepted measure of variability is the standard deviation, 
which is derived by taking the square root of the average 
of the squares of the individual deviations from the aver- 



Table No. 1 

Distribution of the intelligence scores of the main groups 
of the principal sample on the combined scale. 











PROPORTION IN EACH 




ACTUAL 


DISTRIBUTION 1 








COMBINED 








TEN THOUSAND 


SCALE 














INTERVALS 
















WHITE 


WHITE 


NEGRO 


WHITE 


WHITE 


NEGRO 




OFFICERS 


DRAFT 


DRAFT 


OFFICERS 


i DRAFT 


DRAFT 


24.0-24.9 


6 


1 




4 






23.0-23.9 


106 


18 


.... 
.... 


68 


\ "k 




22.0-22.9 


612 


124 




S94 


t 13 


.... 


21.0-21.9 


1648 


444 


7 


106C 


) 48 


"3 


20.0-20.9 


2522 


1006 


16 


wrc 


J 107 


7 


19.0-19.9 


2836 


1804 


35 


1824 


t 192 


15 


18.0-18.9 


2698 


2996 


81 


i73e 


J 319 


34 


17.0-17.9 


2155 


4687 


172 


138'; 


^ 499 


73 


16.0-16.9 


1454 


6847 


330 


93^ 


i 729 


140 


15.0-15.9 


837 


9328 


600 


538 


} 993 


254 


14.0-14.9 


412 


12019 


1031 


26^ 


) 1279 


437 


13.0-13.9 


179 


14659 


1793 


11^ 


) 1560 


760 


12.0-12.9 


60 


14002 


2572 


3f 


) 1490 


1090 


11.0-11.9 


14 


9481 


2951 




) 1009 


1251 


10.0-10.9 


3 


6227 


3187 




I 662 


1351 


9.0-9.9 


1 


4433 


3319 




L 472 


1406 


8.0-8.9 


.... 


2876 


2891 




306 


1225 


7.0-7.9 


.... 


1683 


2149 




179 


911 


6.0-6.9 




814 


1315 




87 


557 


5.0-5.9 


.... 


334 


684 




36 


290 


4.0-4.9 


.... 


122 


302 




13 


128 


3.0-3.9 


.... 


37 


112 




4 


48 


2.0-2.9 


.... 


11 


38 




1 


16 


1.0-1.9 





2 


10 






4 


No. cases 


15543 


93955 


23596 








Average 


18.84 


13.54 


10.41 








S.D 


2.10 


2.92 


2.79 









80 




.^ g o fl 



82 AMERICAN INTELLIGENCE 

age. The size of the standard deviation indicates the vari- 
abihty of the group. In this case the white officers are more 
homogeneous (less variable) than the white draft and the 
negro draft (standard deviation of 2.10 compared with 2.92 
and 2.79). Furthermore, the negro draft is more homogen- 
eous than the white draft (2.79 compared with 2.92). 

v^^ Figure 31 shows graphically the proportions given in the 
last columns of Table 1. The horizontal line shows the com- 
bined scale intervals and the vertical lines the number in 
each ten thousand. These curves show very clearly the dis- 
tribution of intelligence in the three groups. In general, the 
distributions are similar in shape, but they differ markedly 
in their position on the scale of intelligence. The differences 
are very great. Of the officers, 98.87% are above the average 
of the white draft, and 99.97% are above the average of the 
negro draft. Of the white draft, 86.31% are above the aver- 
age of the negro draft. Only 13.13% of the negro draft are 
above the average of the white draft. This method of figur- 
ing gives us some indication of the differences between the 
groups. If the distribution of intelligence in two groups were 
the same, 50% of either group would exceed the average 
of the other group. If the distributions were absolutely dis- 
tinct, and there was no over-lapping, then 100% of one 
group would exceed the highest man in the other group. 
This last case would only occur if we compared very ex- 
treme groups (such as officers and idiots), and the conven- 
tional method is that of expressing the difference on a scale 
of 50%, i. e. the per cent, of one group above the average 

^ of the other group. 

^ The most reliable method of determining the relation- 
ship between two groups is that of comparing the differ- 
ence between the averages with the probable error of the 
difference. This method takes account of the variability of 
the original measures in each series, and also the reliability of 



AMERICAN INTELLIGENCE 83 

the measures as determined by both the variability and the 
number of cases. If the difference is 2 and the probable 
error of the difference is 1, the difference would be written 
2 1, meaning that the chances are even that it would 
not be less than 1, or more than 3. Difl^erences which 
are not at least four times as great as the probable error of 
the difference are not conventionally accepted as significant. 

Applying this method to the groups under consideration, /^-~ 
we find the following differences : 
Between white officers and white draft 5.30 .0131. 

(The difference is 405 times the probable error of the 

difference.) 
Between white officers and negro draft 8.43 .0167. 

(The difference is 505 times the probable error of the 

difference.) 
Between white draft and negro draft 3.13 .0138. 

(The difference is %%1 times the probable error of the 

difference.) 



SECTION III 

ANALYSIS OF THE WHITE DRAFT INTO 
FOREIGN AND NATIVE BORN 

The next problem is that of breaking up the white draft 
into its constituent elements as far as possible. In Chapter 
6, Part 3 of Memoir XV we find tables showing the dis- 
tribution of the scores of 12,492 recruits who reported that 
they were born in foreign countries. The tables in Memoir 
XV give the scores of all but 800 who reported foreign 
birth. As all of the 12,492 reported cases were in Groups 
I, II and III, they may be deducted from those groups, 
leaving a remainder composed of (1) native born, and (2) 
foreign born who failed to report on their psychological ex- 
amination blanks the fact that they were born in some 
country other than the United States. How large this latter 
group is we have no way of estimating. In the following 
tabulations the term "native born" is defined as all who 
stated that they were born in the United States, plus all 
who failed to record the country of their birth. 

The original data giving the score distributions of the 
12,492 foreign born were obtained from the following tables: 

Alpha: Table 207 (p. 693) for men who took 

alpha only 4,191 

Beta: Table 208 (p. 694) for men who took beta 

only, or alpha and beta 7,264 

Individual: 

Table 209 (p. 694) for men who took Stanford- 
Binet examination only, or follow- 
ing alpha, or following beta, or fol- 
lowing alpha and beta 207 

84 



AMERICAN INTELLIGENCE 85 

Table 210 (p. 596) for men who took perform- 
ance scale examination only, or fol- 
lowing alpha, following beta, or fol- 
lowing alpha and beta 830 



Total foreign born white draft 12,492 

The actual distributions of the scores of the native born 
and foreign born are shown in the first two columns of 
Table 2, while the last two columns in this table give the 
proportion in ten thousand scoring at each class interval. 
These distributions are also shown graphically in Figure 32. 
Here again it is apparent that we have two groups that are 
markedly different. Of the native born 74.8% exceed the 
average of the foreign born. The difference between the 
native born and the foreign born is 1.72 .0186, a differ- 
ence that is 92^/^ times the probable error of the difference. 

Comparing the native born white draft with the negro 
draft shown in the preceding section, we find that 88.76% 
of the native born white draft exceed the average of the 
negro draft. The difference between these two groups is 

3.36 ==.014, a difference that is 240 times the probable 
error of the difference. 

Comparing the foreign born with the negro draft, we find-^^ 
that 70.44% of the foreign born exceed the average of the 
negro draft. The difference between these two groups is 

1.64 .0212, a difference that is 77 times the probable 
error of the difference. 



Table No. 2 

Analysis of the white draft into foreign born and native 
born groups. Distribution of each group on the combined 
scale. 









PROPORTION IN EACH 




ACTUAL DISTRIBUTION 1 










TEN THOUSAND 


COMBINED 








SCALE 










INTERVALS 


NATIVE 


FOREIGN 


NATIVE 


FOREIGN 




BORN 


BORN 


BORN 


BORN 




WHITE 


WHITE 


WHITE 


WHITE 




DRAFT 


DRAFT 


DRAFT 


DRAFT 


24.0-24.9 


1 


..... 






23.0-23.9 


18 




"k 


i 


22.0-22.9 


120 


4 


15 


8 


21.0-21.9 


428 


16 


52 


13 


20.0-20.9 


971 


35 


119 


21 


19.0-19.9 


1733 


71 


213 


57 


18.0-18.9 


2850 


147 


349 


117 


17.0-17.9 


4403 


284 


540 


227 


16.0-16.9 


6345 


502 


779 


402 


15.0-15.9 


8537 


791 


1048 


633 


14.0-14.9 


10870 


1148 


1334 


919 


13.0-13.9 


13066 


1593 


1604 


1275 


12.0-12.9 


12220 


1782 


1500 


1426 


11.0-11.9 


7885 


1596 


968 


1277 


10.0-10.9 


4801 


1425 


589 


1141 


9.0-9.9 


3178 


1254 


390 


1004 


8.0-8.9 


1985 


892 


244 


714 


7.0-7.9 


1153 


530 


142 


424 


6.0-6.9 


556 


259 


68 


207 


5.0-5.9 


228 


106 


28 


94 


4.0-4.9 


83 


39 


12 


31 


3.0-3.9 


25 


12 


3 


10 


2.0-2.9 


7 


4 


1 


8 


1.0-1.9 


2 


1 




1 


No. of cases. . . . 


81465 


12492 




Average 


13.77 


12.05 






S. D 


2.86 


2.88 











86 



j9? 






I 

/I 

/ / 






N 



\ 



\ 



\ 



\ 



\ 



\ 

\1 



\ 






^ o 

o ^ 



^ ^ .S ^ i 3 



0) >. 



a; ^ ?^ fl, ^ ^ 



<i; .-S 



i % 



rl3 o 
G CO 






44H O 

.2 Id 

f3 en 

GO 4) 



cnH ^ I "^ ^ 

^ '^ ^ I a,t 

H ^ +S g 53 03 



(D 



en 



'-' ^ >^ 



o o 






OJ W5 4J 



be 



<U o -4^ 






^ be 






<v 



S a; ^ 



fl.5 fl 



^ ^ 






<u 



(U 



=^ be tn rrt - ^ 



b d fl (u be ;h 2 
g g^'S 3 'S .2 " 

O i-i n =d - '^ 



b* ^ 



Pi Ph O H 4-l 4H 



o .<u ^ 



SECTION IV 

ANALYSIS OF THE FOREIGN BORN WHITE 

DRAFT INTO YEARS OF RESIDENCE 

GROUPS 

The next problem is that of breaking up the foreign born 
white draft into its sub-groups, in order to discover the rea- 
son for the discrepancy between the main group of foreign 
born and the native born white draft. In Chapter 7, Part 3 
of Memoir XV we find tables showing the scores of 11,295 
foreign born (included in Groups I, II and III of the prin- 
cipal sample) classified according to the number of years of 
residence. The scores are tabulated for the following groups: 

1st : a six year period to 5 years of residence 

2nd: a five year period .... 6 to 10 years of residence 
Srd: a five year period .... 11 to 15 years of residence 
4th: a five year period .... 16 to 20 years of residence 
5th: a ten year period. . . .Over 20 years of residence 

As the age limit of the draft was 31, the last group would 
include cases who have been in the United States since 
childhood. It is probable that all of the 11,295 cases are in- 
cluded in the 12,492 cases shown in the preceding section. 
The discrepancy in numbers between the two groups is 
probably due to the fact that some foreign born reported 
the country of their birth, but failed to report the number 
of years they had been residents of the United States. 

The scores reported in Chapter 7 have been re-figured on 
the combined scale. The following tables were used for the 
original data: 



AMERICAN INTELLIGENCE 89 

Alpha :Table 219 (p. 701) for men who took alpha 

only 3,619 

Beta: Table 220 (p. 702) for men who took beta 

only, or alpha and beta 7,264 

Individual : 

Table 221 (p. 702) for men who took performance 
scale examination only, or following 
alpha, following beta, or following alpha 
and beta 802 

Table 222 (p. 703) for men who took Stanford- 
Binet examination only, or following 
alpha, or following beta, or following 
alpha and beta 207 

Total cases in all residence groups 11,295 

The distributions of the scores of the five years of resi- 
dence groups on the combined scale are shown in Table 3. 
This table shows a very remarkable fact, viz., a steady 
increase in the average scores with increasing years of 
residence, the averages being: 

TO 5 YBS. 6 TO 10 YRS. 11 TO 15 YRS. 16 TO 20 YES. OVER 20 YES. 

11.41 (2.85) 11.74 (2.80) 12.47 (2.77) 13.55 (2.60) 13.82 (2.71) 

From to 20 years of residence, the average rises steadily 
and the variability becomes less and less. 

Table 4 gives the difference between each group and 
every other group, together with the probable error of the 
difference, and the ratio of the difference to the probable 
error of the difference. All of the differences shown in Table 
4 are significant except the difference between groups "16 
to 20 yrs." and "over 20 yrs.," this difference (0.27) being 
only three times the probable error of the difference 
(0.0915). 



Table No. 3 

Analysis of the foreign born white draft by years of resi- 
dence in the United States. Distribution of each resi- 
dence group on the combined scale. 







TEARS 


OF RESIDENCE 




COMBINED 












SCALE 












INTEKVALS 


TO 5 


6 TO 10 


11 TO 15 


16 TO 20 


OVER 20 




YEARS 


YEARS 


YEARS 


YEARS 


YEARS 


4.0-24.9 






0.1 






23.0-23.9 


.... 


o'.i 


0.5 


.... 


.... 


22.0-22.9 


0.6 


0.6 


0.8 


0.5 


0.5 


21.0-21.9 


2.6 


2.9 


2.6 


1.6 


3.6 


20.0-20.9 


5.8 


8.2 


6.3 


3.6 


7.5 


19.0-19.9 


14.4 


18.6 


13.1 


8.3 


12.1 


18.0-18.9 


27.6 


37.6 


27.9 


18.0 


24.7 


17.0-17.9 


55.5 


72.0 


50.6 


36.7 


41.1 


16.0-16.9 


104.5 


141.8 


83.1 


61.2 


66.2 


15.0-15.9 


172.4 


240.7 


131.3 


87.3 


86.1 


14.0-14.9 


261.2 


355.2 


198.7 


106.1 


106.3 


13.0-13.9 


374.4 


506.2 


282.0 


128.6 


128.6 


12.0-12.9 


444.8 


597.0 


309.1 


124.5 


113.7 


11.0-11.9 


457.5 


572.1 


245.6 


82.6 


69.0 


10.0-10.9 


471.0 


533.0 


189.8 


47.4 


42.9 


9.0-9.9 


453.3 


479.2 


152.3 


29.5 


29.5 


8.0-8.9 


342.6 


347.2 


102.1 


18.1 


18.2 


7.0-7.9 


212.9 


209.6 


58.5 


9.9 


9.2 


6.0-6.9 


106.8 


102.4 


27.0 


4.5 


3.4 


5.0-5.9 


44.7 


41.7 


10.5 


1.7 


1.0 


4.0-4.9 


16.5 


14.9 


3.6 


0.6 


0.3 


3.0-3.9 


5.0 


4.4 


1.1 


0.2 


0.1 


2.0-2.9 


1.5 


1.3 


0.3 


0.1 




1.0-1.9 


0.4 


0.3 


0.1 






No. of cases 


3576 


4287 


1897 


771 


764 


Average 


11.41 


11.74 


12.47 


13.55 


13.82 


S. D 


2.85 


2.80 


2.77 


2.60 


2.71 







90 



Table No. 4 

Comparison of the average scores on the combined scale 
of the five years of residence groups of the foreign born 
white draft. 



YEARS OF RESIDENCE 
GROUPS 


TO 5 YRS. 
AVE. 11.41 

3576 CASES 


6 TO 10 YRS. 
AVE. 11.74 

4287 CASES 


11 TO 15 YRS. 

AVE. 12.47 
1897 CASES 


16 TO 20 YRS. 

AVE. 13.55 

771 CASES 


6 to 10 yrs. (Diff. 
Ave. 11.74 (P. E. Diff. 
4287 cases (Ratio 


+0.33 
0.0431 
7.7 








11 to 15 yrs. (Diff. 
Ave. 12.47 (P. E. Diff. 
1897 cases (Ratio 


+1.06 
0.0536 
19.8 


+0.73 
0.0517 
14.1 






16 to 20 yrs. (Diff. 
Ave. 13.55 (P. E. Diff. 
771 cases (Ratio 


+2.14 
0.0709 
30.2 


+1.81 
0.0695 
26.0 


+1.08 
0.0764 
14.1 




Over 20 yrs. (Diff. 
Ave. 13.82 (P. E. Diff. 
764 cases (Ratio 


+2.41 
0.0735 
32.8 


+2.08 
0.0721 
28.8 


+1.35 
0.0788 
17.1 


+0.27 
0.0915 
3.0 



91 



Table No. 5 

Comparison of the average scores on the combined scale 
of the native born white draft with the five years of resi- 
dence groups of the foreign born white draft. 





NATIVE BORN WHITE DRAFT. 81465 






CASES, AVE. 


13.77 

RATIO OF 






PROBABLE 


DIFFERENCE 


YEARS OF RESIDENCE 
GROUPS 


DIFFERENCE 


ERROR OF 
THE DIF- 


TO PROBABLE 
ERROR OF 






FERENCE 


THE DIF- 
FERENCE 


to 5 yrs. 








Ave. 11.41 


-2.36 


0.0104 


226.9 


3576 cases 








6 to 10 yrs. 








Ave. 11.74 


-2.03 


0.0296 


68.6 


4287 cases 








11 to 15 yrs. 








Ave. 12.47 


-1.30 


0.0434 


30.0 


1897 cases 








16 to 20 yrs. 








Ave. 13.55 


-0.22 


0.0636 


3.5 


771 cases 








Over 20 yrs. 








Ave. 13.82 


+0.05 


0.0664 


0.75 


764 cases 









93 



AMERICAN INTELLIGENCE 93 

Table 5 shows the relationship of the five years of resi- 
dence groups to the native born white draft. With increase 
in the time of residence, the differences between the native 
born and the foreign born become increasingly less signifi- 
cant. The difference between the native born and the "16 
to 20 yrs." group of foreign born is slight (0.22) and is less 
than four times the probable error of the difference 
(0.0636). The foreign born group in this country over 20 
years have an average score identical with the average 
score of the native born, the actual difference (0.05) being 
smaller than the probable error of the difference ( == 0.0664) . 

Figure 33 shows graphically the relationships between 
the averages of the years of residence groups and the native 
draft. In this graph, the horizontal line represents in- 
creasing length of residence and the vertical line repre- 
sents increase in the average score on the combined scale. 

This very remarkable fact of increase in the intelligence 
score with years of residence was commented on by the 
army authors in Memoir XV as follows : 

"It is not possible to state whether the difference is 
caused by the better adaptation of the more thoroughly 
Americanized group to the situation of the examination or 
whether some other factor is operative. It might be, for in- 
stance, that the more intelligent immigrants succeed and 
therefore remain in this country, but this suggestion is 
weakened by the fact that so many successful immigrants 
do return to Europe. At best we can but leave for future 
decision the question as to whether the differences repre- 
sent a real difference of intelligence or an artifact of the 
method of examination." (p. 704.) 

If our results reflect another factor independent of in- 
telligence, which might be designated "the better adapta- 
tion of the more thoroughly Americanized group to the 
situation of the examination," we have no means of con- 











NATIVE BORN WHITE DRAFT 











/ 


-^ 




.4^ 


/ 




^^^ 


^r 






' 









0TO5 



6 TO 10 n TO 15 16 TO 20 

YEARS RESIDENCE IN UNITED STATES 



OVER 20 



94 



Figure 33. Apparently increasing average intelligence with in- 
creasing years of residence. The horizontal scale reads from 
left to right according to increasing length of residence. The 
vertical scale represents score on the combined scale. For 
purposes of comparison, the position of the native born white 
draft on the combined scale is shown by a continuous line. The 
group of immigrants who have been in this country from 16 to 
20 years have an average intelligence almost as high as that of 
the native born^ while immigrants in this country over 20 years 
test the same as native born. 



95 



96 AMERICAN INTELLIGENCE 

trolling this factor. Ultimately, the validity of our conclu- 
sions from this study rests on the validity of alpha, beta, 
and the individual examinations. It is sometimes stated 
that the examining methods stressed too much the hurry- 
up attitude frequently called typically American. The ad- 
justment to test conditions is a part of the intelligence test. 
We have, of course, no other measure of adjustment aside 
from the total score on the examinations given. If the tests 
used included some mysterious type of situation that was 
"typically American," we are indeed fortunate, for this is 
America, and the purpose of our inquiry is that of obtain- 
ing a measure of the character of our immigration. Inability 
to respond to a "typically American" situation is obviously 
an undesirable trait. 

For our purposes then we will accept the definition of in- 
telligence given on page 573 of Memoir XV, viz., "by 'in- 
telligence' we mean the ability that manifests itself quanti- 
tatively in a set of consistent scores in all of the types of 
examination upon which our data are based." We are 
forced to include the "adjustment to test conditions" in 
our definition of intelligence. And we hope, probably in the 
teeth of the facts, that the adjustment to test conditions 
involved a situation that was "typically American." 

