ALFRED P. SLOAN SCHOOL OF MANAGEMENT
INSTITUTE OF TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASSACHUSETTS 02139
^ts&. INST, rfc^j^
DEC 1.2 1975 J
MASS. INST. TECH.
INDUSTRIALIZATION AND VARIATION IN SOCIAL
STRUCTURE: AN EMPIRICAL TEST OF THE
By Stephen J. Kobrin
INDUSTRIALIZATION AND VARIATION IN SOCIAL
STRUCTURE: AN EMPIRICAL TEST OF THE
By Stephen J. Kobrin
^£) , 825- ^s■
DEC 10 1975 ':
In Industrialism and Industrial Man , Kerr, Dunlop, Harbison, and Myers
concluded that social systems become more uniform and societies in general
become more alike as they industrialize. While the convergence hypothesis
has generated a good deal of discussion and controversy in the intervening
fifteen years, reports of empirical tests of the concept have been limited.
This paper summarizes such an attempt utilizing the techniques of cross-
national research across a group of fifty-nine developing countries. The
theoretical underpinnings of the hypothesis will first be discussed, the
concept will then be restated in terms of testable hypotheses, the empirical
findings will be reviewed, and, last, the conclusions will be presented.
II, Industrialization and social change
Reduced to its essence, industrialization entails the use of inanimate
sources of power--tools and machines--to multiply human effort in production.
As its raison d'etre is an increase in output per unit of (human) input,
broadscale industrialization results in an increased societal emphasis on
efficiency which, given the nature of the machine pleads to larger scale
productive units. There is thus a central logic to industrialization, a
logic which leads to an increased emphasis on efficiency and scale.
As Moore has observed, it is not reasonable to expect industrialization
to be neutral in its social consequences. Rather, evidence from both the
West and the developing countries indicates that industrialization tends
to be a "universal social solvent." The logic of the machine, the pressure
it exerts for increases in efficiency and scale, is inconsistent with the
* The author would like to thank Charles A, Myers and Clark Kerr for their
comments and encouragement.
basic structure of traditional society.
Widespread mechanization of production is difficult ( if not impossible)
to achieve within the bounds of a traditional society where roles are diffused
and ascribed, and the economic, social, and religious aspects of life
are often interwoven as a whole cloth. This conflict between mechanization
and social structure results in what is perhaps the most important immediate
effect of industrialization, the different iation of roles or the division of labor.
Development of a mechanized and occupationally differentiated system
of production, in turn, results in pressures for increases in both horizontal
and vertical mobility and for changes in the basic patterns of social organ-
ization, A relatively rigid and stable stratification system, with both
status and roles ascriptively determined, directly conflicts with the
"functional requisites of a society with expanding and progressing economic
activities," To the extent efficiency is important, positions must be
filled on the basis of skills and abilities rather than status; if large
scale productive organizations are to develop, workers must be free to
Similarly, industrialization has placed severe disintegrative pressure
on extended families, tribes, and clans. As Tumin has noted, industrial-
ization violates the corporate nature of the clan; its organization is
inconsistent with the use of individuals as units of work or remuneration.
The nuclear or conjugal family is much better suited to broadscale mechan-
ization. It is geographically and socially mobile, allows for individual
specialization, and importantly for individual remuneration and accumulation.
Two points emerge from this brief summary of the relationship between
industrialization and social change. First, industrialization results in
pressures for change in the structure of traditional society, a process defined
in this paper as social modernization. Second, there is a central logic to
industrialization resulting from the emphasis on efficiency and scale which
directs--within limits--the process of change. The social structures which
evolve must be consistent with the requisites of industrialization. We would
have to agree with Levy that: "(T)he emphasis associated with high levels
of modernization on rationality, universalism, and functional specificity,
for example, limit the possibilities of relationships in terms of which
manufacturing can be carried out." As one author has noted, there are
many ways of skinning a cat, but very few ways of operating a zipper,
In their final review of Industrialism and Industrial Man . Dunlop et al
conclude that, "...the logic of industrialization results in advanced indus-
trial societies becoming more alike, despite cultural and political differ-
ences, and certainly more alike than any one of them is like a less developed
country." However, they note that given political, social and cultural
differences, there will never be total convergence; that convergence is
towards a range of alternatives rather than a single point.