The hypothesis that the more intelligent immigrants re- 
main in this country while the more stupid ones go home, 
which was offered by the army authors to account for the 
increase of intelligence scores with increasing years of resi- 
dence, can not be checked from the data available in this 
study, and the emigration statistics give us little help. 
Table 6 shows the ratio between emigrant aliens and im- 
migrant aliens from each country from 1908 to 1917. No 
figures for emigrant aliens are available prior to 1908. 

Table 6 shows that since 1908 a very considerable num- 
ber (approximately one third) of immigrants have event- 



AMERICAN INTELLIGENCE 97 

ually returned to their native countries. If the selection factor 
mentioned were operating, and in the long run the depart- 
ing group contained more persons of lower intellectual capac- 
ity than the remaining group, an intelligence measurement 
would show an increase in the direction shown in Table 3, 
but the departing third would have to be very heavily 
weighted with low grade individuals to make any consider- 
able difference in the distribution of the remaining two 
thirds. 

Under the conditions of this study then, the hypothesis 
that the more intelligent immigrants succeed and therefore 
remain in this country must remain in the realm of specu- 
lation, as it can not be demonstrated as a fact. For our 
purposes the converse hypothesis, that the successful im- 
migrants save their earnings and return to Europe to live 
in comfort for the rest of their lives, is equally cogent. 
Either hypothesis is a legitimate speculation. We must 
agree with the army authors that the first hypothesis is 
weakened by the fact that so many successful immigrants 
do return to Europe. The first hypothesis is also weakened 
by the fact that if the more inferior individuals left, the 
distribution of the intelligence of the departing third would 
have to be very markedly skewed at the lower end of the 
scale. The distribution curve of the departing third would 
be skewed to such an extent that only 10% or 15% of this 
group would exceed the median of the remaining two 
thirds. We must conclude that the selection factor mention- 
ed might produce a slight change in the direction noted, but 
that it is highly improbable that any such factor could pro- 
duce a change in the amount observed. 

The important problem which we are facing is that of 
interpreting the fact of increase of intelligence test scores 
with increasing years of residence. Does our curve in Fig- 
ure 33 represent the growth of intelligence with increasing 



Table No. 6 

Per cent, that emigration was of immigration for fifteen 
countries since 1908. The figures were obtained by di- 
viding the number of emigrant ahens departed by the 
number of immigrant ahens admitted. The result is the 
ratio of emigration to immigration. The ratio 100 would 
mean that the number of alien immigrants admitted 
equalled the number of alien emigrants departed. 







AUSTRIA 


BELGIUM 


CANADA 


DENMARK 


ENGLAND 


GERMANY 


GREECE 


1908. 




78.2 
29.1 


20.5 
11.7 


6.8 

58.6 


13.9 
10.5 


10.6 
9.4 


21.0 
19.2 


28.5 


1909. 




40.0 


1910. 




18.3 


12.1 


60.4 


22.5 


9.7 


19.9 


31.4 


1911. 




54.2 


17.8 


86.9 


6.2 


10.3 


18.8 


35.7 


1912. 




49.5 


26.5 


60.0 


10.7 


16.6 


20.8 


54.4 


6Yr. 


Period 


42.8 


17.5 


57.9 


12.8 


11.3 


19.9 


37.2 


1913. 




23.0 


10.8 


63.6 


9.4 


13.7 


13.8 


134.0 


1914. 




27.0 


19.9 


37.0 


10.0 


20.2 


14.4 


31.0 


1915. 




74.7 


13.9 


28.3 


12.4 


35.8 


18.2 


77.8 


1916. 




16.1 


2.4 


15.5 


15.4 


39.8 


15.3 


17.8 


1917. 




18.9 


3.8 


18.1 


17.8 


33.5 


16.9 


8.5 


5Yr. 


Period 


26.7 


13.7 


30.4 


12.0 


23.6 


14.6 


47.7 


10 Yr 
1908- 


. Period 
1917 


36.8 


15.9 


40.3 


12.5 


15.7 


18.0 


42.8 



















TOTAL ALL 


HOLLAND 


IRELAND 


ITALY 


NORWAY 


RUSSIA 


SCOTLAND 


SWEDEN 


TURKEY 


COUNTRIES 


5.6 


6.6 


129.9 


18.3 


24.1 


11.1 


20.1 


15.4 


50.5 


6.6 


5.5 


45.4 


9.8 


16.4 


6.0 


8.0 


17.6 


30.1 


6.1 


5.9 


24.3 


5.9 


9.3 


5.5 


4.2 


9.2 


19.4 


5.5 


6.8 


39.8 


10.0 


17.0 


8.1 


7.8 


22.5 


33.6 


8.5 


11.9 


68.9 


26.6 


21.3 


15.1 


19.6 


23.5 


39.8 


6.4 


7.3 


55.8 


12.6 


17.4 


8.9 


10.5 


17.2 


33.8 


8.7 


10.4 


33.2 


19.9 


9.3 


17.2 


11.5 


15.4 


25.7 


10.9 


14.7 


29.7 


33.6 


18.5 


23.1 


15.1 


12.2 


24.9 


19.5 


15.6 


194.5 


15.2 


69.9 


39.5 


14.5 


12.7 


62.2 


12.1 


15.1 


215.6 


26.2 


66.8 


50.1 


22.6 


1.2 


43.4 


10. 1 


19.1 


36.3 


35.1 


46.7 


36.5 


15.2 


4.6 


22.4 


11.5 


13.7 


53.2 


25.1 


17.5 


25.0 


14.7 


13.3 


30.3 



8.4 9.6 54.6 16.9 17.4 13.7 12.1 15.7 



99 



100 AMERICAN INTELLIGENCE 

length of residence, does it represent an error in the method 
of measuring inteUigence, or, looked at from another angle, 
does it show the gradually decreasing inteUigence of the 
more recent immigrants examined in the army? 

The hypothesis of growth of intelligence with increasing 
length of residence may be identified with the hypothesis 
of an error in the method of measuring intelligence, for we 
must assume that we are measuring native or inborn intel- 
ligence, and any increase in our test score due to any other 
factor may be regarded as an error. It is therefore necessary 
to examine two hypotheses, viz., (1) a defect in the measur- 
ing scale, and (2) a change in the character of the immi- 
grants examined, in order to decide which is correct, or, in 
case both factors are operative, to estimate quantitatively 
the magnitude of one of the factors, so that allowance may 
be made for that factor and the weight of the other factor 
thus determined. 

The most probable source of error in our measure of in- 
telligence is that arising from the different types of exam- 
ination. Examination alpha involves the use of English, 
and the ability to use English is a function of intelligence 
and education in its broadest sense. Examination beta in- 
volves no English, and the tests can not be considered as 
educational measures in any sense. The individual exami- 
nations were adapted to the linguistic ability of the person 
examined. We would therefore expect to find an error in 
two places only, first, in the selection of men for alpha and 
beta, and second in the relationship between alpha and beta 
as expressed on the combined scale. 

If all members of our five years of residence groups had 
been given alpha, beta and individual examinations in 
equal proportions, then all would have been treated alike, 
and the relationship shown would stand without any pos- 
sibility of error. But this is not the case. The actual per- 



AMERICAN INTELLIGENCE 101 

centage of men in each residence group taking each type of 
examination was as follows : 



LENGTH OB' RESIDENCE 


ALPHA 


BETA 


STANFORD- 
BINET 


PERFORMANCE 
SCALE 


to 5 years . . . . 


19% 


68% 


2% 


11% 


6 to 10 years . . . . 


26% 


65% 


2% 


7% 


11 to 15 years. . . . 


41% 


54% 


1% 


4% 


16 to 20 years . . . . 


66% 


32% 


1% 


1% 


Over 20 years 


73% 


26% 


y2% 


K% 



Recognizing a variation in the type of examination given, 
our problem becomes that of determining whether or not 
any injustice has been done by converting results from 
these different types of examination into the combined 
scale. If the language and educational factors account for 
the rise in the average score on the combined scale with in- 
creasing years of residence, then we should expect that the 
contribution made to the combined scale score through ex- 
amination beta would remain constant, and the contribution 
from examination alpha would increase very rapidly. On 
the other hand, if the rise is independent of the language 
and educational factor and due to the greater native intel- 
ligence of the older groups, we should expect the contribu- 
tion from both types of examination to remain the same. 
In other words, if the combined scale is accurate, the dis- 
parity between proportions taking different types of ex- 
aminations would make no difference in the final results. 
At the same time we would not expect to find the average 
scores on the combined scale made by way of alpha and 
beta to be the same, for beta was given not only to those 
who were illiterate, but also to the dull and stupid who 
failed to make a good score on alpha. 

Computing, then, the average score on the combined scale 
made by way of examination alpha, beta, and the Stanford- 
Binet (which includes the results from the performance 



102 AMERICAN INTELLIGENCE 

scale) by each of the five years of residence groups, we find 
the following: 

TEABS OF SCORES DERIVED 

JV!^^.T^^ SCORES DERIVED SCORES DERIVED IT" , tiLt! 

BESIDENCE FROM THE 

** FROM ALPHA FROM BETA i^xvwm xxiji, 

GROUPS .tx^xi^ ^Kj ^ STANFORD-BINET 

to 5 years 13.46 (S.D. 2.47) 11.11 (S.D. 2.73) 9.99 (S.D. 2.21) 

6 to 10 years. . . 13.57 (S.D. 2.45) 11.29 (S.D. 2.65) 9.86 (S.D. 2.22) 

11 to 15 years. . . 13.91 (S.D. 2.25) 11.62 (S.D. 2.50) 10.19 (S.D. 2.09) 

16 to 20 years. . . 14.31 (S.D. 2.19) 12.22 (S.D. 2.68) 

Over 20 years... 14.56 (S.D. 2.32) 11.93 (S.D. 2.70) 



There is a steady progress in the scores in examination 
alpha from "0 to 5 yrs." up to "over 20 yrs.," the total 
gain being 1.12 points on the combined scale. There is an 
equal amount of progress in the scores from examination 
beta, a gain of 1.11 points on the combined scale being made 
in a shorter period of time, i. e. from "0 to 5 yrs." to "16 
to 20 yrs." If the increase in the average score on the com- 
bined scale from 11.41 to 13.82 were due to the language 
and educational factor, then the gain should come from 
alpha and not from beta, for alpha involves language and 
(indirectly) education, and beta does not. We actually find 
that the gain from each type of examination is about the 
same. This indicates, then, that the five years of residence 
groups are groups with real differences in native intellig- 
ence, and not groups laboring under more or less of a lin- 
guistic and educational handicap. 

There remains but one hypothesis that might establish 
the fact that the increase in the score on the combined scale 
with increasing length of residence was due to an error in 
the measuring scale, and that is the hypothesis that the 
combined scale was constructed in such a fashion that it 
penalized individuals born in non-English speaking coun- 
tries. It will be remembered that the combined scale was 
constructed from Group X, a special experimental group 
to which were given all three types of examination. Group 



AMERICAN INTELLIGENCE 103 

X was composed of 1047 individuals all of whom were born 
in English speaking countries. 

Fortunately we are able to test the reliability of the 
combined scale under the most severe conditions. On page 
654 of Memoir XV, Table 166 shows the scores on both 
alpha and beta of all individuals in Groups I, II and III 
of the principal sample who had been given both alpha and 
beta. This group includes 4893 cases. It is obvious that we 
may figure these 4893 cases as either alpha cases or beta 
cases and convert them into the combined scale either by 
Table 159 (the alpha conversion table) or Table 162 (the 
beta conversion table). This the army writers have done, 
and the results are given in Table 167 on page 655. 

It is found that when we treat the 4893 cases as alpha 
cases the average score on the combined scale is 10.775 
(S.D.I. 64). When we treat the same4893 cases as beta cases, 
the average score on the combined scale is 12.158 (S. D. 
2.63). The actual difference between the two averages is 
1.383 (=1=0.0298). In commenting on this result, the army 
writers state: 

"At first glance these results seem rather startling, for 
one might suppose that going from alpha (for a given num- 
ber of cases) to the combined scale ought to yield the same 
results as going from beta to combined scale. The facts are 
quite the contrary. However, this difference in no wise dis- 
credits the method. It must be remembered that in a group 
of this sort there is a large percentage of illiterates; thus 
the group no doubt includes a considerable proportion of 
the cases who made unsatisfactory scores in alpha and 
were recalled to beta not because of stupidity but because 
of language difficulty. When they reached beta, they were 
able to make scores more consistent with their ability. It 
is precisely this element of the group that causes the dif- 
ference in the two means on the combined scale. The same 



104 AMERICAN INTELLIGENCE 

fact explains the wide differences in the standard devia- 
tions. The standard deviation of the combined scale dis- 
tribution when reached by way of beta is larger than by 
way of alpha. Here the difference is no doubt due to the 
fact that in alpha both the stupid and the non-English 
speaking piled up in the lower class intervals, while in beta 
the stupid remained in the lower ranges and the more in- 
telligent went higher, thus increasing the standard devia- 
tion." (p. 655.) 

It is also possible to study the effect of using different 
conversion tables in the case of Group X, the special ex- 
perimental group which was composed of individuals who 
were born in English speaking countries. The army writers 
report on page 645 the analysis of Group X in this manner. 
The following averages and standard deviations are re- 
ported for the different methods of treating the data : 

A\'ERAGE S. D. 

(1) Treating all 1047 cases as measured by alpha only . . . 13.82716 3.03940 

(2) Treating all cases scoring less than 25 points in alpha 

as beta cases and the remainder as alpha 

cases 13.88606 3.03776 

(3) Treating all cases scoring less than 50 points in alpha 

as beta cases and the remainder as alpha 

cases 13.94350 3.20690 

(4) Treating all cases scoring less than 75 points in alpha 

as beta cases and the remainder as alpha 

cases 13.96981 3.30997 

The approximate agreement of the averages derived by 
the four different methods is considered by the army 
writers to be a proof of the validity of the transformation 
tables which they publish for converting alpha, beta, and 
Stanford-Binet distributions into combined scale distri- 
butions. 

We thus have two extreme instances of the results of 



AMERICAN INTELLIGENCE 105 

treating the same groups in different ways. If we take all 
cases in the principal sample to whom were given both 
alpha and beta, we find a difference of 1.383 (0.0298) in 
the combined scale score of the group when figured first as 
alpha cases and then as beta cases. On the other hand, 
since the combined scale was empirically derived from the 
1047 cases in Group X, it makes very little difference how 
we treat the results of that group. We must now determine 
whether the distributions of alpha scores of the foreign 
born groups most closely resemble the distribution of alpha 
scores in Group X, or the distribution of the alpha scores 
of the 4893 cases who had both alpha and beta. This ques- 
tion is very easily answered by turning to Table 7, which 
shows the alpha score distribution of the following groups : 

(1) The 4893 cases who had both alpha and beta. 

(2) The 1047 cases in the special experimental group. 

(3) The 679 cases of foreign born who had been in this 

country from to 5 years. 

(4) The 1098 cases of foreign born who had been in this 

country from 6 to 10 years. 

(5) The 1777 cases of foreign born who had been in this 

country from to 10 years, a group obtained by com- 
bining (3) and (4). 
All five distributions have been made comparable by re- 
ducing them to the proportion in each 1000 scoring at 
each class interval of the alpha examination. Figure 34 
shows graphically the relationships between the 4893 alpha 
and beta cases, the 1047 cases in the special experimental 
group and the 1777 foreign born who had been in this 
country for 10 years or less. In Figure 34 the horizontal 
line indicates alpha scores and the vertical line shows the 
proportion in each 1000. 

A hasty survey of Table 7 shows that our group of 4893 
cases is a very specially selected group. 97.3% of the alpha 



Table No. 7 

Distribution of alpha scores of (1) all cases in Groups I, II, 
and III given both alpha and beta, (2) special experi- 
mental group, (3) foreign born in U. S. to 5 years, (4) 
foreign born in U. S. 6 to 10 years, (5) foreign born in 
U. S. to 10 years. All distributions reduced to common 
denominator of number per 1000 at each class interval. 
Actual distributions may be found on pages 621, 654 and 
701 of Memoir XV. 





4893 CASES 


1047 CASES 


679 CASES 


1098 CASES 


1777 CASES 


ALPHA 


TAKING BOTH 


IN SPECIAL 


IN u. a. 


IN U. 8. 


IN U. 8. 


CT,ASS 


ALPHA AND 


EXPERIMENTAL 


TO 5 


6 TO 10 


TO 10 


INTEBVAL8 


BETA 


GROUP 


TEARS 


TEARS 


TEARS 


185-189 




3 






.... 


180-184 




1 


"2" 


i 




175-179 




1 




.... 
.... 


.... 


170-174 




2 


"4' 




"2" 


165-169 




7 


2 


1 


1 


160-164 




4 




3 


2 


155-159 




6 


"k' 


2 


2 


150-154 




6 




4 


2 


145-149 




7 


"e 


2 


3 


140-144 


.... 


9 


12 


5 


7 


135-139 




12 


6 


8 


7 


130-134 




9 


3 


7 


6 


125-129 


.... 


20 


7 


7 


7 


120-124 




13 


3 


6 


5 


115-119 




19 


9 


11 


10 


110-114 


*'i* 


12 


19 


10 


14 


105-109 


.... 


21 


9 


14 


12 


100-104 




19 


15 


17 


17 


95-99 


1 


39 


26 


16 


19 


90-94 


1 


24 


26 


27 


26 


85-89 


1 


30 


22 


37 


31 


80-84 


2 


41 


27 


40 


34 


75-79 


1 


25 


32 


31 


31 


70-74 


2 


39 


53 


29 


39 


65-69 


1 


38 


29 


48 


40 


60-64 


3 


44 


47 


61 


60 


55-59 


7 


38 


52 


48 


60 


50-54 


6 


35 


66 


65 


69 


45-49 


8 


40 


72 


56 


62 


40-44 


12 


44 


68 


75 


73 


85-39 


19 


63 


81 


75 


78 


30-34 


23 


53 


49 


61 


66 


25-29 


59 


48 


57 


68 


64 


20-24 


104 


53 


44 


62 


55 


16-19 


149 


31 


29 


27 


28 


10-14 


286 


42 


22 


29 


26 


6-9 


207 


46 


29 


21 


24 


0-4 


104 


67 


72 


47 


67 



106 



AMERICAN INTELLIGENCE 107 

scores of the 4893 cases are below a total alpha score of 50. 
On the other hand, the distributions of alpha scores in our 
groups of foreign born very closely resemble the distribu- 
tion of alpha scores of Group X, the special experimental 
group on which the combined scale was based. If we com- 
pute the percentage of cases in each residence group graded 
on alpha falling below a total alpha score of 50, we find the 
following: 

PER CENT. GRADED BY ALPHA 
YEARS OF RESIDENCE GROUP AND FALLING BELOW A TOTAL 

ALPHA SCORE OF 50 

to 5 years 52.4% 

6 to 10 years 52.2% 

11 to 15 years 48.7% 

16 to 20 years 39.1% 

Over 20 years. 36.0% 

Group X shows 47.7% of the 1047 cases falling below a 
score of 50 on alpha. This group is approximately the same 
as the group of foreign born we are studying. On the other 
hand, the group of 4893 cases shows 97.3% of the alpha 
scores below 50, and a very marked piling up of cases below 
25 on alpha. The relationships between the distributions 
of scores in the three groups are very clearly shown in 
Figure 34. 

It is clear that the combined scale would penalize the 
foreign born only if all individuals who took alpha and 
beta had been scored as alpha cases. As a matter of fact, 
the opposite is true, for every individual who took both 
alpha and beta was scored as a beta case, and the alpha 
score was disregarded. On this account our results on the 
foreign born are not subject to the distortion shown by 
treating 4893 alpha and beta cases as alpha cases. 

The 4893 cases treated as beta cases, not as alpha cases, 
are in our group of 93,955 cases representing the white 



FOREIGN BORN 0-10 YEARS IN U. S. 




109 



Figure 34. Distributions of alpha scores of three groups: (A) 
foreign born individuals in this country from to 10 years, (B) 
group X, the special experimental group, (C) all cases who took 
both alpha and beta. The horizontal scale shows alpha scores, 
and the vertical scale proportions in each thousand. The distri- 
butions are drawn from the figures given in Table 7. To prove 
that the combined scale does not penalize the foreign born, it is 
only necessary to show that the distribution of alpha scores for 
the foreign born is approximately similar to that of group X, 
and unlike that of all alpha-beta cases. This similarity is appar- 
ent from our chart. The men given both alpha and beta would 
have been penalized if they had been scored through the com- 
bined scale as alpha cases. This did not happen, however, for 
the alpha scores of these men were disregarded, and only the 
beta record used. 



110 AMERICAN INTELLIGENCE 

draft, and the chances are that most of them are among 
the 12,492 members of this group who were born in foreign 
countries. By treating the group as alpha cases, we show 
that the maximum correction factor that it would be neces- 
sary to apply to a group graded wrongly by alpha would be 
1.383 points on the combined scale. But inasmuch as all 
alpha and beta cases have been correctly graded by using 
the beta conversion tables, there remain only those cases 
graded by alpha alone who should have been given exam- 
ination beta as well. It is of course impossible to estimate 
the number of cases of this sort. If we make the very ex- 
travagant assumption that 25% of the cases graded by 
alpha alone in the residence group "0 to 5 yrs." should have 
been graded by beta, and credit each of these cases with 
1.383 points, the final correction that it would be necessary 
to apply to our average would be 0.064 points, a quantity 
which is negligible in view of the magnitude of the differ- 
ences under consideration. Furthermore, we should not be 
justified in using a correction factor of any sort unless it 
could be shown that the distribution of the alpha scores 
of the foreign born groups was very unlike that of the dis- 
tribution of alpha scores of the special experimental group. 
As a matter of fact, these distributions as shown in Figure 
34 are very much alike. 

We have therefore demonstrated the accuracy of the 
combined scale as a measure of the intelligence of the 
groups under consideration. We must therefore accept the 
conclusion that under the conditions of this experiment the 
differences shown in the average scores of the five years 
of residence groups indicate real differences in intelligence 
and not a defect in the measuring scale. Instead of con- 
sidering that our curve (Figure 33) indicates a growth of 
intelligence with increasing length of residence, we are 
forced to take the reverse of the picture and accept the 



AMERICAN INTELLIGENCE 111 

hypothesis that the curve indicates a gradual deterioration 
in the class of immigrants examined in the army, who came 
to this country in each succeeding ^ve year period since 
1902. 



SECTION V 

ANALYSIS OF IMMIGRATION TO THE 
UNITED STATES 



The fact that the average inteUigence of the immigrants 
examined in the army who came to this country in each 
successive five year period since 1902 becomes progres- 
sively lower with each succeeding period indicates that an 
explanation of this phenomenon might be found in a change 
in the character of immigration. We must, therefore, turn 
to the statistics on immigration to see if any such change 
can be detected. Table 8 shows the percentage of the total 
immigration coming from various countries in the periods 
roughly corresponding to the five years of residence periods 
covered in the army statistics. The data on immigration 
were obtained from the Statistical Abstract of the United 
States for the years 1900, 1910, and 1920. Table 8 reads 
as follows: in the years 1887 to 1897, the period roughly 
corresponding with our residence group "over 20 yrs.," 
10.9% of our total immigration came from England, 2.7% 
from Scotland, 1% from Holland, etc. The relations shown 
in Table 8 are shown graphically in Figure 35. Each com- 
plete bar in Figure 35 represents 100%. The per cent, which 
each country has contributed to the total immigration of 
each period has been scaled off proportionately in each bar. 
These figures show that the most abrupt change in the 
character of immigration came between the periods 1887- 
1897 and 1898-1902. These periods show a very marked 
decrease in the proportion of the immigration from Eng- 
land and Germany, and a substantial decrease in the pro- 
portion of immigration from Scotland, Sweden, and 

112 



AMERICAN INTELLIGENCE 113 

Table No. 8 

Per cent, of total immigration coming from various coun- 
tries during periods roughly corresponding to the five 
years of residence groups. 

1913-1917 1908-1912 1903-1907 1898-1902 1887-1897 
TO 5 6 TO 10 11 TO 15 16 TO 20 over 20 

YEARS YEARS YEARS YEARS YEARS 

England 3.7 5.1 4.6 2.6 10.9 

Scotland 1.0 1.8 1.4 0.5 2.7 

Holland 0.6 0.8 0.5 0.4 1.0 

Germany 2.5 3.5 4.0 4.8 18.7 

Denmark 0.7 0.7 0.8 0.8 1.7 

Canada 13.5 6.0 0.6 0.2 No record 

Sweden 1.5 2.0 2.8 4.6 7.3 

Norway 1.1 1.5 2.3 2.4 2.7 

Belgium 0.5 0.5 0.5 0.3 0.6 

Ireland 2.4 3.3 3.8 7.2 11.9 

Austria 16.7 21.8 24.9 23.6 11.6 

Turkey 2.6 3.4 1.9 0.7 No record 

Greece 3.7 2.5 1.8 1.1 0.2 

Russia 17.8 18.3 18.3 17.8 12.0 

Italy 20.0 20.2 23.7 25.8 12.4 

Allothers 11.7 8.6 8.1 7.2 6.3 

Ireland. On the other hand, the proportion of immigrants 
coming from Austria, Russia, and Italy showed a marked 
increase at this time. In general the following relations 
held: 

England showed a decided drop in the proportion of im- 
migrants furnished between the period 1887-1897 and the 
period 1898-1902. There has been a slight increase since 
1898, but the proportion is less than 5%, when formerly it 
was over 10%. 

Scotland contributed 2.7% of our total immigration in 
the period 1887-1897, and since that time, never more than 

Holland never contributed more than 1% of our total 
immigration in any period covered by these figures. 

Germany contributed 18.7% of our immigration in the 
period 1887-1897, 4.8% in the period 1898-1902, and since 
that time the proportion has decreased with each succeed- 
ing period. 



DATES 1887 

1897 

YEARS ovEn 



ENGLAND 



SCOTLAND 
HOLLAND 



GERMANY 

DENMARK 
SWEDEN 



NORWAY 
BELGIUM 



IRELAND 

AUSTRIA 

GREECE 
RUSSIA 

ITALY 



ALL OTHERS 




TO 
1912 

6 10 



^ -=;^^- - ^ ^ GERMAN Y 

-^-^^.^_ _ r^^^-.^-c::-p DENMARK 




1918 DATES 

TO 

1917 

0-5 YEARS 
~JH| ENGLAND 
. - SCOTLAND 
rteBHOLLAND 
^m GER>IAN 



M 



CANADA 



SWEDEN 
NORWAY 
BELGIUM 
IRELAND 



AUSTRIA 



M TURKEY 
GREECE 



RUSSIA 



L^"i 



ITALY 



ALL OTHERS 



114 



Figure 35. Analysis of immigration by countries. Each vertical 
bar represents 100%, and each subdivision represents the per 
cent, of the total immigration that each country furnished in the 
period under consideration. It is apparent that the most sudden 
change in the character of immigration came between the per- 
iods 1887-1897 and 1898-1902. On the other hand, the curve 
showing the relationship between years of residence and average 
intelligence shows no correspondingly large difference in the in- 
telligence of immigrants who came here in these periods (Figure 
22, Groups "16 to 20" and "over 20"). Figure 22 shows a large 
difference between the "16 to 20 yrs." and "11 to 15 yrs." 
groups, but the chart above shows no marked difference in the 
immigration coming to this country in the corresponding five- 
year periods, 1898-1902 and 1903-1907. 



115 



116 AMERICAN INTELLIGENCE 

Denmark contributed 1.7% of our immigration in the 
period 1887-1897 and since that time less than 1% in each 
period. 

The immigration from Canada was not reported prior to 
1896. There was a very marked increase in the percentage 
of our total immigration coming from Canada in the period 
1908-1912 and again in the period 1913-1917. 

The proportion of immigration coming from Sweden has 
decreased steadily from 7.3% in the period 1887-1897 to 
1.5% in the period 1913-1917. A similar decrease from 
2.7% to 1.1% is shown by Norway during the same periods. 

Belgium has contributed less than 1% to the total immi- 
gration in each period under consideration. 

The immigration from Ireland has decreased in propor- 
tion from 11.9% to 2.4% in the five year periods shown. 

All other countries show a gain in the proportion of im- 
migrants which they supply. Austria supplied 11.6% in the 
period 1887-1897, this proportion jumping to 23.6% in the 
period 1898-1902 and remaining above 20% until the last 
period 1913-1917, a period which reflects the war conditions 
in Europe. 

The proportion of immigration from Russia has increased 
from 12% prior to 1898 to 18% since that time. The pro- 
portion of immigration from Italy, which was about 12% 
prior to 1898, has never been below 20% since. Turkey and 
Greece show a small but increasing proportion in the suc- 
cessive periods covered. 

Enough evidence has been cited to show that there has 
been some change in the character of our immigration dur- 
ing the periods covered in the army report. The gradual 
decline in the average inteUigence of the more recent im- 
migrants examined in the army might be due to these 
changes in the source of supply. If this hypothesis is correct, 
we should expect to find differences between the scores of 



AMERICAN INTELLIGENCE 117 

the representatives of each country making up our total 
group of the foreign born white draft. 



SECTION VI 



ANALYSIS OF THE FOREIGN BORN WHITE 
DRAFT BY COUNTRY OF BIRTH 



The army reports show distributions of psychological 
test scores for all of the 12,492 foreign born cases classified 
according to the country of birth, although these figures 
for each country are not sub-divided again into years of 
residence groups. As a matter of fact, when we break up 
the 12,492 cases according to the country of birth, the 
figures from certain countries become very small; so that 
further sub-division would make the results valueless. We 
can, however, examine the figures which give us an intelli- 
gence measure of the foreign born men of our army and take 
them for what they are worth. Even though the number 
of cases is very small for certain nationalities, we previous- 
ly saw that the reliability of the difference does not depend 
entirely on the number of cases, but on three factors the 
size of the difference, the variability of each series of meas- 
urements, and the number of cases in each series. 

The same tables from which we derived the 12,492 for- 
eign born cases (Tables 207, 208, 209 and 210; pp. 692, 693 
and 694) give the classification of test scores by country of 
birth. From these tabulations the combined scale score of 
each nativity group has been computed. The actual dis- 
tributions of these combined scale scores are shown in 
Table 9. These distributions reduced to percentages are 
shown in Table 10. 

The differences found, expressed in terms of the per cent, 
from each country who exceed the average native born 
American, are as follows: 

118 



AMERICAN INTELLIGENCE 



119 



England 67.3% 


Belgium 


35.3% 


Scotland 58.8% 


Austria 


28.2% 


Holland 58.1%) 


Ireland 


26.2% 


Germany 48.7% 


Turkey 


25.3% 


Denmark 47.8% 


Greece 


21.3% 


Canada 47.3% 


Russia 


18.9% 


Sweden 41.7% 


Italy 


14.4% 


Norway 37.3% 


Poland 


12.2% 



|. The actual diifferences between the average scores on the 
combined scale of each country and every other country, 
together with the probable errors of the differences, and 
the ratios of the probable errors of the differences to the 
differences are shown in Tables 11 to 27 inclusive. In these 
tables, the convention has been followed of marking a dif- 
ference "unreliable" if the actual difference was less than 
four times the probable error of the difference. The coun- 
tries tabulated are arranged in the order of the increasing 
reliability of the differences found, above and below the 
average of the country with which they are compared. 

The relations between the averages shown in Table 9 are 
shown graphically in Figure 36. For the convenience of 
those who use Stanford-Binet "mental ages," one side of the 
scale in Figure 36 has been drawn so as to read in "mental 
ages," and the other side to read in combined scale units 
the units in which our averages and measures of variability 
have been calculated. The Stanford-Binet "mental age" 
scale was calculated from the regression equation given on 
page 654 of Memoir XV: 

Mental age (in years) = 0.778 C.S.-|-2.606. 



Table No. 9 

Analysis of foreign born white draft by country of birth. 
Distribution of scores on the combined scale, of men 
born in each country. 



COMBINED 


















SCALE 


ENGLAND 


SCOTLAND 


HOLLAND 


GERMANY 


DENMARK 


CANADA 


SWEDEN 


NORWAY 


INTERVALS 


















24.0-24.9 


0.1 
















23.0-23.9 


0.7 






.... 


.... 


.... 






22.0-22.9 


1.7 


6.5 


o.i 


0.5 




0.6 






21.0-21.9 


SO 


1.4 


0.7 


1.2 


'o'.i 


4.3 


0.5 


o'.i 


20.0-20.9 


6.0 


1.9 


1.6 


2.4 


0.5 


8.7 


2.2 


1.2 


19.0-19.9 


10.7 


2.6 


3.0 


5.1 


1.9 


15.8 


5.4 


3.2 


18.0-18.9 


21.1 


3.7 


4.5 


7.1 


5.8 


28.2 


10.2 


8.4 


17.0-17.9 


33.7 


8.9 


7.2 


14.7 


13.7 


47.8 


21.3 


18.4 


16.0-16.9 


48.9 


17.9 


12.9 


24.7 


26.4 


77.2 


44.3 


34.0 


15.0-1.5.9 


64.2 


20.8 


20.4 


34.4 


39.9 


104.4 


72.6 


59.1 


14.0-14.9 


71.5 


22.9 


25.1 


46.3 


52.0 


135.2 


101.9 


81.4 


13.0-13.9 


65 7 


23.0 


25.5 


57.1 


64.0 


163.2 


125.6 


98.5 


12.0-12.9 


40.9 


18.0 


18.6 


53.2 


54.2 


144.4 


118.3 


101 2 


11.0-11.9 


19.2 


11.0 


10.2 


31.5 


31.2 


93.5 


79.3 


79.5 


10.0-10.9 


10.2 


6.1 


5.8 


16.4 


18.2 


60.2 


51.4 


55.8 


9.0-9.9 


6.1 


3.5 


2.8 


8.0 


10.3 


41.7 


31.7 


36.4 


8.0-8.9 


3.7 


2.1 


1.1 


3.5 


4.6 


25.0 


16.2 


19.8 


7.0-7.9 


2.2 


1.1 


0.4 


1.3 


1.7 


13.0 


7.0 


9.2 


6.0-6.9 


1.0 


0.5 


0.1 


0.5 


0.4 


5.6 


2.4 


3.4 


5.0-5.9 


0.3 


0.1 




0.1 


0.1 


2.1 


0.6 


1.0 


4.0-4.9 


0.1 










0.8 


0.1 


3 


S.0-3.9 








'.'.'.'. 




0.2 






2.0-2.9 












0.1 




.... 


1.0-1.9 






.... 







.... 






No. cases 


411 


146 


140 


308 


325 


972 


691 


611 


Average 


14.87 


14.34 


14.32 


13.88 


13.69 


13.66 


13.30 


12.98 


S. D.. 


2.57 


2.63 


2.39 


2.43 


2.23 


2.67 


2.38 


2.47 







150 



BELGIUM 


IHF.T.AND 


AUSTRIA 


TURKEY 


GREECE 


RUSSIA 


ITALY 


POLAND 






6!i 









O.Q 


b'.k 








1.1 


'6'.4 


0.3 


.... 


2.3 


0.8 








2.0 


1.4 


0.8 


0.2 


3.8 


2.1 




o.k 


3.3 


2.0 


1.5 


0.8 


7.7 


5.9 


o'.k 


0.8 


7.5 


3.0 


4.3 


3.0 


20.7 


14.4 


0.9 


2.4 


13.5 


5.0 


9.6 


8.2 


40.0 


32.8 


2.7 


7.0 


23.7 


10.1 


15.6 


14.8 


62.7 


70.0 


5.8 


13.3 


38.2 


20.1 


24.4 


26.5 


97.7 


136.4 


11.1 


18.6 


59.4 


32.1 


36.9 


49.8 


152.8 


235.0 


18.3 


22.3 


102.0 


46.8 


57.2 


83.0 


233.6 


377.2 


33.1 


22.6 


122.6 


50.8 


67.4 


99.2 


301.1 


508.7 


47.2 


16.5 


94.0 


38.2 


56.9 


88.5 


317.0 


569.2 


53.8 


11.0 


68.9 


29.3 


47.8 


73.1 


316.7 


596.3 


56.0 


7.3 


55.0 


24.7 


41.0 


57.3 


299.2 


573.6 


55.2 


4.4 


35.7 


17.5 


28.9 


35.5 


226.0 


423.4 


43.4 


2.6 


18.6 


10.6 


17.3 


19.0 


141.3 


255.7 


28.5 


1.3 


8.0 


5.4 


8.2 


8.5 


71.1 


126.6 


15.2 


0.5 


3.0 


2.3 


3.3 


3.2 


29.9 


53.0 


6.8 


0.2 


1.0 


0.9 


1.2 


1.0 


11.2 


19.5 


2.6 




0.3 


0.3 


0.3 


0.3 


3.4 


6.0 


0.9 




.'.'. 


0.1 


0.1 


0.1 


1.0 
0.2 


1.8 
0.4 


0.3 


129 


658 


301 


423 


572 


2340 


4009 


382 


12.79 


12.32 


12.27 


12.02 


11.90 


11.34 


11.01 


10.74 


2.42 


2.60 


2.75 


2.75 


2.45 


2.83 


2.60 


2.59 



121 



Table No. 10 

Analysis of foreign born white draft by country of birth. 
Per cent, from each country scoring at each interval on 
the combined scale. 



COMBINED 


















SCALE 


ENGLAND 


SCOTLAND 


HOLLAND 




DENMABE 


CANADA 


SWEDEN 


NOBWAY 


miEBVALS 


















23.0-23.9 


0.2 


.... 






.... 








22.0-22.9 


0.4 


0.3 


'6!i 


0.2 




o'.i 






21.0-21.9 


0.7 


1.0 


0.5 


0.4 


.... 


0.4 


'6!i 


.... 


20.0-20.9 


1.5 


1.3 


1.1 


0.8 


'6!2 


0.9 


0.3 


*6.2 


19.0-19.9 


2.6 


1.8 


2.1 


1.7 


0.6 


1.6 


0,8 


0.5 


18.0-18.9 


5.1 


2.5 


3.2 


2.3 


1.8 


2.9 


1.6 


1.4 


17.0-17.9 


8.2 


6.1 


6.1 


4.8 


4.2 


4.9 


3.1 


3.0 


16.0-16.9 


11.9 


12.2 


9.3 


8.0 


8.2 


7.9 


6.4 


5.6 


16.0-15.9 


15.6 


14.3 


14.6 


11.2 


12.3 


10.8 


10.6 


9.7 


14.0-14.9 


17.4 


15.7 


17.9 


15.0 


16.0 


13.9 


14.8 


13.3 


13.0-13.9 


16.0 


15.8 


18.2 


18.5 


19.7 


16.8 


18.1 


16.1 


12.0-12.9 


10.0 


12.3 


13.3 


17.3 


16.6 


14.9 


17.2 


16.6 


11.0-11.9 


4.7 


7.5 


7.3 


10.2 


9.6 


9.6 


11.4 


13.0 


10.0-10.9 


2.5 


4.2 


4.1 


5.3 


5.6 


6.2 


7.6 


9.1 


9.0-9.9 


1.5 


2.4 


2.0 


2.6 


3.2 


4.3 


4.6 


6.0 


8.0-8.9 


0.9 


1.4 


0.8 


1.1 


1.4 


2.6 


2.8 


3.8 


7.0-7.9 


0.5 


0.8 


0.3 


0.4 


0.5 


1.3 


1.0 


1.6 


6.0-6.9 


0.2 


0.4 


0.1 


0.2 


0.1 


0.8 


0.3 


0.6 


5.0-5.9 


0.1 










0.2 


0.1 


0.2 


4.0-4.9 












0.1 




.... 


S.0-3.9 


.... 


.... 


.... 


.... 


.... 


.... 




.... 


2.0-2.9 





















Ifd 



BELGIUM 


IBELAND 


AU8TBIA 


TUEKET 


GREECE 


RUSSIA 


ITALY 


POLAND 


'. . . . 


6.2 


'6!i 


o'.i 





'6!i 


.... 







0,3 


0.5 


0.2 


.... 


0.2 






'0.2 


0.5 


0.7 


0.4 


'6!i 


0.4 


'6!i 


'6]i 


0.6 


1.1 


1.0 


1.0 


0.5 


0.9 


0.4 


0.2 


1.9 


2.1 


1.7 


2.3 


1.4 


1.7 


0.8 


0.7 


5.4 


3.6 


3.4 


3.7 


2.6 


2.6 


1.7 


1.5 


10.3 


5.8 


6.6 


5.8 


4.7 


4.2 


3.4 


2.9 


12.9 


9.0 


10.6 


8.7 


8.7 


6.5 


5.8 


4.8 


17.3 


15.5 


15.6 


13.5 


14.5 


10.1 


9.4 


8.7 


17.5 


18.7 


16.9 


15.9 


17.4 


12.9 


12.7 


12.3 


12.8 


14.3 


12.7 


13.4 


15.4 


13.5 


14.2 


14.1 


8.5 


10.5 


9.7 


11.3 


12.8 


13.5 


14.9 


14.6 


5.6 


8.3 


8.2 


9.7 


10.0 


12.8 


14.3 


14.4 


3.4 


5.4 


5.8 


6.8 


6.2 


9.7 


10.5 


11.4 


2.0 


2.8 


3.5 


4.1 


3.3 


6.0 


6.4 


7.5 


1.0 


1.2 


1.8 


1.9 


1.5 


3.0 


3.1 


4.0 


0.4 


0.5 


0.8 


0.8 


0.6 


1.3 


1.7 


1.8 


0.2 


0.2 


0.3 


0.3 


0.2 


0.5 


0.5 


0.7 







0.1 


0.1 


0.1 


0.1 


0.1 



0.2 
0.1 



123 



USCOfSccnl IQ^ 



Enghnd 14^ 

Scdhnd 1434-^ 
Holland 14^^ 

U.STwhitcJ 13.77 - 

DanmarK I5e9 <^ 
Canada 1566 ^ 
^*^den 030 - 

Norv^ay I2.S8 - 
Belgium 18.79 - 

Jrvhnd IZZe ^ 
Austria I^.ZT^ 

TUrKeyl^OZ- 
Greece 1130 - 



18 



16 



17 



16 



16 



14 



17 



16 



le- 



w- 



is 



13 



/^sjia II.S4 
Italy 11.01 
AAtnd 10.74 
US(Cohnd) I04t 



12 



12 



11- 



I 10- 



1^4 



Figure 36. The relative standing of the nativity groups according 
to their average intelligence. The averages of the nativity groups 
are taken from Table 9. The averages of the vrhite officers and 
negro draft (from Table 1) and the native born white draft 
(from Table 2) are also shown. The left hand scale reads in 
units of the combined scale. The right hand scale reads in 
units of "mental age" representing what would be the approxi- 
mately equivalent scores on the Stanford revision of the Binet- 
Simon scale. In interpreting the differences shown, it must be 
remembered that all the differences are not equally reliable, for 
the reliability of the measurements depends on the number of 
cases in each group and the variability of the group. The relia- 
bility of all the differences is shown in Tables 11 to 27. 