There has been a good deal of support for the concept of convergence
in the literature, Karsh and Cole note that the application of science and
technology implies a level of standardization more universal than in pre-
industrial society. Haskins finds, "(F)undamentally convergent pressures
immanent in the nature and structure of science and technology." Levy
concludes that, "(F)or all of their important differences, all relatively
modernized societies are in fact becoming more and more alike all the time
and the range of possible variation among them decreases all the time."
According to Moore and Feldman, "virtually no one rejects the notion
that industrial societies share a core set of social structures that together
provide an extended operational definition of industrialism itself."
However, Moore had earlier (in 1965) concluded that the weight of the evidence
was against convergence. He felt that while structural interdependence and
functional integration exert pressures for convergence, they are more than
offset by differences in preindustrial society and differences in the routes
and processes of change. Upon closer examination, it is not clear that
this position differs from that contained in Industrialism and Industrial
Man and its various postscripts. It will be recalled that Dunlop et al
noted that cultural, social and political differences prevent complete
uniformity. Furthermore, they saw the process as so complex and "subject to
such contrary internal pressures" that it is inherently pluralistic.
In this research we are concerned with the direct impact of industrial-
ization upon social structure. We are thus not interested in testing (nor
entering the lists on) questions of convergence of political systems, ideol-
ogies, values and attitudes, or cultural orientations. We would agree that
both preindustrial social structures and differences in the path of industrial-
ization impinge on the process of social modernization. However, it appears
obvious that once a society embarks on a course of broadscale industrialization,
with its emphasis upon efficiency and scale, and pressures for role differen-
tiation, mobility, and changes in social organization, the range and variation
of social structure becomes increasingly circumscribed.
In operational terras the convergence hypothesis implies differences in
the variation of an indicator (or indicators) of social structure at differ-
ent levels of industrialization. One would expect decreases in the variation
of an indicator of social structure (about a central measure) at higher levels
of industrialization. Furthermore, assuming one could isolate and quantify
various aspects of social structure (e.g. mobility or family organization)
one would also expect less variation among the various aspects of social
structure at higher levels of industrialization. Thus we would hypothesize
(1) The variation about the mean for a given indicator of social
structure decreases as the level of industrialization increases.
(2) There is less variation between the indicators of various aspects
of social structure at higher levels of industrialization.
(3) If countries are divided into relatively less and relatively more
industrialized groups, there is less variation within than between
The methodology is cross-national and cross sectional with most data
collected between 1962-1965. All non-socialist bloc developing countries
that were sovereign states as of 1965 and met population and economic size
minimums were included in the analysis. The final country list (see Table 4)
contains fifty-nine countries.
Raw indicators of social structure that are directly affected by indus-
trialization were obtained from published sources, such as the various yearbooks
published by international organizations, university data banks, and collections
of qualitative indicators such as those compiled by Adelman and Morris or
Banks and Textor, The original list included twenty- two indicators; after
redundant and apparently spurious indicators were deleted, seventeen were
retained for further analysis. The appendix contains a detailed description
of the indicators and their sources.
The seventeen raw indicators of social structure consist of a relatively
large number of variabljes which are to a large extent interdependent and
correlated with one another. Factor analysis was thus used to reduce the
data to a smaller number of independent, and hopefully conceptually more
meaningful, aspects of social structure.
Quantification of industrialization was straightforward. Its definition
(the application of inanimate sources of power to production) suggests power
consumption per capita as a basic indicator. To provide a measure of the
breadth of industrialization, the size of the manufacturing sector, as a
percentage of GDP, was added. The index of industrialization thus consists
of an unweighted linear average of the two variables.
Given both the somewhat diffuse nature of the convergence hypothesis and
the problems of accuracy and comparability of data which arise when one
works with statistical indicators across a relatively large number of devel-
oping countries, relatively simple and straightforward techniques are used to
test the hypotheses. These can best be duscussed in the context of the findings,
As noted above, factor analysis (a principal axis solution and a varimax
rotation) was used to reduce the seventeen raw indicators to a smaller number
of aspects of social structure. The matrix of rotated factor loadings is
- 7 -
shown as Table 1. The loadings are a measure of the degree to which a given
variable is associated with a given factor; they are analogous to correlation
coefficients. The column headed by h contains communalities or the percentage
of the common variance of a given variable accounted for by the factors in
total. As the factors were orthogonally rotated, the loadings define the
major clusters of interrelationships among the variables and the factors are
independent. While interpretation is necessarily subjective, the conceptual
content of a factor is typically inferred from those variables loading most
highly on it.