12A 



Table No. 11 

Differences between ENGLAND and other countries 

Number of cases 411 
Average score 14.87 
Standard deviation 2.57 



Scotland 


-0.53 0.1695(3.1) Difference unreliable 


Holland 


-0.55 0.1606(3.4) Difference unreliable 


Germany 


-0.99 0.1264 (7.8) 


Denmark 


-1.18 0.1192(9.9) 


Canada 


-1.21 0.1030(11.7) 


Belgium 


-2.08 0.1669(12.5) 


United States 


-1.10 0.0855(12.9) 


Sweden 


-1.57 0.1049(14.9) 


Norway 


-1.89 0.1087(17.4) 


Austria 


-2.60 0.1368 (19.0) 


Turkey 


-2.85 0.1241 (22.9) 


Ireland 


-2.55 0.1094 (23.3) 


Greece 


-2.97 0.1097(27.0) 


Poland 


-4.13 0.1236(33.4) 


Russia 


-3.53 0.0940(37.5) 


Italy 


-3.86 0.0897(43.0) 



I 



126 



Table No. 12 



Differences between SCOTLAND and other countries 

Number of cases 146 
Average score 14.34 
Standard deviation 2.63 



England 

Holland 

Germany 

United States 

Denmark 

Canada 

Sweden 

Belgium 

Norway 

Austria 

Ireland 

Turkey 

Greece 

Russia 

Poland 

Italy 



+0.53 
-0.02 
-0.46 
-0.57 
-0.65 
-0.68 
-1.04 
-1.55 
-1.36 
-2.07 
-2.02 
-2.32 
-2.44 
-3.00 
-3.60 
-3.33 



0. 
0, 

















.1695 (3.1) 
.1999 (0.1) 
.1736 (2.6) 
.1466 (3.9) 
.1685 (3.9) 
.1575 (4.3) 
.1587 (6.5) 
.2050 (7.6) 
.1612 (8.4) 
.1813 (11.4) 
.1617 (12.5) 
.1720 (13.5) 
.1619 (15.0) 
.1517 (19.8) 
.1716 (20.9) 
.1491 (22.3) 



Difference unreliable. 
Difference unreliable. 
Difference unreliable. 
Difference unreliable. 
Difference unreliable. 



127 



Table No. 13 



Differences between HOLLAND and other countries 

Number of cases 140 
Average score 14.32 
Standard deviation 2.39 



England 

Scotland 

Germany 

Denmark 

United States 

Canada 

Sweden 

Belgium 

Norway 

Austria 

Ireland 

Turkey 

Greece 

Russia 

Poland 

Italy 



+0.55 
+0.02 
-0.44 
-0.63 
-0.55 
-0.66 
-1.02 
-1.53 
-1.34 
-2.05 
-2.00 
-2.30 
-2.42 
-2.98 
-3.58 
-3.31 




=tO 
0, 












0, 



1606 (3.4) 
1999 (0.1) 
1649 (2.7) 
1595 (3.9) 
1362 (4.0) 
1479 (4.4) 
1491 (6.8) 
1977 (7.7) 
1518 (8.8) 
1731 (11.8) 
1523 (13.1) 
1632 (14.1) 
1525 (15.9) 
1417 (21.0) 
1628 (21.9) 
1388 (23.9) 



Difference unreliable. 
Difference unreliable. 
Difference unreliable. 
Difference unreliable. 
Difference unreliable. 



1^8 



Table No. 14 



Differences between GERMANY and other countries 

Number of cases 308 
Average score 13.88 
Standard deviation 2.43 



England 
Holland 
Scotland 
United States 
Denmark 
Canada 
Sweden 
Belgium 
Norway- 
Austria 
Ireland 
Turkey 
Greece 
Poland 
Russia 
Italy 



+0.99 
+0.44 
+0.46 
-0.11 
-0.19 
-0.22 
-0.58 
-1.09 
-0.90 
-1.61 
-1.56 
-1.86 
-1.98 
-3.14 
-2.54 
-2.87 



0.1264 
0.1649 
0.1736 
0.0934 
0.1249 
0.1097 
0.1114 
0.1711 
0.1150 
0.1418 
0.1157 
0.1297 
0.1159 
0.1291 
0.1012 
0.0972 



(7.8) 

(2.7) 

(2.6) 

(1.1) 

(1.5) 

(2.0) 

(5.2) 

(6.3) 

(7.8) 

(11.3) 

(13.5) 

(14.3) 

(17.1) 

(24.3) 

(25.1) 

(29.5) 



Difference unreliable. 
Difference unreliable. 
Difference unreliable. 
Difference unreliable. 
Difference unreliable. 



129 



Table No. 15 



Differences between the UNITED STATES and other 
countries. (Native born white draft used.) 

Number of cases 81,465 
Average score 13.77 
Standard deviation 2.86 



England 

Holland 

Scotland 

Germany 

Denmark 

Canada 

Belgium 

Sweden 

Norway 

Austria 

Turkey 

Ireland 

Greece 

Poland 

Russia 

Italy 



+ 1.10 
+0.55 
+0.57 
+0.11 
-0.08 
-0.11 
-0.98 
-0.47 
-0.79 
-1.50 
-1.75 
-1.45 
-1.87 
-3.03 
-2.43 
-2.76 



0.0855 
0.1362 
0.1466 
0.0934 
0.0835 
0.0582 
0.1437 
0.0614 
0.0678 
0.1071 
0.0904 
0.0688 
0.0693 
0.0896 
0.0400 
0.0285 



(12.9) 

(4.0) 

(3.9) 

(1.1) 

(1.0) 

(1.9) 

(6.8) 

(7.6) 

(11.6) 

(14.0) 

(19.3) 

(21.2) 

(26.9) 

(33.8) 

(60.7) 

(96.8) 



Difference unreliable. 
Difference unreliable. 
Difference unreliable. 
Difference unreliable. 
Difference unreliable. 



ISO 



Table No. 16 
Differences between DENMARK and other countries 





Number of cases 325 




Average score 13.69 




Standard deviation 2.23 


England 


+ 1.18 0.1192(9.9) 


Scotland 


+0.65 0.1685(3.9) Difference unreliable. 


Holland 


+0.63 0.1595(3.9) Difference unreliable, 


Germany 


+0.19 0.1249(1.5) Difference unreliable. 


United States +0.08 0.0835 (1.0) Difference unreliable, 


Canada 


-0.03 0.1013(0.3) Difference unreliable. 


Sweden 


-0.39 0.1032(3.7) Difference unreliable. 


Belgium 


-0.90 0.1659(5.4) 


Norway 


-0.71 0.1071 (6.6) 


Austria 


-1.42 0.1355(10.5) 


Ireland 


-1.37 0.1079(12.7) 


Turkey 


-1.67 0.1228(13.6) 


Greece 


-1.79 0.1081(16.5) 


Poland 


-2.95 0.1221 (24.1) 


Russia 


-2.35 0.0921 (25.5) 


Italy 


-2.68 0.0877(30.5) 



131 



Table No. 17 
Differences between CANADA and other countries 

Number of cases 972 
Average score 13.66 
Standard deviation 2.67 



England 


+ 1.21 0.1030(11.7) 


Holland 


+0.66 0.1479(4.4) 


Scotland 


+0.68 0.1575(4.3) 


Germany 


+0.22 0.1097(2.0) Difference unreliable, 


United States +0.11 0.0582(1.9) Difference unreliable, 


Denmark 


+0.03 0.1013(0.3) Difference unreliable. 


Sweden 


-0.36 0.0840(4.3) 


Belgium 


-0.87 0.1547(5.6) 


Norway 


-0.68 0.0888(7.6) 


Austria 


-1.39 0.1215(11.4) 


Ireland 


-1.34 0.0896(14.9) 


Turkey 


-1.64 0,1071(15.3) 


Greece 


-1.76 0.0900(19.6) 


Poland 


-2.92 0.1064(27.4) 


Russia 


-2.32 0.0700(33.1) 


Italy 


-2.65 0.0641 (41.3) 



132 



Table No. 18 
Differences between SWEDEN and other countries 

Number of cases 691 
Average score 13.30 
Standard deviation 2.38 



England 


+ 1.57 0.1049(14.9) 


United States +0.47 0.0614 (7.6) 


Holland 


+ 1.02 0.1491(6.8) 


Scotland 


+ 1.04 0.1587(6.5) 


Germany 


+0.58 0.1114(5.2) 


Canada 


+0.36 0.0840(4.3) 


Denmark 


+0.39 0.1032(3.7) Difference unreliable, 


Belgium 


0.51 0.1559(3.3) Difference unreliable, 


Norway 


-0.32 0.0910(3.5) Difference unreliable. 


Austria 


-1.03 0.1231(8.4) 


Ireland 


-0.98 0.0918(10.6) 


Turkey 


-1.28 0.1089(11.8) 


Greece 


-1.40 0.0921(15.2) 


Poland 


-2.56 0.1082(23.6) 


Russia 


-1.96 0.0727(26.9) 


Italy 


-2.29 0.0671(34.1) 



133 



Table No. 19 
Differences between NORWAY and other countries 

Number of cases 611 
Average score 12.98 
Standard deviation 2.47 



England 


+ 1.89 =i= 0.1087 (17.4) 


United States +0.79 0.0678 (11.6) 


Holland 


+ 1.34 0.1518(8.8) 


Scotland 


+ 1.36 0.1612(8.4) 


Germany 


+ 0.90 0.1150(7.8) 


Canada 


+0.68 0.0888(7.6) 


Denmark 


+0.71 0.1071(6.6) 


Sweden 


+0.32 0.0910(3.5) Difference unreliable 


Belgium 


-0.19 0.1586(1.2) Difference unreliable 


Austria 


-0.71 0.1264(5.6) 


Ireland 


-0.66 0.0961 (6.8) 


Turkey 


-0.96 0.1126(8.5) 


Greece 


-1.08 0.0965(11.2) 


Poland 


-2.24 0.1119(20.1) 


Russia 


-1.64 0.0795(20.6) 


Italy 


-1.97 0.0729(27.0) 



134 



Table No. 20 



Differences between BELGIUM and other countries 

Number of cases 129 
Average score 12.79 
Standard deviation 2.42 



England 

Holland 

Scotland 

United States 

Germany 

Canada 

Denmark 

Sweden 

Norway 

Ireland 

Austria 

Turkey 

Greece 

Russia 

Poland 

Italy 



+2.08 
+ 1.53 
+ 1.55 
+0.98 
+ 1.09 
+0.87 
+0.90 
+0.51 
+0.19 
-0.47 
-0.52 
-0.77 
-0.89 
-1.45 
-2.05 
-1.78 








==0 






=*=o 









,1669 (12.5) 
,1977(7.7) 
,2050 (7.6) 
1437 (6.8) 
1711 (6.3) 
,1547 (5.6) 
1659 (5.4) 
,1559 (3.3) 
,1586 (1.2) 
1590 (2.9) 
,1789 (2.9) 
1695 (4.5) 
1592 (5.6) 
,1488 (9.8) 
,1691 (12.1) 
,1462 (12.2) 



Difference unreliable. 
Difference unreliable. 
Difference unreliable. 
Difference unreliable. 



135 



Table No. 21 
Differences between IRELAND and other countries 

Number of cases 658 
Average score 12.32 
Standard deviation 2.60 



England 


+2.55 0.1094(23.3) 


United States +1.45 0.0688 (21.2) 


Canada 


+ 1.34 0.0896(14.9) 


Germany 


+ 1.56 0.1157(13.5) 


Holland 


+2.00 0.1523(13.1) 


Denmark 


+ 1.37 0.1079(12.7) 


Scotland 


+2.02 0.1617(12.5) 


Sweden 


+0.98 0.0918(10.6) 


Norway 


+0.66 0.0961 (6.8) 


Belgium 


+0.47 0.1590(2.9) Difference unreliable 


Austria 


0.05 0.1269(0.4) Difference unreliable 


Turkey 


0.30 0.1132(2.7) Difference unreliable 


Greece 


-0.42 0.0972(4.3) 


Russia 


-0.98 0.0791 (12.4) 


Poland 


-1.58 0.1126(14.0) 


Italy 


-1.31 0.0739(17.7) 



1S6 



Table No. 22 



Differences between AUSTRIA and other countries 

Number of cases 301 
Average score 12.27 
Standard deviation 2.75 



England 

United States 

Holland 

Canada 

Scotland 

Germany 

Denmark 

Sweden 

Norway 

Belgium 

Ireland 

Turkey 

Greece 

Russia 

Poland 

Italy 



+2.60 
+ 1.50 
+2.05 
+ 1.39 
+2.07 
+ 1.61 
+ 1.42 
+ 1.03 
+0.71 
+0.52 
+0.05 
-0.25 
-0.37 
-0.93 
-1.53 
-1.26 



0. 
0. 
0. 
0. 
0. 
0. 
0. 
0. 
0. 
0. 
0. 
0. 
0. 
0. 
0. 
0. 



1368 (19.0) 
1071 (14.0) 
1731 (11.8) 
1215 (11.4) 
1813 (11.4) 
1418 (11.3) 
1355 (10.5) 
1231 (8.4) 
1264 (5.6) 
1789 (2.9) 
1269 (0.4) 
1398 (1.8) 
1272 (2.9) 
1139 (8.2) 
1393 (10.9) 
1104 (11.4) 



Difference unreliable. 
Difference unreliable. 
Difference unreliable. 
Difference unreliable. 



137 



Table No. 23 

Differences between TURKEY and ot^er countries 

Number of cases 423 
Average score 12.02 
Standard deviation 2.75 



England 


+2.85 0.1241 (22.9) 


United States +1.75 0.0904 (19.3) 


Canada 


+ 1.64 0.1071 (15.3) 


Germany 


+ 1.86 0.1297(14.3) 


Holland 


+2.30 0.1632(14.1) 


Denmark 


+ 1.67 0.1228(13.6) 


Scotland 


+2.32 0.1720(13.5) 


Sweden 


+ 1.28 0.1089(11.8) 


Norway 


+0.96 0.1126(8.5) 


Belgium 


+0.77 0.1695(4.5) 


Ireland 


+0.30 0.1132(2.7) Difference unreliable 


Austria 


+0.25 0.1398(1.8) Difference unreliable 


Greece 


-0.12 0.1135(1.0) Difference unreliable 


Russia 


-0.68 0.0984(6.9) 


Poland 


-1.28 0.1269(10.1) 


Italy 


-1.01 0.0944(10.7) 



138 



Table No. 24 
Differences between GREECE and other countries 

Number of cases 572 
Average score 11.90 
Standard deviation 2.45 



England 


+2.97 0.1097(27.0) 


United States +1.87 0.0693 (26.9) 


Canada 


+ 1.76 0.0900(19.6) 


Germany 


+ 1.98 0.1159(17.1) 


Denmark 


+ 1.79 0.1081(16.5) 


Holland 


+2.42 0.1525(15.9) 


Sweden 


+ 1.40 0.0921(15.2) 


Scotland 


+2.44 0.1619(15.0) 


Norway 


+ 1.08 0.0965(11.2) 


Belgium 


+0.89 0.1592(5.6) 


Ireland 


+0.42 0.0972(4.3) 


Austria 


+0.37 0.1272(2.9) Difference unreliable. 


Turkey 


+0.12 0.1135(1.0) Difference unreliable. 


Russia 


-0.56 0.0795(7.1) 


Poland 


-1.16 0.1129(10.3) 


Italy 


-0.89 0.0744(11.9) 



139 



Table No. 25 
Differences between RUSSIA and other countries 

Number of cases 2340 
Average score 11.34 
Standard deviation 2.83 

United States +2.43 0.0400 (60.7) 
England +3.53 0.0940 (37.5) 

Canada +2.32 0.0700 (33.1) 

Sweden +1.96 0.0727 (26.9) 

Denmark +2.35 0.0921 (25.5) 
Germany . +2.54 0.1012 (25.1) 

Holland +2.98 0.1417 (21.0) 

Norway +1.64 0.0795 (20.6) 

Scotland +3.00 0.1517 (19.8) 

Ireland +0.98 0.0791 (12.4) 

Belgium +1.45 0.1488 (9.8) 

Austria +0.93 0.1139 (8.2) 

Greece +0.56 0.0795 (7.1) 

Turkey +0.68 0.0984 (6.9) 

Poland -0.60 0.0977 (6.1) 

Italy -0.33 0.0483 (6.9) 



140 



Table No. 26 
Differences between ITALY and other countries 

Number of cases 4009 
Average score 11.01 
Standard deviation 2.60 

United States +2.76 0.0285 (96.8) 

England +3.86 0.0897 (43.0) 

Canada +2.65 0.0641 (41.3) 

Sweden +2.29 0.0671 (34.1) 

Denmark +2.68 0.0877 (30.5) 

Germany +2.87 0.0972 (29.5) 

Norway +1.97 0.0729 (27.0) 

Holland +3.31 0.1388(23.9) 

Scotland +3.33 0.1491 (22.3) 

Ireland +1.31 0.0739(17.7) 

Belgium +1.78 0.1462 (12.2) 

Austria +1.26 0.1104 (11.4) 

Turkey +1.01 0.0944(10.7) 

Greece +0.89 0.0744 (11.9) 

Russia +0.33 0.0483 (6.9) 

Poland -0.27 0.0936(2.9) Difference unreliable. 



141 



Table No. 27 
Differences between POLAND and other countries 

Number of cases 382 
Average score 10.74 
Standard deviation 2.59 

United States +3.03 0.0896 (33.8) 

England +4.13 0.1236 (33.4) 

Canada +2.92 0.1064 (27.4) 

Germany +3.14 0.1291 (24.3) 

Denmark +2.95 0.1221 (24.1) 

Sweden +2.56 0.1082 (23.6) 

Holland +3.58 0.1628 (21.9) 

Scotland +3.60 0.1716 (20.9) 

Norway +2.24 0.1119 (20.1) 

Ireland +1.58 0.1126 (14.0) 

Belgium +2.05 0.1691 (12.1) 

Austria +1.53 0.1393 (10.9) 

Greece +1.16 0.1129 (10.3) 

Turkey +1.28 0.1269 (10.1) 

Russia +0.60 0.0977 (6.1) 

Italy +0.27 0.0936(2.9) Difference unreliable. 



U3 



I 



AMERICAN INTELLIGENCE 143 

Tables 11 to 27 give the most accurate interpretation of 
the differences found between the various nativity groups 
that it is possible to derive from the army data. It is desir- 
able, however, to attempt to estimate the meaning of these 
differences in terms of standards which have some popular 
significance. For this reason, the combined scale distribu- 
tions in this study have been converted into estimates of 
the per cent, of A and B men in each group, and the per 
cent, of D, D and E men. 

It should be remembered that the army letter ratings 
are arbitrary ratings and have no real significance aside 
from the tests from which they were derived. The army 
rating "A" represents a certain score on the tests that 
should have been reached by 4% or 5% of the whole army 
group. In the same way the rating "B" was fixed so as to 
include the next 8% or 10%. It is of course absurd to deplore 
the fact that only 4% or 5% of the army were A men, when 
A was fixed so that only 4% or 5% could receive that rating. 
At the other end of the scale, the ratings D and D were 
fixed so that they would include approximately 20% of the 
total group, and the E rating was reserved for those recom- 
mended for development battalions, special service organi- 
zations, rejection or discharge. The estimates made at the 
time the examinations were being standardized proved to 
be about right. The A and B groups which should have in- 
cluded 12% to 15% of the draft actually included 12%, and 
the D, D and E groups, which should have included 20% 
to 25%, actually included 24%. 

Table 28 gives the per cent, of cases in each nativity 
group who would be classified A or B according to the 
criterion of the upper 12% of the total unselected white 
draft. Table 29 gives the per cent, who would be classified 
as D, D and E according to the criterion of 24% of the 
unselected group. The relations between the various nativ- 



144 AMERICAN INTELLIGENCE 

ity groups given in Tables 28 and 29 are shown graphically 
in Figure 37. 

Another criterion that probably represents intellectual 
ability of a high order is the per cent, at or above the 
average of the white officers. The classification of the 
nativity groups according to this criterion is given in Table 
30. At the other end of the scale, a criterion of inferiority 
that has n certain social significance is the per cent, at or 
below the average of the negro draft. Table 31 shows the 
nativity groups classified according to this criterion. The 
results given in Tables 30 and 31 are shown graphically in 
Figures 38 and 39. 

It is not possible to determine accurately the percentage 

Table No. 28 
Per cent, of each nativity group in the A and B groups 

England 19.0 

Native Born White Draft 13.2 

Scotland 13.1 

Holland 12.4 

Total White Draft 12.0 

Canada 11.1 

Germany 10.1 

Denmark 7.0 

Sweden 5.9 

Norway 5.3 

Foreign Born White Draft 4.6 

Ireland 4.3 

Austria 4.1 

Turkey 4.0 

Russia 3.3 

Belgium 2.9 

Greece 2.2 

Italy 1.5 

Colored Draft 1.4 

Poland 1.1 



AMERICAN INTELLIGENCE 145 

of feeble-mindedness in each group. The selection by the 
draft boards probably excluded all idiots and many im- 
beciles. The diagnosis of the border-line cases of feeble- 
mindedness is, in the last analysis, a social diagnosis, and 
can not be based on intelligence tests alone. It has been 
found, however, that a "mental age" of eight indicates an 
order of intelligence so low that the individual has diffi- 
culty in adjusting himself to his environment adequately. 
Table 32 gives a conservative estimate of the per cent, in 
each nativity group below an approximate "mental age" of 
eight. The percentages in Table 32 are shown graphically 
in Figure 40. 

Table No. 29 
Per cent, of each nativity group in the D, D , and E 

groups 

England 8.8 

Holland 12.0 

Scotland 13.5 

Germany 16.2 

Denmark 17.0 

Native Born White Draft 21.0 

Canada 21.6 

Sweden 23.2 

Total White Draft 24.1 

Norway 29.0 

Belgium 29.3 

Ireland 38.1 

Austria 38.4 

Turkey 43.6 

Greece 44.6 

Foreign Born White Draft 44.6 

Russia 55.7 

Italy 60.5 

Poland 63.8 

Colored Draft 67.5 



ENGLAND 

V//////////////////////////////////^^^^^ 

HOLLAND 

Y /////////////////////////////////////////////^^^^ 

SCOTLAND 



yy////////////////////////////////////^^^^^ - 

GERMANY 

y///////////////////////////////^^^^^ 

DENMARK 

y// / //////////////////////////////^^^^^ z 

NATIVE BORN WHITE DRAFT 

y//////////////////////////////////////////////^^^^ 

CANADA 

V/////////////////////////////////////////^^^^ 

SWEDEN 

V////////////////////////////////////^^^^ 

TOTAL WHITE DRAFT 




NORWAY 

V//////////////////////////////////////////////y^ 
BELGIUM 

W/////////////////////////////////////^^ I 

IRELAND 

y/////////////////////////////////////^^^^ I 

AUSTRIA 

VTTPTZ//////////////////// ////////^^^^ 

TURKEY 

W7/////////////////////////////////?7r\ 

GREECE 

V777P77/////////////////////////'/////A I 



FOREIGN BORN WHITE DRAFT 

MM II V//////////y/////A I 



RUSSIA 



ITALY 



POLAND 



V/////////// / /////////////A\ 
Wy///////////////////////A I 
I II 



NEGRO DRAFT 



20 30 40 50 60 

146 



100 



A&B 



Figure 37. The relative standing of the various nativity groups in 
the proportions of A and B men^ and D, D , and E men. In 
interpreting this charts it should be remembered that A and B^ 
and D, D , and E do not represent absolute intelligence stand- 
ards, but rather standards arbitrarily fixed. In this case the 
standards were fixed by the 93,955 individuals making up the 
sample of the total white draft, A and B representing scores 
obtained by the upper 12% of this group, while D, D , and E 
represent scores obtained by the lower 24%. The comparison is 
relative and not altogether reliable, for it fails to take into con- 
sideration the average, the number of cases, and the variability. 
Tables 28, 29, and this chart have been presented for the 
convenience of the reader, and to supplant Table 217, and Fig- 
ure 19, on pages 697 and 698 of Memoir XV, which are not 
based on combined scale results. 



147 



Table No. 30 

Per cent, of each nativity group at or above the average 
of the white officers 

England 6.2 

Scotland 4.8 

Native Born White Draft 4.6 

Holland. 4.3 

Total White Draft 4.1 

Canada 3.5 

Germany 3.4 

Austria 1.5 

Sweden 1.4 

Foreign Born White Draft 1.3 

Ireland 1.2 

Denmark 1.1 

Norway 1.0 

Turkey 0.8 

Russia 0.8 

Belgium 0.3 

Italy 0.3 

Greece 0.3 

Poland 0.1 



Figure 38. The proportion of each nativity group obtaining intel- 
ligence scores at or above the average of the white officers 
(18.84). Reference to Figure 31 will show that this criterion 
indicates a relatively high order of intelligence. In comparing 
this Figure with Figures 37 and 39, it should be noted that each 
of the three figures has been drawn to a different scale. Our in- 
terpretation of these figures must be made with caution, for we 
are comparing extremes of the distribution curves without ref- 
erence to the position of the average, the variability about the 
average, or the number of cases in the various groups. 

148 



07o 1% 2% 37o 4% 5% 67o 7% 



ENGLAND 

SCOTLAND 

NATIVE BORN WHITE DRAFT 

HOLLAND 

TOTAL WHITE DRAFT 

CANADA 

GERMANY 

AUSTRIA 

SWEDEN 

FOREIGN BORN WHITE DRAFT 

IRELAND 

DENMARK 

NORWAY 

TURKEY 

RUSSIA 

BELGIUM 

ITALY 

GREECE 

POLAND 




0% l7o 2% 



*% 5% Qfo 1% 



149 



Table No. 31 

Per cent, of each nativity group at or below the average 
of the negro draft 

England 4.3 

Holland 4.9 

Germany 6.5 

Scotland 6.8 

Denmark. '. 7.5 

Native Born White Draft. . 7.6 

Sweden 11.5 

Canada 11.6 

Total White Draft 13.7 

Norway 15.2 

Belgium 16.0 

Ireland 22.8 

Austria 24.5 

Greece 27.1 

Turkey 28.2 

Foreign Born White Draft 29.5 

Russia 39.0 

Italy 42.3 

Poland 46.0 



Figure 39. The proportion of each nativity group at or below the 
average of the negro draft. Reference to Figure 31 will show 
that this criterion indicates a rather low order of intelligence. 
If 50% of any nativity group were at or below the negro aver- 
age, the two distributions would be approximately identical. 
Russia shows 39% below the negro average, Italy 42.3%, and 
Poland 46%. 



150 



07o 



109& 209& 80% 40% 50^ 




0% \07o 207e 30% 40%- 50% 



151 



Table No. 32 



Per cent, of each nativity group below the approximate 
"mental age" of eight 

Holland 0.1 

Germany 0.2 

Denmark 0.2 

England 0.3 

Scotland. . : 0.4 

Sweden 0.4 

Norway 0.8 

Canada 0.9 

Native Born White Draft 1.1 

Total White Draft 1.4 

Belgium 1.6 

Ireland 1.9 

Greece 2.3 

Austria 2.7 

Turkey 3.1 

Foreign Born White Draft 3.2 

Russia 5.0 

Italy 5.2 

Poland 6.8 

Colored Draft 10.0 

Figure 40. Proportion of each nativity group testing below the ap- 
proximate "mental age" of eight. This criterion indicates intelli- 
gence of a very low order. These individuals are probably capa- 
ble of adjusting themselves only to the simplest form of environ- 
ment, occupation and conditions of living. Few of them would be 
able to manage their affairs with ordinary prudence. Many of 
them should be in custodial institutions. 



159 




POLAKD 
NEGRO DRAFT 



"1% 2% ^ *%> ^ 6% 7% 8% fo 10% n% 



1.W 



SECTION VII 
RELIABILITY OF THE RESULTS 



The results of the army psychological examination fig- 
ured by means of the combined scale give us the best avail- 
able measures of the intelligence of the various groups ex- 
amined. Do these results apply to the army as a whole? 
The logic underlying the answer to this question is the same 
as that underlying the judgment of the whole by one of 
its parts. The tea taster samples the tea to be graded. He 
does not need to brew a whole bale of tea to find its worth. 
In this experiment we have sampled the entire army by 
taking 15,543 white oflScers, 93,955 white recruits and 
23,596 negro recruits. Our group of white recruits was sub- 
divided into 81,465 native born and 12,492 foreign born. 

No one would hesitate to accept the results of the 
81,465 native born as typical of the army as a whole. If we 
continued sampling indefinitely, our results would increase 
in reliability only as the square root of the number of 
cases, and 81,465 cases constitute a sample that is a luxury 
from the point of view of size. In the same way, no one 
could seriously question the reliability of our sampling of 
15,543 oflScers, 23,596 negroes and 12,492 foreign born as 
typical of the remainder of the oflScers, negroes, and foreign 
born whites in the army. 

The results from the 81,465 cases in the native born 
white draft may be taken as typical of white males between 
the ages of 21 and 31 and above the idiot or imbecile grade. 
In making our comparisons between other groups, we 
know that the Selective Service Act called all men to the 
colors impartially. The same regulations drew the Italians, 
the negroes, the native whites, the Polish, and all other 

154 



AMERICAN INTELLIGENCE 155 

groups into the army. The method of sampling all the 
psychological records again drew these cases impartially. 
If our theory of sampling is correct, then we may accept 
the army results as very approximately typical of the male 
population as a whole. 

For instance, our figures in Table 3 show characteristic 
differences in the average score on the army tests of foreign 
born individuals in this country from to 5 yrs., and those 
in this country from 6 to 10 yrs., etc. The same factors 
which determined the sampling of the 3,576 cases in this 
country to 5 yrs. determined the sampling of the 4,287 
cases in this country 6 to 10 yrs. As long as the principles 
of sampling are the same, we may take our small sample as 
typical of the group as a whole. 

The results of the psychological tests of foreign born 
individuals classified according to length of residence, 
taken as typical of our foreign born population as a whole, 
indicate definitely that the average intelligence of succeed- 
ing waves of immigration has become progressively lower. 
Immigrants coming to this country between 1913 and 1917 
have a lower average intelligence than those coming to 
this country in the years 1908 to 1912. The group coming 
to this country in the years 1903 to 1907 had a higher 
average intelhgence that the 1908 to 1912 group, and a 
lower average intelligence than immigrants coming to this 
country in the years 1898 to 1902. 

In drawing these conclusions we are taking the groups 
examined in the army as typical of the corresponding 
groups in the entire population. During the years 1913 to 
1917, about 3 1/3 millons of immigrants came to this 
country. We are actually using 3,576 cases or about 0.1% 
as typical of the whole group. It may very properly be 
objected that this is too small a sample on which to base 
definite conclusions. We must therefore state our conclu- 



156 AMERICAN INTELLIGENCE 

sions less dogmatically, and with the proviso that if the 
groups examined in the army are typical of the immigra- 
tion coming to this country in the same periods, then we 
know that our more recent periods of immigration give 
us an average intelligence which becomes progressively 
lower and lower. This tentative conclusion will be modified 
by any evidence which tends to support the hypothesis 
made. 

The same kind of argument from the sample to the group 
holds in our interpretation of the differences in the average 
intelligence scores of groups in our army born in different 
countries. For instance, in the period under consideration 
from 1887 to 1917 there have been about 3 7/8 millions of 
Italians, and over 3 million Russians who have come to 
this country. We are actually using 4,009 Italians and 
2,340 Russians as typical samples of these groups. Of 
course no one would maintain that these 4,009 Italians are 
typical of the population of Italy. There are so many vari- 
able factors determining immigration that the immigrants 
can not themselves be taken as representative of the 
country as a whole. The question at issue is that of accept- 
ing 4,009 Italians as typical of the 3 7/8 millions who have 
come to this country since 1887. The chief claim to reli- 
ability of our sample from each country is the fact that the 
sample was drawn at random from the army group, and 
the fact that the Selective Service Act drew the men from 
each country impartially. 



SECTION VIII 
THE RACE HYPOTHESIS 

The results of the examination of the nativity groups 
suggest immediately that the race factor may underlie the 
large differences found. If we do find the common factor of 
race underlying the differences between the various nativ- 
ity groups, it will give our results much greater reliability, 
for the chance factors of sampling particularly inferior or 
superior groups in the small nativity samples would dis- 
appear in combination. Our figures are based on country 
of birth and no statistics are available for race. The race 
hypothesis must therefore be examined indirectly. 

Writers on immigration, for the most part, divide the 
countries of Europe into two groups (1) Northern and 
Western, and (2) Eastern and Southern, and usually as- 
sume that the immigration from Northern and Western 
Europe has been mostly Nordic. This traditional method 
is open to two very serious objections. In the first place, 
the classification fails to differentiate the Alpine and Medi- 
terranean race groups. In the second place the assumption 
that the immigration from Northern and Western Europe 
was mostly of a pure Nordic type is unwarranted, for this 
classification includes Germany and Ireland, two countries 
that have contributed very largely to our immigration in 
the past. The following figures show the size of the Irish 
and German immigration: 



157 



158 



AMERICAN INTELLIGENCE 







PER CENT. 


PER CENT. 


PER CENT. 




TOTAL 






FROM 


DECADE 


IMMIGRATION 


FROM 
IRELAND 


FROM 
GERMANY 


IRELAND AND 
GERMANY 


1820-1830 


143,439 


35% 


5% 


40% 


1831-1840 


599,125 


35% 


25% 

25% 


71% 


1841-1850 


1,171,251 


46% 


1851-1860 


2,598,214 


35% 


37% 


72% 


1801-1870 


2,314,824 


19% 


34% 


53% 


1871-1880 


2,812,191 


15% 


26% 


41% 


1881-1890 


. 5,246,613 


12% 


28% 


40% 


1891-1900 


3,844,420 


11% 


14% 


25% 


1901-1910 


8,795,386 


4% 


4% 


8% 


1911-1920 


5,735,811 


23^% 


2>^% 


5% 



These figures show clearly the fallacy of assuming that the 
immigration from Northern and Western Europe has been 
predominately Nordic, for Ireland is largely Mediterranean 
and Germany largely Alpine. 

If we wish to obtain even approximate estimates of the 
contributions of each of the three European races to our 
importations, it is necessary to abandon the Northern and 
Western, and Eastern and Southern classification and try 
another method. If it were possible to make even approxi- 
mate estimates of the percentage of Nordic, Alpine and 
Mediterranean blood in each of the European nations send- 
ing immigrants to this country, such approximate estimates 
would be very much superior to the present method. 

In collaboration with students of this subject, I have 
constructed Table 33 which contains tentative estimates of 
the present blood constitution of the countries sending im- 
migrants to this country. This table is, of course, only an 
approximation to the truth, and many persons will dis- 
agree with the estimates. For this reason, I am re-publishing 
in Table 34, Table 68, page 100, of the Statistical Abstract 
for the United States for 1920, which shows the arrivals of 
alien passengers and immigrants by nationalities and by 



Table No. 33 

Tentative estimates of the proportion of Nordic, Alpine 
and Mediterranean blood in each of the European coun- 
tries. 



Austria-Hungary 

Belgium 

Denmark 

France 

Germany 

Greece 

Italy 

Netherlands 

Norway 

Sweden 

Russia (including Poland) 

Poland 

Spain 

Portugal 

Roumania 

Switzerland 

Turkey (unclassified) 

Turkey (in Europe) (including Serbia, 

Montenegro and Bulgaria) 

Turkey (in Asia) 

England 

Ireland 

Scotland 

Wales 

British North America 



EB CENT. 


PER CENT. 


PER CENT. 


NORDIC 


ALPINE 


MEDITERRANEAN 


10 


90 





60 


40 





85 


15 





30 


55 


15 


40 


60 








15 


85 


5 


25 


70 


85 


15 





90 


10 





100 








5 


95 





10 


90 





10 


5 


85 


5 





95 





100 





35 


65 








20 


80 





60 


40 





10 


90 


80 





20 


30 





70 


85 





15 


40 





60 


60 


40 






159 



Table No. 34 

No. 68 ARRIVALS OF ALIEN PASSENGERS AND IMMIGRANTS, 1820 TO 1920: Bt Nationau- 

TIE8 AND BT DeCADES. 

[Sources: Records of the Bureau of Statistics prior to 1896; subsequently, reports of the Commissioner General 
of Immigration, Department of Labor. The figures represent "alien passengers" from Oct. 1, 1820, to Dec. 31, 
1867; "immdgrants" from Jan. 1, 1868, to date. 



COUNTBY OF LAST PERMANENT 
RESIDENCE 


OCT. 1, 1820, 
TO SEPT. 30, 1830 


OCT. 1, 1830, 
TO DEC. 31, 1840 


JAN. 1, 1841, 
TO DEC. 31, 1850 


JAN. 1, 1851, 
TO DEC. 31, 1860. 














27 

169 

8,497 

6,761 


22 

1,063 

45,575 

152,454 




5,074 

639 

77,262 

434,626 


4,738 


Denmark 


3,749 




76,358 


Germany 


951,667 


Greece 1 . ... 




Italy 


408 
1,078 

} 

91 
} 2,622 


2,253 
1,412 

1,201 

646 

2,954 


1,870 
8,251 

13,903 

656 

2,759 


9,231 


Netherlands 


10,789 


Norway 






20,931 


Russia, including Russian Poland^ 

Spain^ 


1,621 


Portugal* 


10,353 






Switzerland 


3,226 


4.821 


4,644 


25,011 


T'lirtpv in RiirnnpS 














United Kingdom: 

England 


22,167 

2,912 

50,724 


73,143 

2,667 

207,381 


263,332 

3,712 

780,719 


385,643 


Scotland 


38,331 




914,119 


Wales* 














Total United Kingdom 


75,803 


283,191 


1,047,763 


1,338,093 








43 


96 


155 


116 






Total Europe 


98,816 


495,688 


1,597,502 


2,452,657 


British North America^ 


2.277 
4,817 

105 
3,834 

531 


13,624 

6,599 

44 

12,301 

856 


41,723 

3,271 

368 

13,528 
3,579 


59,309 


Mexico^ 


3,078 




449 


West Indies: Bermuda and Miquelon. 


10,660 

1,224 






Total America^ 


11,564 


33,424 


62,469 


74,720 


Islands of the Atlantic 


352 


103 


337 


3,090 


China 


2 


8 


35 


41,397 


India^ 
























Other Asia 


8 


40 


47 


61 






Total Asia 


10 


48 


82 


41,458 




2 

16 

32,679 


9 

52 

69,801 


29 
55 

52,777 


158 


Total Africa 


210 


All other countries 


25,921 


Grand total 


143,439 


699,125 


1,713,251 


2,598,214 







ilncluded in "Europe, not specified," prior to 1891-1900. ^Includes also Finland after 1872. 
Includes Canary and Balearic Islands after 1900. , . , i * . 

Figures include the Azores and Cape Verde Islands after 1879, they being classed with Portugal so far as that 
country is separately shown. 

160 



JAN. 1, 1861, 


TEARS ENDED JUNE 30 


TO JUNE 30, 1870 


1871 TO 1880 


1881 TO 1890 


1891 TO 1900 


1901 TO 1910 


1911 TO 1920 


7.800 

6,734 

17,094 

85,984 

787,468 


72,969 

7,221 

31,771 

72,206 

718,182 


353,719 

20,177 

88,132 

50,464 

1,452,970 


597,047 

20,062 

52,670 

36,006 

543,922 

15,996 

655,694 

31,816 

( 95,265 

\ 230,679 

693,703 

( 6,723 

( 23,010 

14.559 

33.149 

2.562 


2,145,266 

41,635 

65,285 

73,379 

341,498 

167,519 

2,045,877 

48,262 

190,505 

249,534 

1.597,306 

27,935 

69,149 

53,008 

34,922 

118,202 


896,342 

33,746 

41,983 

61,897 

143,945 

184,201 


11,728 
9,102 

109,298 
4,536 
8,493 


55,759 
16,541 

211,245 

52,254 

9,893 


307,309 
53,701 

568,362 

265,088 

6,535 


1,109,524 
43,718 
66,395 
95,074 
921,957 
68,611 
89,732 
13,311 


23,286 


28,293 


81,988 


23,091 
77,098 










668,128 

38,768 

435,778 


460.479 

87,564 

436,871 


657,488 
149,869 
655,482 


271,094 
60,053 

403,496 
11,186 


388,017 

120,469 

339,065 

17,464 


249,944 
78,601 

145,937 
13,107 










1,042,674 


984.914 


1,462,839 


745,829 


865,015 


487,589 


210 


656 


10,318 


4,370 


1,719 


18,350 


2,064,407 


2.261,904 


4,721,602 


3,703,061 


8,136,016 


4,376.564 


153,871 

2,191 

96 

9,043 

1,396 


383.269 

5,362 

210 

13,957 

928 


392,802 

1,913 

462 

29,042 

2,304 


2,631 

746 

1,183 

35,040 

3,059 


179,226 

49,642 

8,112 

107,548 
17,280 


742,185 
219,004 

17,159 
123,424 

41,899 


166,597 


403,726 


426,523 


42,659 


361,808 


1,143,671 


3,446 


10,056 


15,798 
















64,301 


123,201 


61,711 


23,166 

26 

26,855 

8,398 

28,370 


20,605 

4,713 

129,797 

77,393 

11,059 


21,278 
2,082 








83,837 








79,389 


308 


622 


6,669 


5,973 


64,609 


123.823 


68,380 


86,815 


243,567 


192,559 


221 

312 

15,232 


10,913 

229 

1,540 


12,574 

437 

1,299 


8,793 
1,343 
1,749 


12,973 

7,368 

33,654 


13,427 
8,443 
1,147 


2,314,824 


2,812,191 


5,246,613 


3,844,420 


8,795,386 


6,735.811 



^Includes Serbia. Bulgaria, and Montenegro prior to 1920; included in "Europe, not specified," prior to 1891- 
1900; also, after 1919, Czechoslovakia, Poland, and the Kingdom of the Serbs, Croats, and Slovenes. 
Not separately stated prior to 1891-1900. 