The variables which load most highly on the first factor represent the
movement from diffused and traditional agricultural roles into more differ-
entiated roles in the mining and manufacturing industries and commercial
agriculture, and the associated general societal changes. It is thus named
role differentiation or roles . The second factor is clearly defined; it
captures variables which are measures or requisites of vertical and horizontal
mobility and it is so named.
The last factor can be interpreted as encompassing the transition from
tribes and extended families to nuclear families, a corresponding lessening of
the influence of non-associational groups, and increasing integration and
homogeneity of society as cultural and linguistic differences break down. The
third factor thus describes basic changes in the organization of society and
is named social organization .
The three factors represent major components of the changes in social
structure which result from industrialization; increased differentiation of
roles, increased mobility, and a shift to a more homogeneous society composed
of nuclear family units. Scores for each of the factors were obtained by
utilizing the loadings as weights and summing for the (suitably rescaled)
variables which load highest on each factor.
The first hypothesis posits less variation in each of the indicators of
social structure at higher levels of industrialization. This suggests a
comparison of their dispersion at several levels of
industrialization. While the standard deviation provides an absolute measure
of dispersion, a relative measure is required for comparison. To this end,
the coefficient of variation, which is the standard deviation expressed as
a percentage of the arithmetic mean, was computed for each aspect at four
levels of industrialization. The results are presented in Table 2.
In each instance there is less dispersion (relative to the mean) in a
given indicator at higher than at lower levels of industrialization. Consider-
ing each indicator in turn, countries are more alike at higher than at lower
levels of industrialization. It is interesting to note that both the absolute
and relative measure of dispersion for social organization is larger, at each
level of industrialization, than for either roles or mobility. This is
consistent with what the theory would predict as: 1) there tends to be a
greater variation in this aspect of social structure in preindustrial societies
than for roles or mobility and 2) changes in social organization are a function
of changes in the other two factors which are more immediately affected by
The second hypothesis states that there will be less variation between
the various aspects of social modernization as nations industrialize. If the
process of social modernization is a coherent phenomenon, and if the three
factors are actual aspects of modernization, then a country which is more
modernized than another should score higher on each of the three aspects of
modernization than the other. The three aspects or factors would then con-
stitute what is known as a unidimensional scale.
In actual practice we would expect deviations from a unidimensional scale.
We would expect to find errors or a less than perfect correspondence between
an ideal or predicted scale and the actual scale. In terms of the problem at
hand, we would expect this correspondence to vary at different levels of
industrialization; we would expect a better fit between ideal and actual
scales at higher levels of industrialization.
This correspondence, or the degree of reproducibility of the ideal scale,
may be computed by a technique known as scaleogram analysis. Guttman scales
are utilized to determine if responses of subjects to given items form a
scale; if they can be ordered along an underlying dimension. In a perfect
Guttman scale, responses of a subject to all items can be reproduced from the
knowledge of its rank order alone. The technique thus involves construction
of an ideal scale, comparing the actual and ideal scale (counting errors),
and then determining if the degree of reproducibility is satisfactory.
In the case at hand, we thus want to determine if country scores on roles,
mobility, and social organization form a scale and if the degree of reproduci-
bility varies at different levels of industrialization. The three aspects of
modernization were converted to ordinal scales (quartiles) by reference to
their distributions. The ideal or predicted scale was constructed by util-
izing the bar chart method and is shown as Table 3. The lowest value for a
given aspect is zero and the highest is three; the total scale score is simply
the row sum. The ideal scale appears to be consistent with theory; where
deviations from a homogeneous pattern (eg 3,3,3) occur, roles always has
a value equal to or greater than the other two indicators. This is to be
expected if role differentiation is the first order effect of industrial-
The actual scale is presented as table 4. The countries are grouped
by four levels of industrialization based upon a division of the ordinal
scale for that variable into four equal parts. The errors are simply the
deviations from the closest ideal pattern. The coefficient of reproduce-
ability (CR) which is one less the number of errors as a percentage of
the number of responses of all subject to all items, is then used to
judge whether the actual responses constitute a scale. Originally
Guttman selected a CR of 0.85 as the dividing line separating scales from
non-scales, but more recently the standard has been tightened to 0.90.