'Immigrants from British North America and Mexico were not reported from 1886 to 1895, inclusive. 
8Not separately enumerated prior to 1899. 

161 



162 AMERICAN INTELLIGENCE 

decades from 1820 to 1920. My own Tables 9 and 10 give 
the distribution of the intelhgence scores on the combined 
scale for the nativity groups we are studying. Anybody 
who disagrees with the estimates given in Table 33 may 
take these tables and split them according to any other 
estimates he wishes to make. However, minor changes in 
the proportions given in Table 33 would make very little dif- 
ference in, the final results. The figures which follow are 
merely estimates based on Table 33. I am not claiming 
that these figures are absolutely reliable, but merely that 
they represent very much closer approximations to the 
truth than would be obtained from the Northern and West- 
ern, and Southern and Eastern classification. 

To obtain an estimate of the proportion of Nordic, Al- 
pine, and Mediterranean blood in our immigration since 
1840, the immigration figures by countries, given in Table 
34, have been cut according to the proportions given in 
Table 33 and re-combined into percentage estimates which 
are given in Table 35. These estimates show in general an 
immigration prior to 1890 which ran 40% or 50% Nordic 
blood. 

Since 1890, the proportion of Nordic blood has dropped 
to 20% or 25%, the Alpine stock now constituting about 
50% of the total and the Mediterranean 20% or 25%. 

The proportions given in Table 35 are shown graphically 
in Figure 41. The percentage estimates, given in Figure 35 
and shown graphically in Figure 41, should be considered 
in connection with the total volume of immigration for 
each decade given in Table 34 and shown graphically in 
Figure 42. 



Table No. 35 

Estimate of the amount of Nordic, Alpine and Mediter- 
ranean blood coming to this country from Europe in each 
decade since 1840. 







PEB CENT. 


PEB CENT. 


PER CENT 


PER CENT. 




TOTAL 


NORDIC 


ALPINE 


MEDITERRANEAN 


OTHERS AND 


DECADE 


IMMIGRATION 


BLOOD 


BLOOD 


BLOOD 


UNCLASSIFIED 


1841-1850 


1,713,251 


40.5 


19.0 


36.2 


4.3 


1851-1860 


2,598,214 


42.3 


25.5 


28.9 


S.3 


1861-1870 


2,314,824 


50.6 


26.0 


19.2 


4.2 


1871-1880 


2,812,191 


48.8 


28.5 


16.7 


6.0 


1881-1890 


5,246,613 


46.1 


85.2 


16.0 


2.7 


1891-1900 


3,844,420 


30.2 


43.8 


22.5 


8.5 


1901-1910 


8,795,386 


19.8 


51.3 


24.3 


4.6 


1911-1920 


5,735,811 


22.6 


44.0 


23.7 


9.7 



163 




^^ 


S 






z 


1 






< 








H 
















tf 




u 


1^ 


tf 




1 s 


;5 


H 


^ I 


Q 


H 





1 


PU 


*^ 


W ' 


1 o 


H^ 


Q 


W ' 


z 


^ 


1 


,g' 



go ^^13^^ 

^ en 43 G^_fl ^ 

o a, ^ N '^ ^ ^ 

S ^ ^ vQ O (u 
1? > X 



.t^ C 550 3 (^ - -g 0^ 



^ C 3 W 

03 a a g 

M ^! '^ "73 -^ ^ '*^ 



SV, f:2 a fl <i^ -=? " 
-^ a; 9 o .2 



O ^ c3 O CJ >H 
. - ^ =3 



^ ^ '3 -^ -^ - H 

:3 O fl >H rj O 
^' ^ 5yo 0, - .S o 



C3 C S3 






en 

> "Jj O -N >^ 

o a^ 



w S e_ 



>> O 

.I 

03 a; 



.2 o "^iij M 

CO ^j 

<U CO >! 



H a _ri ,_( O ' I 

oc > <u a g 

b3 _^ > O c3 ^ 






o G 



a :: 



CO 
CO 

^ ^ .3^ 1 S p:5 



-^ '-^ ^ 



. 2 o S>~ C 

553 ^ r^ -H ^ a 

cj ^ <i; ^ o '^ 

-^ rrt -^ T3 ^ 

.2 n3 fl w a < 

-M o c3 3 ^ 
C3 ^ ^ > ^ 

^^ S o o o 

2 a T3 00 '-' ? 

-' <^ 3 o W O 

^ ;3 c ^ .9 o 



00 



05 ^pq C5 



. ^ S ^ O T! . 



-~ w (U 

s ;r -^ -^ 






168 AMERICAN INTELLIGENCE 

In order to obtain an estimate of the intelligence of the 
three European races in this country, the distributions of 
the intelligence scores on the combined scale given in 
Table 9 were cut according to the proportions given in 
Table 33, and re-combined into Nordic, Alpine, and Medi- 
terranean groups. The final distributions are, of course, 
neither purely Nordic, Alpine, nor Mediterranean, but the 
sample of individuals we have thus selected as Nordic is 
undoubtedly more typical of the Nordic race type than 
it is of the Alpine and Mediterranean types. In the same 
way, the Alpine and Mediterranean groups are more typi- 
cal of each of these race types than they are of either of the 
other two. With thus much of apology for the method, I 
will, in the following pages, simply for brevity of expres- 
sion, call these groups Nordic, Alpine, and Mediterranean. 
The reader must bear in mind that the distributions are 
only approximate samplings. 

The actual distributions on the combined scale of the 
three race groups so selected are given in Table 36, togeth- 
er with the proportions in each thousand. The distribution 
curves of the three groups are shown in Figure 43, in which 
the horizontal direction represents scores on the com- 
bined scale, and the vertical direction proportions in each 
thousand making each intelligence score. 

The differences found are very marked. The difference 
between the Nordic and Alpine group is 1.61 =1=0.042, a 
difference which is 38.3 times the probable error of the 
difference. The difference between the Nordic and Medi- 
terranean group is 1 .85 =*= 0.042, a difference which is 44 times 
the probable error of the difference. The Alpine and Medi- 
terranean groups are, on the other hand, very much closer 
together, the difference being 0.24 =*= 0.04, a difference which 
is 6 times the probable error of the difference. 

The easiest and most obvious objection that can be made 



! 



Table No. 36 

Analysis of the foreign born white draft by races. Distri- 
butions of the inteUigence scores of the Nordic, Alpine 
and Mediterranean groups. 



COMBINED 


ACTUAL DISTRIBUTION 


PROPORTION IN EACH 


SCALE 










THOSUAND 




INTERVALS 


NORDIC 


ALPINE 


MEDITER- 
RANEAN 


NORDIC 


ALPINE 


MEDITER- 
RANEAN 


24.0-24.9 














23.0-23.9 




.... 


.... 


.... 


.... 


.... 


22.0-22.9 


"3 


1 


1 




.... 




21.0-21.9 


8 


5 


2 


2 






20.0-20.9 


19 


11 


5 


5 


2 


"'2 


19.0-19.9 


37 


22 


11 


11 


5 


3 


18.0-18.9 


71 


47 


26 


21 


10 


6 


17.0-17.9 


135 


90 


55 


39 


19 


13 


16.0-10.9 


238 


155 


103 


69 


32 


24 


15.0-15.9 


357 


246 


180 


103 


51 


43 


14.0-14.9 


469 


372 


296 


136 


78 


71 


13.0-13.9 


566 


544 


408 


164 


114 


111 


12.0-12.9 


528 


650 


591 


153 


136 


141 


11.0-11.9 


371 


628 


590 


107 


132 


140 


10.0-10.9 


260 


595 


509 


75 


125 


136 


9.0- 9.9 


184 


546 


523 


53 


115 


125 


8.0- 8.9 


112 


403 


376 


32 


85 


90 


7.0- 7.9 


59 


248 


223 


17 


52 


53 


6.0- 6,9 


26 


124 


108 


8 


26 


26 


5.0- 5.9 


9 


52 


47 


3 


11 


11 


4.0- 4.9 


3 


19 


16 


1 


4 


4 


3.0- 3.9 


1 


6 


5 


.... 


2 


1 


2.0- 2.9 




2 


1 


.... 


.... 


.... 


1.0- 1.9 


.... 


.... 


.... 


.... 


.... 


.... 


No. of cases. 


3456 


4766 


4196 








Average .... 


13.28 


11.67 


11.43 








S.D 


2.70 


2.87 


2.70 










d 3i S 

s 2 o 2 

w ^ -S^ ^ 5^ g 

5 ;^ "^ *^. ^ 

O "P ^ CO <u 

*^ *H 0) a; <u 

V "^ <^ 5 ^ 

I- ^2:^ ? 



O H W . 

C 03 ^ d <^ 
DC o (U OJ >^ 



<u 



O e3 : 0) -^ 

C a O c3 o 

'C -^ ^ ^ 

^ o ^ 

"^ ^g MH <D ^ 

^^^ a; 



AMERICAN INTELLIGENCE 171 

to these findings is that the superiority of the Nordic group 
is due to the fact that it contains so many EngUsh speaking 
persons, and that lack of facihty in the use of EngHsh is a 
handicap to the non-Enghsh speaking foreign born in the 
army tests. We have previously examined this hypothesis 
in connection with the argument establishing the fact that 
each succeeding five year period since 1902 shows a gradual 
deterioration in the intelligence of the immigrants examined 
in the army, and have definitely shown that the language 
factor does not distort the scores of the years of residence 
groups. There is, however, a considerable amount of wish- 
ful thinking on the subject of race, and it is well to make 
assurance doubly sure by testing the hypothesis that the 
superiority of the Nordic group is caused by the presence 
in the group of English speaking populations. 

It is possible to split the Nordic distribution in such a 
way that one group will contain representatives from 
countries which are predominantly English speaking (Eng- 
land, Scotland, Ireland and Canada), while the other group 
will contain representatives from countries which are pre- 
dominantly non-English speaking (Holland, Denmark, 
Germany, Sweden, Norway, Belgium, Austria, Russia, 
Italy and Poland) . This we have done, and the results are 
given in Table 37, the two distributions being shown in 
Figure 44. 

The distributions of the English speaking Nordic group 
and the non-English speaking Nordic group show a differ- 
ence of 0.87=1=0.065, a difference which is 13.4 times the 
probable error of the difference. There are, of course, cogent 
historical and sociological reasons accounting for the in- 
feriority of the non-English speaking Nordic group. On the 
other hand, if one wishes to deny, in the teeth of the facts, 
the superiority of the Nordic race on the ground that the 
language factor mysteriously aids this group when tested, 



Table No. 37 

Analysis of the total Nordic sample into an English speak- 
ing Nordic group and a non-English speaking Nordic 
group. 



COMBINED 


ACTUAL DISTRIBUTION 


PROPORTION IN EACH 


SCALE 






THOUSAND 


INTEKVAT,a 


ENGLISH 


NON-ENGLISH 


ENGLISH 


NON-ENGLISH 




SPEAKING 


SPEAKING 


SPEAKING 


SPEAKING 




NORDIC 


NORDIC 


NORDIC 


NORDIC 


24.0-24.9 


.... 








23.0-23.9 










22.0-22.9 


2 




"2 


.... 


21.0-21.9 


7 


' k 


6 




20.0-20.9 


12 


6 


10 


8 


19.0-19.9 


21 


16 


17 


7 


18.0-18.9 


39 


82 


32 


14 


17.0-17.9 


67 


67 


54 


30 


16.0-16.9 


108 


181 


87 


59 


15.0-15.9 


143 


214 


116 


96 


14.0-14.9 


176 


298 


148 


132 


13.0-13.9 


201 


865 


168 


164 


12.0-12.9 


172 


856 


189 


160 


11.0-11.9 


109 


262 


88 


118 


10.0-10.9 


70 


189 


57 


85 


9.0- 9.9 


49 


185 


40 


61 


8.0- 8.9 


31 


82 


25 


87 


7.0- 7.9 


16 


48 


18 


19 


6.0- 6.9 


7 


19 


6 


9 


5.0- 5.9 


2 




2 


8 


4.0- 4.9 


1 




.... 


2 


3.0- 3.9 


.... 




.... 





2.0- 2.9 


.... 


.... 


.... 


.... 


1.0- 1.9 


.... 


.... 


.... 


.... 


No. of cases. . . 


1234 


2222 






Average 


13.84 


12.97 






S.D 


2.79 


2.60 







173 




O <L> jQ pfi .JH 



be ^ O 



52 ^-Sst! 



2 .2 o bcna 



^ i3!tj ^ t: 

^ E2 :i b^ S S 



JZ! i-M 'ft 

si " ""I 

ej .rt ^_> -^ 

1 O O g ^ C 

O O bC-^ g^ 

.^^ g S. 2 .3^ 

.s :3 .S^ ;S ^ ^ 

'^ 'IS . .S ^ V5 



d 'bc5 *<-i 



';P go 



rt .S ^ ;=3 H c 



- _r .-;:; fl" C 

QJ "^ i-H jj ~ 



-< -' tJ r! 

tn < Ji o 



{>..S:J O 






S fl a. 



^ p g .^:S S S^ 
V. o .S ^ 9^ o 

- \g So 



S JH 03 
.Oi bC^ 



174 AMERICAN INTELLIGENCE 

he may cut out of the Nordic distribution the English 
speaking Nordics, and still find a marked superiority of the 
non-English speaking Nordics over the Alpine and Medi- 
terranean groups. The difference between the non-English 
speaking Nordic group and the Alpine group is 1.30 =*= 0.047, 
a difference which is 27.6 times the probable error of the 
difference. The difference between the non-English speak- 
ing Nordic group and the Mediterranean group is 1.54=1= 
0.047, a difference which is 31.3 times the probable error 
of the difference. The distributions are shown graphically 
in Figure 45. Discarding the English speaking Nordics 
entirely, we still find tremendous differences between the 
non-English speaking Nordic group and the Alpine and 
Mediterranean groups, a fact which clearly indicates that 
the underlying cause of the nativity differences we have 
shown is race, and not language. 

It may be convenient for some to interpret the differ- 
ences found between the representatives of the three Euro- 
pean races in this country in terms of the standards having 
popular significance which were used in Section VI. The cri- 
teria of the per cent. A and B, and the per cent. D, D 
and E give the following results: 

PER CENT. PER CENT. 

A AND B D,D AND E 

English speaking Nordic 12 . 3 19.9 

Total Nordic 8.1 25.8 

Non-English speaking Nordic . 5.7 29 . 1 

Alpine 3.8 50.3 

Mediterranean 2.5 53 . 6 

The criteria of the per cent, at or above the average 
white officer, and at or below the average of the negro 
draft give the following results: 




S3 t^ 






-^ ^^ 






C -M 



o o 



bf) 'J5 a C O rQ c >j 

Q, a> <u ;:5 "^ -^ ^^ ^ 

'^^ fl '^ _- -^-' Jzj 
<v .ti -^f^ Ti -^ *' 

Ph.22 "*^ ea "1^ o 
c 

5 "XJ . >^ C oj a 
2 ^ d .t^ ^ C C 

g ;^. ^ .2 P -^^g^ 
g bc rt S 2 *^ ^ 
c .S ^ &- ^ ^ 

o ^ ;:5 bc q ?J 0) 
^W I i S " c 
-M G o 2; = o fl 

S -p. ^ ^ ^ ^ 2 

^ < ^ -B '? ^ a ^ 

bo p.^ S c' o o 

I H j^ M M 






^ .2 






be Sf 

C3 rO 



o o 

bC'-S 



o ^ 



176 AMERICAN INTELLIGENCE 

PER CENT. PER CENT. 

AT OR ABOVE AT OR BELOW 

AVERAGE AVERAGE OF 

WHITE THE NEGRO 
OFFICER DRAFT 

English speaking Nordic 4.0 10 . 9 

Total Nordic 2.3 14.5 

Non-English speaking Nordic . 1.3 16.5 

Alpine 1.0 34.5 

Mediterranean 0.5 36 . 5 

The criterion of the per cent, below an approximate 
**mental age" of eight gives the following results: 

PER CENT. 
BELOW 

"mental 
age" 8 

English speaking Nordic 0.8 

Total Nordic 1.1 

Non-English speaking Nordic . 1.3 

Alpine 4.2 

Mediterranean 4.2 



SECTION IX 

RE-EXAMINATION OF PREVIOUS CONCLUSIONS 
IN THE LIGHT OF THE RACE HYPOTHESIS 

It is now necessary to retrace our steps for a moment to 
examine some of our previous conclusions in the light of 
this new hypothesis. The hypothesis that the differences 
between the nativity groups found in the army tests are 
due to the race factor may be used to re-test our previous 
conclusions that each succeeding five year period of immi- 
gration since 1902 has given us an increasingly inferior 
selection of individuals (Section IV) . The periods which we 
sample by means of the army data, and the average score 
on the combined scale of each sample are as follows : 



PERIOD 


NUMBER OF 
CASES 


COMBINED 

SCALE 
AVERAGE 


1887-1897 


764 


13.82 


1898-1902 


771 


13.55 


1903-1907 


1897 


12.47 


1908-1912 


4287 


11.74 


1913-1917 


3576 


11.41 



Table 35, which gives our estimates of the per cent, of 
Nordic, Alpine and Mediterranean blood coming to this 
country, shows that the big change in immigration came 
between the decades 1881-1890 and 1891-1900, the per- 
centage of Nordic blood which formerly ran from 40% to 
50% having dropped to 30% in the decade 1891-1900, and 
to approximately 20% or 25% in the two subsequent dec- 
ades. On the other hand, the big drop in the intelligence 
of immigrants arriving came after 1902. The change in 

177 



178 AMERICAN INTELLIGENCE 

character of the immigration would account for part of the 
decHne in the average intelligence of succeeding periods of 
immigration, but not for all of it. The decline in intelligence 
is due to two factors, the change in the races migrating to 
this country, and to the additional factor of the sending 
of lower and lower representatives of each race. 

The only tendency which would relieve this deplorable 
situation would be a current of emigration strong enough 
to counteract the current of immigration. Table 6 preced- 
ing shows the ratio between emigration and immigration 
for each of the nativity groups involved in this study, and 
we find in general between 1908 and 1917 a return current 
approximately one third of the arriving current. 

Unfortunately, no emigration statistics are available 
prior to 1908, and the figures after 1912 are distorted by 
the Balkan and European wars. The only sample that we 
can take that is comparatively free from outside influences 
is the sample 1908-1912. Taking the figures of arrivals and 
departures for this period, and dividing them into Nordic, 
Alpine and Mediterranean groups according to the method 
previously outlined, we obtain the following percentage 
estimates : 





ALIEN 


ALIEN 






TMMTGRANTS 
ADMITTED 


EMIGRANTS 
DEPARTED 


NET 
IMMIGRATION 


Per cent, of Nordic 








blood 


21.2 


16.0 


23.9 


Per cent, of Alpine 








blood 


50.4 


50.6 


50.2 


Per cent, of Mediter- 








ranean blood 


23.2 


28.6 


20.5 


Per cent, others and 








unclassified 


5.2 


4.8 


5.4 



AMERICAN INTELLIGENCE 179 

The sample from this five year period shows a shght 
change (approximately 3%) in favor of the Nordic type 
and against the Mediterranean type, the Alpine immigra- 
tion holding its own. There is therefore no relief from our 
receding curve of intelHgence from emigration, if this five 
year period be taken as typical of the outward alien pas- 
senger movement in other years. 

I?- It will be remembered that the army authors tentatively 
offered the hypothesis that the more intelhgent immigrants 
remained in this country, while the more stupid ones went 
home, as a possible method of accounting for the increase 
of intelhgence scores with increasing years of residence. 
The gain of 3% in favor of the Nordic immigration would 
produce a very slight tendency in this direction, but not 
enough to account for the actual increase of intelligence 
scores found with increasing years of residence, 11.41 
(1913-1917) to 13.82 (1887-1897). 

It will also be remembered that the army writers offered 
the hypothesis of the better adaptation of the more thor- 
oughly Americanized group to the situation of the examin- 
ation to account for the increases shown. The factor of the 
adaptation to the situation of the examination cannot be 
dissected out of the total scores of the test. If such a factor 
were present, it would fall equally heavity on Nordic, Alpine 
and Mediterranean alike, unless the change in the character 
of immigration were so complete that the groups sampled 
at the two extremes of the residence groups (1887-1897 
and 1913-1917) represented different race groups. 

But the difference between these two years of residence 
groups (2.41 =t 0.0735) is so marked that it would be neces- 
sary to assume (if our Nordic group were the more thor- 
oughly Americanized) that the 1887-1897 group was com- 
posed entirely of English speaking Nordics or their equiva- 
lent in intelligence, and that our 1913-1917 group was 



180 AMERICAN INTELLIGENCE 

composed entirely of Mediterraneans or their equivalent 
in intelligence, assumptions quite unwarranted in view 
of the fact that in the two years of residence groups 1887- 
1897 and 1898-1902 we sampled 1545 individuals, while 
our Nordic group includes 3456 cases, and also in view of 
the fact that the Nordic immigration has dropped, in the 
period observed, at the outside from 45% to 20%. We may 
therefore conclude that the intangible factor of "the more 
thoroughly Americanized group" can not be used to ex- 
plain the high test record of the Nordic group. 

There is only one other possible escape from the con- 
clusion that our test results indicate a genuine intellectual 
superiority of the Nordic group over the Alpine and Medi- 
terranean groups, and that is the assumption that the 
situation of the examination involved a situation that was 
"typically Nordic." This assumption of course lands us in 
a perfect circle of reasoning. It would leave us with the 
conclusion that there was something mysteriously Nordic 
about alpha and beta that favored this race. We should 
have to assume that the Nordic, no matter where he may 
be, in the Canadian Northwest, in the Highlands of Scot- 
land, or on the shores of the Baltic, is always ready for an 
intelligence test. Perhaps it would be easier to say that the 
Nordic is intelligent. A situation "typically Nordic" could 
not be used, however, to account for the slight but real dif- 
ference between the English speaking Nordic and the non- 
English speaking Nordic groups. It is therefore best to 
abandon the attempt to account for the differences by the 
more or less feeble hypotheses that would make these dif- 
ferences an artifact of the method of examining, and recog- 
nize the fact that we are dealing with real differences in the 
intelligence of immigrants coming to our shores. 

We have previously noted the fact that the foreign born 
in the army sampled as representative of the immigrants 



AMERICAN INTELLIGENCE 181 

coming to this country between 1887 and 1897 were statis- 
tically identical with the native born white draft. The 
change in the character of our immigration came between 
1890 and 1900. The real drop in the curve of intelligence, 
however, started about 1900. We, therefore, cannot account 
for the drop in the intelligence of the immigrants sampled 
as representatives of those coming to this country in each 
five year period since 1902 by the race hypothesis entirely. 



SECTION X 

COMPARISON OF OUR RESULTS WITH THE 

CONCLUSIONS OF OTHER WRITERS 

ON THE SUBJECT 

In a very definite way, the results which we obtain by 
interpreting the army data by means of the race hypothesis 
support Mr. Madison Grant's i thesis of the superiority of 
the Nordic type: "The Nordics are, all over the world, a 
race of soldiers, sailors, adventurers, and explorers, but 
above all, of rulers, organizers, and aristocrats in sharp 
contrast to the essentially peasant and democratic char- 
acter of the Alpines. The Nordic race is domineering, in- 
dividualistic, self-reliant, and jealous of their personal 
freedom both in political and religious systems, and as a 
result they are usually Protestants. Chivalry and knight- 
hood and their still surviving but greatly impaired counter- 
parts are peculiarly Nordic traits, and feudalism, class 
distinctions, and race pride among Europeans are traceable 
for the most part to the north." (p. 228.) "The pure Nordic 
peoples are characterized by a greater stability and steadi- 
ness than are mixed peoples such as the Irish, the ancient 
Gauls, and the Athenians, among all of whom the lack of 
these qualities was balanced by a correspondingly greater 
versatility." (pp. 228-229.) 

Our results based on the army data also support Mr. 
Grant's estimates of the Alpine race: "The Alpine race is 
always and everywhere a race of peasants, an agricultural 
and never a maritime race. In fact they only extend to 
salt water at the head of the Adriatic and, like all purely 

iMadison Graot. The Passing of the Great Race. New York, 1922, Pp. 476. 

182 



AMERICAN INTELLIGENCE 183 

agricultural communities throughout Europe, tend toward 
democracy, although they are submissive to authority both 
political and religious, being usually Roman Catholics in 
western Europe. This race is essentially of the soil, and 
in towns the type is mediocre and bourgeois." (p. ^27.) 

Our results also support de Lapouge^ in his contention 
that the Nordic type is superior to the Alpine. He says con- 
cerning the Alpine: **I1 est le parfait esclave, le serf ideal, 
le sujet modele, et dans les republiques comme la notre, 
le citoyen le mieux vu, car il tolere tons les abus." (p. 233.) 
"Les etats brachycephales, France, Autriche, Turquie, sans 
parler de la Pologne qui n'est plus, sont loin d'offrir la 
vitalite des Etats -Unis ou de I'Angleterre. Cependant 
la mediocrite meme du brachycephale est une force. Ce 
neutre echappe a toutes les causes de destruction. Noiraud, 
courtaud, lourdaud, le brachycephale regne aujourd'hui 
del'Atlantique a la Mer Noire. Comme la mauvaise monnaie 
chasse Tautre, sa race a supplante la race meilleure. II est 
inerte, il est mediocre, mais se multiplie. Sa patience est 
au-dessus des epreuves; il est sujet soumis, soldat passif, 
fonctionnaire obeissant. II ne porte pas ombrage, il ne se 
revolte point." (p. 481.) 

It must, however, be frankly admitted that our results, 
which show the Mediterranean race inferior to the Alpine, 
are in contradiction with those of most writers who have 
inferred the intellectual level of a race from its historical 
achievements. Mr. Grant, for instance, says: "The mental 
characteristics of the Mediterranean race are well known, 
and this race, while inferior in bodily stamina to both the 
Nordic and the Alpine, is probably the superior of both, 
certainly of the Alpines, in intellectual attainments. In the 
field of art its superiority to both the other European races 

1 Georges Vacher de Lapouge. UAryen, son role social. Paris, 1899, Pp. 563. 



184 AMERICAN INTELLIGENCE 

is unquestioned, although in hterature and in scientific re- 
search and discovery the Nordics far excel it." (p. 229). ^ 

The apparent contradiction between our results and the 
estimates of other observers has a very obvious solution, 
viz., that those who draw their conclusions from historical 
data are studying the Mediterranean race as it was at the 
period of its greatest development, when it produced the 
civilizations of Egypt, Mesopotamia, Phoenicia, and Crete, 
and, with a Nordic predominance, gave the civilizations 
of Greece and Rome, while our data sample this race group 
as it is at the present time. 

The sample we have taken as representative of the Medi- 
terranean race, as it is now constituted, is drawn from im- 
migrants in our army born, for the most part, in Greece, 
Ireland, Italy, and Turkey, and inasmuch as the number 
from Italy (4009) is so large, our Mediterranean sample 
is heavily weighted (approximately 2/3) by this nativity 
group. 

In regard to the Irish, Mr. Madison Grant says: "In 
spite of the fact that Paleoliths have not been found there, 
some indications of Paleolithic man appear in Ireland, both 
as single characters and as individuals. Being, like Brittany 
situated on the extreme western outposts of Eurasia, it has 
more than its share of generalized and low types surviving 
in the living populations, and these types, the Firbolgs, 
have imparted a distinct and very undesirable aspect to a 
large portion of the inhabitants of the west and south and 
have greatly lowered the intellectual status of the popula- 
tion as a whole. The cross between these elements and the 



^The quotations I have chosen from Mr. Madison Grant's chapter on Racial 
Aptitudes most certainly do not do justice to that author, but they seemed to 
me to summarize his general position briefly. The entire book should be read to 
appreciate the soundness of Mr. Grant's position and the compelling force of his 
arguments. 



AMERICAN INTELLIGENCE 185 

Nordics appears to be a bad one, and the mental and cul- 
tural traits of the aborigines have proved to be exceedingly 
persistent and appear especially in the unstable tempera- 
ment and the lack of coordinating and reasoning power, 
so often found among the Irish. To the dominance of the 
Mediterraneans mixed with Pre-Neolithic survivals in the 
south and west are to be attributed the aloofness of the is- 
land from the general trend of European civilization and 
its long adherence to ancient forms of religion and even to 
Pre-Christian superstitions." (pp. 202-203.) 

The immigrants in this country from Italy come mostly 
from Southern Italy and Sicily. The following quotation 
from Ripley 1 concerning Sicily is significant: 

"Commanding both straits at the waist of the Mediter- 
ranean, it has been, as Freeman in his masterly description 
puts it, 'the meeting place of the nations.' Tempting, there- 
fore, and accessible, this island has been incessantly over- 
run by invaders from all over Europe Sicani, Siculi, Feni- 
cii, Greeks, and Romans, followed by Albanians, Vandals, 
Goths, Saracens, Normans, and last by the French and 
Spaniards. Is it any wonder that its people are less pure in 
physical type than the Sardinians or even the Calabrians of 
the mainland near by.^^ Especially is this noticeable on its 
southern coasts, always more open to colonization than the 
northern edge. Nor is it surprising, as Freeman rightly 
adds, that 'for the very reason that Sicily has found dwell- 
ing places for so many nations, a Sicilian nation there 
never has been.'" (p. 271.) 

The secret of the whole dilemma is the intermingling of 
races around the Mediterranean littoral in the last 2500 
years. It is beside the point to contrast our results obtained 
by the actual psychological measurements of living repre- 

iWilliam Z. Ripley. The Races of Europe. New York, 1899, Pp. 624. 



186 AMERICAN INTELLIGENCE 

sentatives of this race with the attainments of the tem- 
porary civiHzations that flared up in historical times. The 
whole question of the degeneration of these peoples has 
been discussed by Mr. Charles W. Gould,^ and our results 
from the examinations of drafted men born in these regions 
support his position. 

It is rather difficult to compare our results from the 
race groups with the various hypotheses erected by Pro- 
fessor William McDougall,^ who, while he does not claim 
for the Nordic race "any general innate superiority" (p. 
29), analyzes the mental constitution of this race and the 
other European races in such a way that an examination 
of his theories will be interesting. Professor McDougall's 
hypotheses, very briefly and inadequately stated are: that 
the Nordic is stronger in the instinct of curiosity, the root 
of wonder, than the Mediterranean ; the herd instinct, the 
root of sociability, is stronger in the Mediterranean than in 
the Nordic; the Nordic is constitutionally introvert, the 
Mediterranean constitutionally extrovert; the instinct of 
self-assertion is strong in the Nordic; the Alpine is introvert 
but not so strongly introvert as the Nordic; the Alpine has 
a high degree of sociability, is perhaps relatively weak in 
curiosity, and strong in the instinct of submission. 

In discussing innate differences in instinctive endowment, 
psychologists are still more or less in a speculative realm, 
but the field is open to experimental attack, and a body of 
knowledge based on experimentation is gradually growing. 
At the present time we must rely on consensus of opinion 
rather than experiment. A census of text-books on psychol- 
ogy would show "curiosity" usually listed as an instinctive 

iCharles W. Gould. America, a Family Matter. New York, 1922, Pp. 196. 
^William McDougall. Is America Safe for Democracy. New York, 1921, Pp. 213. 



AMERICAN INTELLIGENCE 187 

tendency. If we follow Professor Thorndikei in his analysis, 
and eliminate many tendencies that others include, we shall 
still have left the instincts of multiform mental and phy- 
sical activity as the potent movers of men's economic and 
recreative activities, (p. 144.) If any instinctive tendency 
finds expression in the tasks assigned by the army tests, it 
is this instinct for multiform mental activity, more vaguely 
termed "curiosity." Our tests, however, measure the end 
result of such a tendency and not the tendency itself, and 
it is only in this vague way that our results showing the 
definite intellectual superiority of the Nordic race can be 
taken as substantiating or contributing to Professor Mc- 
Dougall's hypothesis. 

It is difficult to check our results from the analysis of the 
foreign born white draft by country of birth (reported in 
Section VI) with the results of other investigators, on ac- 
count of the different tests that were used, and the differ- 
ent methods of selecting subjects. Miss Murdoch 2 exam- 
ined, by means of the Pressey group point scale, 500 Jew- 
ish children and 500 Italian children at one school in New 
York City, and 500 American children and 230 negro 
children at another school. The American and Jewish chil- 
dren tested about the same. About 15% of the Italians 
equalled or exceeded the median of the Jews, and about 
30% of the negroes equalled or exceeded the median of the 
Jews. The investigation equalizes the environmental factor 
by selecting, in one instance, Italians and Jews from the 
same school and consequently from the same general 
neighborhood (East 110th St. near 2nd Ave.), and, in the 
other instance, by selecting native white and negro chil- 

lE. L. Thorndike, Educational Psychology^ Vol. I. The Original Nature of Man, 
New York, 1919, Pp. 327. 

2K. Miirdoch. A Study of Race Differences in New York City. School and Society, 
1920, 11, pp. 147-150. 



188 AMERICAN INTELLIGENCE 

dren from the same general neighborhood (West side, 8th 
Ave., near 140th St.), but the American children living in 
this neighborhood can not be taken as typical of the 
country as a whole. 

Miss Arlitti concludes from her examination of 343 chil- 
dren, (191 native born Americans, 87 Italians, and 71 ne- 
groes), by the Stanford-Binet scale, that "there is a marked 
difference in the distribution of intelligence in groups of 
the same race but different social status," and states that 
"race norms which do not take the social status factor into 
account are apt to be to that extent invalid." (p. 183.) 
This position seems to ignore the observation, repeatedly 
confirmed by experiment, that children from the profes- 
sional, semi-professional and higher business classes have, 
on the whole, an hereditary endowment superior to that 
of children from the semi-skilled and unskilled laboring 
classes. Termans states "It has in fact been found wherever 
comparisons have been made that children of superior 
social status yield a higher average mental age than chil- 
dren of the laboring classes. . . . However, the common 
opinion that the child from a cultured home does better in 
tests solely by reason of his superior home advantages is 
an entirely gratuitous assumption. Practically all of the in- 
vestigations which have been made of the influence of 
nature and nurture on mental performance agree in attrib- 
uting far more to original endowment than to environ- 
ment. Common observation would itself suggest that the 
social class to which the family belongs depends less on 
chance than on the parents' native qualities of intellect and 
character." (p. 115.) 



lA. H. Arlitt. On the Need for Caution in Establishing Race Norms. Journal of 

Applied Psychology, 1921, 5, pp. 179-183. 

^L. M. Terman. The Measurement of Intelligence, Boston, 1916, Pp. 362. 



AMERICAN INTELLIGENCE 189 

One frequently hears the opinion expressed in scientific 
circles that differences found between racial groups can 
not be attributed to race unless the individuals examined 
are drawn from the same social milieu. Miss Arlitt finds 
native born white children of inferior and very inferior 
social status above the Italian and negro children in in- 
telligence, but attributes the larger differences found be- 
tween the entire native white group and the Italian and 
negro groups to the fact that three eighths of the native 
white children come from homes of superior and very 
superior social status. In the same way, Miss Murdoch 
finds Jews living near East 110th St. and 2nd Ave. in New 
York City not very inferior to native born whites living 
in the mixed white and negro section around 8th Ave. and 
140th St. The equalization of the environmental factor 
is a necessary control in certain phases of scientific experi- 
ments on race differences, but conclusions as to the intelli- 
gence of racial groups must be drawn from samples taken 
at random from the entire country. These conditions are 
more nearly met by the army sampling of individuals in 
the draft. Our samples of 81,465 native born individuals in 
the white draft, of 12,492 foreign born individuals, and 
23,596 negroes are drawn impartially from every section 
of the country. If we selected our native born Americans 
from those who live in the same squalid conditions in which 
we find most of our negro and foreign bprn population, we 
would not have a fair sample. 

It is unfortunate that our army data classify foreign 
born individuals only by country of origin, so that we have 
no separate intelligence distributions for the Jews. Accord- 
ing to the 1910 census, about 50% of the foreign born popu- 
lation reporting Russia as their country of origin spoke 
Hebrew or Yiddish, about 25% spoke Polish, less than 3% 
spoke Russian, and the rest spoke Lithuanian, Lettish, 



190 AMERICAN INTELLIGENCE 

German, Finnish, Ruthenian and other tongues. From the 
immigration statistics showing aliens admitted classified 
according to race or people, we find about 10% (arriving 
between 1900 and 1920) reported as Hebrew. It is fair to 
assume that our army sample of immigrants from Russia 
is at least one half Jewish, and that the sample we have 
selected as Alpine i is from one fifth to one fourth Jewish. 

Our figures, then, would rather tend to disprove the 
popular behef that the Jew is highly intelligent. Immi- 
grants examined in the army, who report their birthplace 
as Russia, had an average intelligence below those from 
all other countries except Poland and Italy. It is perhaps 
significant to note, however, that the sample from Russia 
has a higher standard deviation (2.83) than that of any 
other immigrant group sampled, and that the Alpine 
group has a higher standard deviation than the Nordic or 
Mediterranean groups (2.60). If we assume that the Jewish 
immigrants have a low average intelligence, but a higher 
variability than other nativity groups, this would reconcile 
our figures with popular belief, and, at the same time, with 
the fact that investigators searching for talent in New York 
City and California schools find a frequent occurrence of 
talent among Jewish children. The able Jew is popularly 
recognized not only because of his ability, but because he 
is able and a Jew. 

Our results showing the marked intellectual inferiority 
of the negro are corroborated by practically all of the in- 
vestigators who have used psychological tests on white and 
negro groups. This inferiority holds even when a low in- 
tellectual sampling of whites is made by selecting onfy 



^There is no serious objection, from the anthropological standpoint, to classi- 
fying the northern Jew as an Alpine, for he has the head form, stature, and color 
of his Slavic neighbors. He is an Alpine Slav. 



AMERICAN INTELLIGENCE 191 

those who hve in the same environment, and who have had 
the same educational opportunities. Professor Ferguson, ^ 
who has studied the problem most carefully, concludes that 
in general 25% of the negroes exceed the median white. Our 
figures show a greater difference than he estimates, less 
than 12% of the negroes exceeding the average of the 
native born white draft. Professor Ferguson also estimates 
that 20% of pure negroes, 25% of negroes three quarters 
pure, 30% of the true mulattoes, and 35% of the quad- 
roons equal or exceed the average score of comparable 
whites. 

The discrepancies between data from various investiga- 
tors as to the amount of difference between negroes and 
whites probably result from different methods of selecting 
whites. If we compare negroes only to those whites who 
live in the same neighborhood, and who have had the 
same educational opportunities, our differences are smaller 
than those obtained by comparing samples of the entire 
white and negro populations. 

Some writers would account for the differences found 
between white and negro by differences of educational 
opportunity alone. The army tests showed the northern 
negro superior to the southern negro, and this superiority 
is attributed to the superior educational opportunities in 
the North. The educational record of the negro sample we 
are studying shows that more than half of the negroes from 
the southern States did not go beyond the third grade, and 
only 7% finished the eighth grade, while about half of the 
northern negroes finished the fifth grade, and a quarter 
finished the eighth grade. That the difference between the 
northern and southern negro is not entirely due to school- 



^G. O. Ferguson. The Mental Status of the American Negro. Scientific Monthly, 
1921, 12, pp. 533-543. 



'2 



192 AMERICAN INTELLIGENCE 

ing, but partly to intelligence, is shown by the fact that 
groups of southern and northern negroes of equal schooling 
show striking differences in intelligence. 

The superior intelligence measurements of the northern 
negro are due to three factors : first, the greater amount of 
educational opportunity, which does affect, to some ex- 
tent, scores on our present intelligence tests; second, the 
greater dmount of admixture of white blood; and, third, 
the operation of economic and social forces, such as higher 
wages, better living conditions, identical school privileges, 
and a less complete social ostracism, tending to draw the 
more inteUigent negro to the North. It is impossible to 
dissect out of this complex of forces the relative weight of 
each factor. No psychologist would maintain that the men- 
tal tests he is now using do not measure educational oppor- 
tunity to some extent. On the other hand, it is absurd to 
attribute all differences found between northern and south- 
ern negroes to superior educational opportunities in the 
North, for differences are found between groups of the same 
schooling, and differences are shown by beta as well as 
by alpha. 