Table 5 summarizes the degree of reproduceability (CR) for the overall
scale and each of the four levels of industrialization. The overall CR
of 0.92 is clearly acceptable; the categorized country scores on the three
aspects of modernization constitute a unidimensional scale. The pattern
of reproduceability over the sub-groups is consistent with the second
hypothesis. The correspondence between the ideal and actual scales is best
at the lowest and at the highest levels of industrialization. The fit, while
still acceptable, is poorer at interim levels.
While we would certainly not claim that the fifteen least industrialized
countries represent "ideal" traditional societies, they are reasonable approx-
imations of preindustrial states. Thus, one would expect a good deal of
homogeneity. In fact, six of the fifteen countries in the group (all of
which rank among the eight least industrialized of the countries studied)
have a scale score of zero.
- 11 -
At interim levels of development, one would expect a poorer corres-
pondence between the predicted and the actual scale patterns; greater
variation in the relationship between the various aspects of social
structure. The immediate effects of the introduction of industrialization
are disintegrative^^ and a breakup of the structures of traditional society
results. While, in general, pressures for social change are exerted through
the increased differentiation of roles, there is no reason to expect either
the path of industrialization or the process of social modernization to
follow the same course in each country.
If the convergence hypothesis is correct, one would expect that as
industrialization proceedes, its requisites would limit the variation be-
tween aspects of social structure. Thus, the improvement in reproduceability
of the predicted scale pattern in the fourth quartile is consistent with the
hypothesis. It should also be noted that thirty-eight percent of the
errors in the fourth quartile are attributable to South Africa, which is
certainly a very singular case. If South Africa is dropped from the
analysis, the CR for the fourth quartile becomes 0.96.
The third hypothesis was tested through cluster analysis which is a
technique used to identify groups of similar individuals within a given
population. Cluster analysis is performed on cases (countries) rather than
variables and essentially groups individuals based upon their profile over
a given set of variables. In this instance, the fifty-nine countries
were clustered on the basis of their profiles over the seventeen raw
indicators of social structure.
While the clustering procedure is hierarchical, building from the most
similar pair of countries, the two penultimate clusters contained all of
the countries analyzed. Table 6 shows the cluster each country is
contained in with the countries grouped by level of industrialization.
If the match was perfect, all countries in quartiles III and IV would
be contained in cluster I and all of the countries in quartiles I and II
in cluster II. On this basis nine or fifteen percent of the countries are
misplaced. However, if one considers only the most and least industrial-
ized quartiles only one country (Ceylon) is contained in the "wrong"
The findings must be considered tentative for several reasons. First,
one must have substantial concerns about the accuracy and comparability
of indicators of social structure collected across a large number of
developing countries. Second, attempting to investigate a longitudinal
phenomena through a cross-sectional methodology requires assumptions about
the nature of the process which may lead to conceptual difficulties.
Third, the tests did not involve rigorous statistical tests of significance.
Thus, one certainly can not state the convergence hypothesis is proven.
However, it is reasonable to conclude that it is consistent with the data
and the findings of this research. The variance of both a given aspect of
social structure and among the three aspects considered together was found
to be smaller at higher levels of industrialization. Furthermore, the
findings of all the tests were consistent lending additional confidence to
It is important to note that convergence is narrowly defined in this
research. Both conceptually and empirically, we are considering convergence
in those social structures which are directly affected by industrialization.
The findings of this research can not be generalized to draw conclusions about
a tendency towards convergence of political systems, cultures or individual
values. We can only conclude that this research supports the notion that
as societies industrialize, the nature of the process places limits on the
pennissable variation in social structure.
This research represents only a preliminary effort at a test of the
convergence hypothesis. We would hope that advances in both the availability
of data and in the methodology of cross-national research will allow
a longitudinal investigation of the topic in the near future. If one
could trace the process of industrialization and social modernization
in a number of countries, a good deal more could be learned about the
process as well as the end result. Additionally it would be interesting
to extend this research to include the most industrialized capitalist
countries as well as the socialist countries.