At the present stage of development of psychological 
tests, we can not measure the actual amount of difference 
in intelligence due to race or nativity. We can only prove 
that differences do exist, and we can interpret these differ- 
ences in terms that have great social and economic signifi- 
cance. The intellectual superiority of our Nordic group over 
the Alpine, Mediterranean, and negro groups has been 
demonstrated. If a person is unwilling to accept the race 
hypothesis as developed here, he may go back to the 
original nativity groups, and he can not deny the fact that 
differences exist. 

When our methods of measuring intellectual capacity 
have been perfected, we will be in a position to determine 



AMERICAN INTELLIGENCE 193 

quantitatively the amount of race differences. Rough group 
tests of the type we are now using will indicate the fact 
that differences exist. However, while scientists are perfect- 
ing their methods of examining, it would be well for them 
to perfect their logic at the same time. Particularly mis- 
leading and unsound is the theory that disregards all dif- 
ferences found between racial groups unless the groups 
have had the same educational and environmental oppor- 
tunities. 

This theory in its most extreme form is set forth by 
Garth 1 as follows: 

"The elements in a study of racial mental similarities or 
differences must be these: (1) Two so-called races Rj and 
R2, (2) an equal amount of educational opportunity, E, 
which should include social pressure and racial patterns of 
thought, and (3) psychological tests, D, within the grasp 
of both racial groups. We should have as a result of our 
experiment Ri E D equal to, greater than, or less than R^ 
E D. In this experiment the only unknown elements should 
be Ri and R^. If E could be made equal the experiment 
could be worked. 

"This element of educational opportunity-nurture is the 
one causing most of the trouble in racial psychology as an 
uncontrollable element. It does not offer quite so much 
difficulty in the study of sex differences, yet it is there 
only in smaller degree than in racial differences, and as it 
is controlled the *sex differences' tend to disappear. Since 
this element of education, or nurture, cannot be eliminated 
it would be safer to take for comparison such racial groups 
as have had as nearly the same educational opportunity as 
is possible having any disparity of this sort well in mind 
when we interpret the results of the experiment. Having 

^T. R. Garth. White, Indian and Negro Work Curves. Journal of Applied Psy - 
chology, 1921, 5, pp. 14-25. 



194 AMERICAN INTELLIGENCE 

done this, we first take the complete distributions on the 
scale of measurement for the groups as statements of the 
true facts of the case, race for race. We then combine these 
distributions into a total distribution of accompHshment 
of all the races taken together to see if we have multimodal 
effects. Should we find these effects we may conclude that 
we have evidence of types, or racial types, and there should 
in this case be one mode for each racial group. But should 
the combined distribution for the several racial groups 
reveal only one mode we may conclude that the test reveals 
no types no real racial differences but rather similarities." 
(p. 16.) 

If intelligence counts for anything in the competition 
among human beings, it is natural to expect that individ- 
uals of superior intelligence will adjust themselves more 
easily to their physical and social environment, and that 
they will endow their children not only with material 
goods, but with the ability to adjust themselves to the 
same or a more complex environment. To select individuals 
who have fallen behind in the struggle to adjust themselves 
to the civilization their race has built as typical of that 
race is an error, for their position itself shows that they 
are, for the most part, individuals with an inferior heredi- 
tary endowment. 

In the same way, our educational institutions are them- 
selves a part of our own race heritage. The average negro 
child can not advance through an educational curriculum 
adapted to the Anglo-Saxon child in step with that child. 
To select children of equal education, age for age, in the 
two groups, is to sample either superior negroes or inferior 
whites. 

The scientific problem is that of eliminating from the 
tests used as measuring instruments those particular tests 
which demonstrably measure nurture, and to measure. 



AMERICAN INTELLIGENCE 195 

with genuine tests of native intelligence, random or im- 
partial samples from each race throughout the entire range 
of its geographical and institutional distribution. 



CONCLUSIONS 

Our study of the army tests of foreign born individuals 
has pointed at every step to the conclusion that the aver- 
age intelligence of our immigrants is declining. This deteri- 
oration in the intellectual level of immigrants has been 
found to be due to two causes. The migrations of the Alpine 
and Mediterranean races have increased to such an extent 
in the last thirty or forty years that this blood now consti- 
tutes 70% or 75% of the total immigration. The represen- 
tatives of the Alpine and Mediterranean races in our 
immigration are intellectually inferior to the representa- 
tives of the Nordic race which formerly made up about 
50% of our immigration. In addition, we find that we are 
getting progressively lower and lower types from each 
nativity group or race. 

In the light of our findings in Sections IV and IX, it is pos- 
sible to re-draw our curve (Figure 33) representing in- 
crease of intelligence score with increasing years of residence 
and to represent it truly as in Figure 46, which shows the 
decline of intelligence with each succeeding period of im- 
migration. 

It is also possible to make a picture of the elements now 
entering into American intelligence. At one extreme we 
have the distribution of the Nordic race group. At the 
other extreme we have the American negro. Between the 
Nordic and the negro, but closer to the negro than to the 
Nordic, we find the Alpine and Mediterranean types. These 
distributions we have projected together in Figure 47. 

Throughout this study all measurements have been made 
in terms of averages and variability about the average. In 
interpreting averages, we must never forget that they stand 

197 



WHITE OFFICERS I 



WHITE DRAFT J^^ 
BORN IN ENGLAND 



FOREIGN BORN 
WHITE DRAFT 



NATIVE BORN 
"WHITE DRAFT 



O CO 

m Q 

pq 12 

en 03 



NEGRO DRAFT I 



OYEB^O 

1897 




16-20 



-10 



11-15 
1898 1903 

TO ijga 

1902 1907 

YEARS RESIDENCE IN UNITED STATES 



TO 
1912 



0-5 
1913 

TO 

1917 



Figure 46. The decline of intelligence with each succeeding period 
of immigration. The apparent increase of intelligence with in- 
creasing length of residence, as shown in Figure SS, has been 
proved to be a progressive decrease in the intellectual level of 
immigrants coming to this country in each succeeding five-year 
period since 1902. The evidence indicates that the immigrants 
prior to 1902 were intellectually equal to the native born white 
draft. The army sample of "native born" includes, besides na- 
tive born of native parentage, the native born of foreign or mixed 
parentage. It is perhaps possible that the native born of native 
parentage might have tested higher than 13.77. The position of 
the white draft born in England is shown above. Although the 
true position of the native born American may be a matter of 
speculation, there is no doubt that the more recent immigrants 
are intellectually closer to the negro than to the native born 
white sample. 






4J "^ 



.2 



2V 

ID 00 



M 



fir -M c3 



53 n :t< ^ 



^.S? 



43 



^ p: c 

p^ < 

S 2 ^ 

83 O O 

1^ 3 -^ p, PI 

O ^-S ^ ^ 
g . P 









o 






(U 









<4-l c IS (D O "u 



03 nd 



0)0*33 






Oh 
O fl 



faCrC 



03 
4J o o 

rH ^ Oh 



O 



1^ IJ 







202 AMERICAN INTELLIGENCE 

for an entire distribution. Careless thinkers are prone to 
select one or two striking examples of ability from a partic- 
ular group, and then rest confidently in the belief that 
they have overthrown an argument based on the total dis- 
tribution of ability. The Fourth of July orator can con- 
vincingly raise the popular belief in the intellectual level 
of Poland by shouting the name of Kosciusko from a high 
platform, but he can not alter the distribution of the intel- 
ligence of the Polish immigrant. All countries send men of 
exceptional ability to America, but the point is that some 
send fewer than others. 

Our distribution curve of intelligence includes ability as 
well as defect. The English speaking Nordic group, for in- 
stance, averages 13.84, and furnishes at one extreme about 
40 men in 1000 who are above the average white officer, 
while at the other extreme, the group furnishes about 8 in 
1000 who are below an estimated "mental age" of eight. A 
distribution further down the scale contributes more to the 
lower orders of intelligence. The distribution of the intel- 
ligence scores of the negro draft, for instance, indicates that 
they contribute only 4 in 1000 above the average white 
officer, while they give us 100 in 1000 below the approxi- 
mate "mental age" of eight. The Alpine and Mediterranean 
races give us only 5 or 10 in 1000 above the average ability 
of the white officer, and about 40 in 1000 below the "mental 
age" of eight. About 350 in 1000 of the Alpine and Mediter- 
ranean types are below the average negro. 

The intellectual characteristics of the immigration to the 
United States as measured by the samples in the draft have 
been reported in this study, first by country of birth, and 
second by race. Parallel with the measurements of intel- 
ligence, the figures on immigration have been presented. 
To complete the picture, there is presented in Table 38 
the population of the United States according to the 1920 



Table No. 38 
Population of the United States in 1920 

Native White of Native Parentage. .58,421,957 
Native White of Foreign Parentage . 15,694,539 
Native White of Mixed Parentage . . 6,991 ,665 

Total Native White 81,108,161 

Foreign Born White 13,712,754 

Negro 10,463,131 

Indian : 244,437 

Chinese 61,639 

Japanese 111,010 

All others : 

Filipinos 5,603 

Hindus 2,507 

Koreans 1,224 

Hawaiians 110 

Malays 19 

Siamese 17 

Samoans 6 

Maoris 2 

Total all others 9,488 

Total Population 105,710,620 



NOTE: Clinton Stoddard Burr in America's Race Heritage (New York: The 
National Historical Society, 1922, pp. 327) estimates that in 1920 there were 
44,689,278 descendents of the old Colonial white stock. 



203 



204 AMERICAN INTELLIGENCE 

census. 1 We have with us approximately 15 7/10 milHons 
of individuals of foreign parentage, 7 millions of mixed 
parentage, 13 7/10 millions of foreign born, and 10 Y2 
millions of negroes. Roughly, in every 100 of our popula- 
tion, ^^ are native born of native parentage, and the other 
45 foreign born, or of foreign, mixed, or colored parentage. 
The group of native born of native parentage includes 
many children of the immigrants coming to this country 
prior to 1890. 

Our immigration figures show a very decided shift from 
the Nordic in favor of the Alpine. The immigration between 
1820 and 1890 probably never contained more than 50% 
or 60% Nordic blood, and prior to 1820 there was very 
little immigration. The earliest settlers were almost pure 
Nordic types, and we may assume the existence by 1820 
of a race as predominantly Nordic as that of England. This 
recent change was, of course, reflected in the cross section 
of the foreign born population taken at 1910, and which 
constitutes the basis of our present immigration act re- 
stricting immigration to 3% of the nationals then resident 
here. A rough estimate of the racial composition of the 
quotas from various countries admissable under the new 
law shows about 35% Nordic blood, 45% Alpine blood and 
20% Mediterranean blood in the annual stream of ap- 
proximately 1/3 of a million that may enter. 

There can be no doubt that recent history has shown a 
movement of inferior peoples or inferior representatives of 

iToo much reliance can not be'placed on the census returns for the foreign bom 
white population. The 1910 census shows the foreign born white population as 
13,345,545, while the 1920 census shows that population as 13,712,754, which 
gives a net increase of 367,209. On the other hand, the figures of the Commis- 
sioner General of Immigration show by actual count at the ports 5,725,811 
aliens admitted and 2,146,994 aliens departed, leaving a net increase of 3,578,817 
for the same period covered by the two censuses (1910 and 1920). Inasmuch as 
the enumerators could not have missed three million, they are probably counted 
among the native white. 



AMERICAN INTELLIGENCE 205 

peoples to this country. Few people realize the magnitude 
of this movement or the speed with which it has taken 
place. Since 1901, less than a single generation, it may be 
estimated that about 10,000,000 Alpine and Mediterran- 
ean types have come to this country. Allowing for the re- 
turn of 1/3 or 3/8 of these, and using our army estimates 
of intellectual ability, this would give us over 2,000,000 
immigrants below the average negro. 

We may consider that the population of the United 
States is made up of four racial elements, the Nordic, 
Alpine, and Mediterranean races of Europe, and the negro. 
If these four types blend in the future into one general 
American type, then it is a foregone conclusion that this 
future blended American will be less intelligent than the 
present native born American, for the general results of the 
admixture of higher and lower orders of intelligence must 
inevitably be a mean between the two. 

If we turn to the history of races, we find that as a general 
rule where two races have been in contact they have inter- 
mingled, and a cross between the two has resulted. Europe 
shows many examples of areas where the anthropological 
characteristics of one race shade over into those of another 
race where the two have intermixed, and, indeed, in coun- 
tries such as France and Switzerland it is only in areas that 
are geographically or economically isolated that one finds 
types that are relatively pure. The Mongol-Tatar element 
in Russia is an integral part of the population. The Mediter- 
ranean race throughout the area of its contact with the 
negro has crossed with him. Some of the Berbers in 
Northern Africa show negroid characteristics, and in India 
the Mediterranean race has crossed with the Dravidians 
and Pre-Dra vidian negroids. The population of Sardinia 
shows a number of negroid characteristics. Turn where we 
may, history gives us no great exception to the general rule 



206 AMERICAN INTELLIGENCE 

that propinquity leads to opportunity and opportunity to 
intermixture. 

In considering racial crosses, Professor Conklini states 
that "It is highly probable that while some of these hy- 
brids may show all the bad qualities of both parents, others 
may show the good qualities of both and indeed in this re- 
spect resemble the children in any pure-bred family. But 
it is practically certain that the general or average results 
of the crossing of a superior and an inferior race are to 
strike a balance somewhere between the two. This is no 
contradiction of the principles of Mendelian inheritance 
but rather the application of these principles to a general 
population. The general effect of the hybridization of races 
can not fail to lead to a lowering of the qualities of the 
higher race and a raising of the qualities of the lower one." 
(pp. 50-51.) 

And as to the possibility of a cross between races in the 
future. Professor Conklin writes: "Even if we are horrified 
by the thought, we cannot hide the fact that all present 
signs point to an intimate commingling of all existing hu- 
man types within the next five or ten thousand years at 
most. Unless we can re-establish geographical isolation 
of races, we cannot prevent their interbreeding. By rigid 
laws excluding immigrants of other races, such as they have 
in New Zealand and Australia, it may be possible for a time 
to maintain the purity of the white race in certain coun- 
tries, but with constantly increasing intercommunications 
between all lands and peoples such artificial barriers will 
probably prove as ineffectual in the long run as the Great 
Wall of China. The races of the world are not drawing apart 
but together, and it needs only the vision that will look 

1 Edwin G. Conklin. The Direction of Human Evolution. New York, 1921, pp. 247. 



AMERICAN INTELLIGENCE 207 

ahead a few thousand years to see the blending of all racial 
currents into a common stream." (p. 52.) 

If we frankly recognize the fact that the crossing of races 
in juxtaposition has always occurred in the past, what 
evidence have we that such crosses have had untoward con- 
sequences? Our own data from the army tests indicate 
clearly the intellectual superiority of the Nordic race 
group. This superiority is confirmed by observation of this 
race in history. The Alpine race, according to our figures, 
which are supported by historical evidence, seems to be 
considerably below the Nordic type intellectually. How- 
ever, our recruits from Germany, which represents a Nordic- 
Alpine cross, are about the same as those from Holland, 
Scotland, the United States, Denmark, and Canada, coun- 
tries which have on the whole a greater proportion of Nordic 
blood than Germany. Again, the Nordic and Alpine mix- 
ture in Switzerland has given a stable people, who have 
evolved, in spite of linguistic differences, a very advanced 
form of government. The evidence indicates that the 
Nordic-Alpine cross, which occurred in Western Europe 
when the Nordics overwhelmed the Alpines to such an ex- 
tent that the type was completely submerged and not 
re-discovered until recently, has not given unfortunate re- 
sults. 

This evidence, however, can not be carried over to indi- 
cate that a cross between the Nordic and the Alpine Slav 
would be desirable. The Alpines that our data sample come 
for the most part from an area peopled largely by a 
branch of the Alpine race which appeared late and radiated 
from the Carpathian Mountains. It is probably a different 
branch of the Alpine race from that which forms the prim- 
itive substratum of the present population of Western 
Europe. Our data on the Alpine Slav show that he is in- 
tellectually inferior to the Nordic, and every indication 



208 AMERICAN INTELLIGENCE 

would point to a lowering of the average intelligence of the 
Nordic if crossed with the Alpine Slav. There can be no 
objection to the intermixture of races of equal ability, pro- 
vided the mingling proceeds equally from all sections of 
the distribution of ability. Our data, however, indicate that 
the Alpine Slav we have imported and to whom we give 
preference in our present immigration law is intellectually 
inferior to the Nordic type. 

The Mediterranean race at its northern extension blends 
with the Alpine very considerably, and to a less extent with 
the Nordic. At the point of its furthermost western expan- 
sion in Europe it has crossed with the primitive types in 
Ireland. Throughout the area of its southern and eastern 
expansion it has crossed with negroid types. In this con- 
tinent, the Mediterranean has crossed with the Amerind 
and the imported negro very extensively. In general, the 
Mediterranean race has crossed with primitive race types 
more completely and promiscuously than either the Alpine 
or the Nordic, and with most unfortunate results. 

We must now frankly admit the undesirable results which 
would ensue from a cross between the Nordic in this coun- 
try with the Alpine Slav, with the degenerated hybrid 
Mediterranean, or with the negro, or from the promiscu- 
ous intermingling of all four types. Granted the undesirable 
results of such an intermingling, is there any evidence 
showing that such a process is going on.^^ Unfortunately the 
evidence is undeniable. The 1920 census shows that we 
have 7,000,000 native born whites of mixed parentage, a 
fact which indicates clearly the number of crosses between 
the native born stock and the European importations. 

The evidence in regard to the white and negro cross is 
also indisputable. If we examine the figures showing the 
proportion of mulattoes to a thousand blacks for each 
twenty year period from 1850 to 1910, we find that in 1850 



AMERICAN INTELLIGENCE 209 

there were 126 mulattoes to a thousand blacks, 136 in 1870, 
179 in 1890 and 264 in 1910. This intermixture of white and 
negro has been a natural result of the emancipation of the 
negro and the breaking down of social barriers against him, 
mostly in the North and West. In 1850, the free colored 
population showed 581 mulattoes to a thousand blacks as 
against 83 in the slave population. At each of the four cen- 
suses (1850, 1870, 1890 and 1910) the South, where the 
social barriers are more rigid than elsewhere, has returned 
the smallest proportion of mulattoes to a thousand blacks. 
The 1910 census showed 201 in the South, 266 in the 
North and 321 in the West, and the West has returned the 
highest proportion at each of the censuses except 1850. 

We must face a possibility of racial admixture here that 
is infinitely worse than that faced by any European country 
to-day, for we are incorporating the negro into our racial 
stock, while all of Europe is comparatively free from this 
taint. It is true that the rate of increase of the negro in this 
country by ten year periods since 1800 has decreased rather 
steadily from about 30% to about 11%, but this declining 
rate has given a gross population increase from approximate- 
ly 1,000,000 to approximately 10,000,000. It is also true that 
the negro now constitutes only about 10% of the total 
population, where he formerly constituted 18% or 19% 
(1790 to 1830), but part of this decrease in percentage of 
the total population is due to the great influx of immi- 
grants, and we favor in our immigration law those coun- 
tries 35% of whose representatives here are below the aver- 
age negro. The declining rate of increase in the negro 
population from 1800 to 1910 would indicate a correspond- 
ingly lower rate to be expected in the future. From 1900 to 
1920 the negro population increased 18.4%, while the na- 
tive born white of native parents increased 42.6%, and the 
native born white of foreign parents increased 47.6%. It is 



210 AMERICAN INTELLIGENCE 

impossible to predict at the present time that the rate of 
infiltration of white blood into the negro will be checked 
by the declining rate of increase in the negro blood itself. 
The essential point is that there are 10,000,000 negroes here 
now and that the proportion of mulattoes to a thousand 
blacks has increased with alarming rapidity since 1850. 

According to all evidence available, then, American in- 
telligence is- declining, and will proceed with an accelerat- 
ing rate as the racial admixture becomes more and more 
extensive. The decline of American intelligence will be more 
rapid than the decline of the intelligence of European 
national groups, owing to the presence here of the negro. 
These are the plain, if somewhat ugly, facts that our study 
shows. The deterioration of American intelligence is not 
inevitable, however, if public action can be aroused to pre- 
vent it. There is no reason why legal steps should not be 
taken which would insure a continuously progressive up- 
ward evolution. 

The steps that should be taken to preserve or increase 
our present intellectual capacity must of course be dictated 
by science and not by political expediency. Immigration 
should not only be restrictive but highly selective. And the 
revision of the immigration and naturalization laws will 
only afford a slight relief from our present difficulty. The 
really important steps are those looking toward the pre- 
vention of the continued propagation of defective strains 
in the present population. If all immigration were stopped 
now, the decline of American intelligence would still be 
inevitable. This is the problem which must be met, and our 
manner of meeting it will determine the future course of 
our national life. 



Uoiversity of Torooto 
library 



DO NOT 

REMOVE 

THE 

CARD 

FROM 

THIS 

POCKET 




Acme Library Card Pocket 

Uader Pat. "Ref. Index FUt" 

Made by LIBRARY BUREAU