Factor Analysis of Social Modernization
% Min. and Mfg.
Enrol. 1 and 2
Non Assoc. Groups
Percent of Total Varianc
-e 29 25 17
Cumulative Percent of
- 15 -
Measures of Dispersion
coeff of var.
coeff of var.
coeff of var.
coeff of var.
- 16 -
Ideal Scale Patterns
Roles Mobility Organization Score
3 3 3 9
3 2 3 8
2 2 2 6
2 2 1 5
- 17 -
Actual Scale Scores
- 18 -
TABLE 4 (cont.)
IV is the highest level of industrialization
- 19 -
Comparison of Errors
I Lowes t
- 20 -
Cluster of Counries versus Level of Industrialization
Sources and Measurement of Social Variables
1. Percentage of the economically active population in traditional
agriculture. (Adelman and Morris, cross-checked against U.N. and I.L.O.
2. Percentage of the economically active population in mining and
manufacturing. (Yearbook of Labor Statistics.)
3. Character of agricultural organization reflecting the range from
peasant farming to modern commercial agriculture. Countries were divided
into groups based upon area studies and the classifications were then
validated through interviews with experts. (Adelman and Morris.)
4. Importance of the indigenous middle class. An estimate based upon
(1) the percentage of the population engaged in middle class occupations
and (2) a qualitative assessment of the importance of expatriates.
(Adelman and Morris.)
5. The extent of dualism. A qualitative estimate of the degree of separa-
tion of the traditional and modern sectors. Countries are divided into
groups (as described under three above) ranging from an overwhelmingly
traditional economy to the relatively complete integration of traditional
and modern sectors. (Adelman and Morris.)
6. Modernization of outlook. A qualitative estimate (perhaps the most
subjective used in this study) of the modernization (in terms of lifestyle)
of educated urban groups and the degree of acceptance of programs of
social and political modernization among both urban and rural populations.
(Adelman and Morris . )
7. The efficiency and modernization of the bureaucracy. An estimate
(countries were divided into four groups) of the efficiency and ascriptive
versus achievement orientation of the civil service. Efficiency is judged
in terms of functionally specific relationships and rational decision making.
(Banks and Textor.)
8. The extent of interest articulation by associational groups. This re-
flects the influence of voluntary groups, such as trade unions and civic
associations. (Banks and Textor.)
9. Literacy. The percentage of the adult population (generally over fifteen
years of age) that meets a given country's standard of literacy. (Taylor
and Hudson. )
10. First and second level school enrollment ratio. The percentage of
appropriate age groups enrolled in primary and secondary schools. (Taylor
11. Third level enrollment ratio. The percentage of appropriate age
groups enrolled in university. (U.N.E.S.C.O. Statistical Yearbook.)
12. An index of human resource utilization. A linear combination of
variables ten and eleven with the latter weighted by a factor of five.
The index is suggested in Harbison, Frederick and Myers, C.A. Education ,
Manpower and Economic Development . New York: McGraw-Hill, 1964.
13. Urbanization. The percentage of the population living in cities of
100,000 or more. (Taylor and Hudson, and Banks.)
14. Transportation. An index, original to this research, intended as
a measure of the potential for horizontal mobility and economic inde-
pendence. The index is composed of a measure of road and rail length
per unit of area multipled by an index of the concentration of the popu-
lation. The latter is scored so that the more concentrated the popula-
tion (a greater proportion living in a few large cities rather than in
many smaller ones) the higher the index. It is assumed that, ceteris
paribus, countries with more diffused populations will tend to have larger
transportation networks relative to area. ( The Statesman's Yearbook .
London: MacMillan, 1967 and 1972; Ginsburg, Norton. Atlas of Economic
Development . Chicago: University of Chicago Press, 1961; and Taylor
and Hudson. )
15. Basic family structure. Countries were grouped into three classes:
those in which tribal allegiances are widespread, those in which the
extended family is the norm and those in which the nuclear family pre-
dominates. (Adelman and Morris.)
16. The extent of interest articulation by non-associational groups.
This reflects the importance of ascriptive groups such as clans and
tribes in a society. The index is qualitative; countries were divided
into groups based upon country studies. (Banks and Textor.)
17. Cultural and linquistic fractionalization. Countries are scored on
a scale ranging from .00 (extremely homogeneous) to .99 (extremely
f ractionalized) . (Atlas Narodov Mira , Academy of Sciences, Moscow,
reported in Taylor and Hudson.)
Irma Adelman and Cynthia Taft Morris, Society, Politics and Economic
Development . (Baltimore: Johns Hopkins University Press, 1967).
Arthur S. Banks, Cross-Polity Time-Series Data . (Cambridge: MIT Press, 1971)
Arthur S. Banks and Robert B. Textor, A Cross-Polity Survey . (Cambridge:
MIT press, 1963).
Charles Lewis Taylor and Michael C. Hudson, World Handbook of Political
and Social Indicators: Second Edition . (New Haven: Yale University Press,
Clark Kerr, John T. Dunlop, Frederick H. Harbison, and Charles
A. Myers, Industrialism and Industrial Man , (Cambridge: Harvard University
Press, 1960; New York, Oxford University Press, 1964), p. 238.
"Marion J. Levy, Jr. , Modernization and the Structure of Societies :
A Setting for International Affairs , (Princeton: Princeton University
Press, 1966; Princeton; Princeton Paperback, 1969), p. 11; Wilbert E. Moore,
Social Change , (Englewood Cliffs: Prentice Hall, 1963), p. 91; James Sidney
Slotkin, From Field to Factory: New Industrial Employees , (Glencoe: The
Free Press, 1960), p. 13.
Kerr et al. Industrialism and Industrial Man, p, 21,
Ibid, p. 15.
Wilbert E. Moore, The Impact of Industry , (Englewood Cliffs:
Prentice Hall, 1965), p. 45.
Levy, Modernization and the Structure of Societies , p. 744.
A traditional society is typically structured on the basis of small,
agarian, often ascriptive, independent and self sufficient, production-
consumption units. This structure (which is obviously idealized, in
Weber's sense) is not compatible with the requisites of broadscale industrial-
In an ideal traditional society, we would envision a rural production
unit which grew their own food, built their own house, and made their own
clothes. Any differentiation of roles which exists would be ascribed on
the basis of age and sex.
Again, the aspects of life as well as occupational roles are diffused.
It is difficult to separate social and economic, or occupational and recrea-
tional activities. The idea of a distinct economic sector is a relatively
Kahl called the occupational division of labor the "...economically
determined skeleton on which the flesh of modern social organization develops.'
Joseph A. Kahl, "Some Social Concomitants of Industrialization and Urbaniza-
tion", Human Organization 18 (Summer, 1959), p. 58; Also see, Gayle D. Ness,
ed. The Sociology of Economic Development: A Reader , (New York: Harper and
Row, 1970), p. 11.
Bert F. Hoselitz, "Stratification and Mobility in Industrial Society",
International Social Science Journal 16 (1964), reprinted in, William A Faunce
and William H. Form, eds.. Comparative Perspectives on Industrial Society ,
(Boston: Little Brown, 1969), p. 173.
If one idea emerges from the literature, it is that industrialization
and the extended family or clan are not compatible. Levy suiranarized the
issue in absolute terms; "(E)very society regardless of the basis from which
change took place, which has changed in the direction of relatively high
levels of modernization has been marked by a change in the ideal structures
for family types toward multilineal conjugal family units." Modernization
and the Structure of Societies , p. 417.
Melvin M. Tumin, "Competing Status Systems", in Labor Commitment
and Social Change in Developing Areas , ed. Wilbert E. Moore and Arnold
Feldman, (New York: Social Science Research Council, 1960), p. 315.
William J. Goode, "Industrialization and Family Change", in North
American Conference on the Social Implications of Technological Change , eds.
Bert F. Hoselitz and Wilbert E. Moore, (Paris: Mouton, 1963), p. 242.
Levy, Modernization and the Structure of Societies , p. 240.
Bruce F. Ryan, Social and Cultural Change , (New York: Ronald Press
Company, 1969), p. 252.
John T. Dunlop, Frederick H. Harbison, Clark Kerr, and Charles A. Myers,
Industrialism and Industrial Man Reconsidered , (Princeton: The Inter-University
Study of Human Resources in National Development, 1975), p. 37.
Ibid, pp. 36 and 37.
Bernard Karsh and Robert E. Cole, "Industrialization and the Convergence
Hypothesis: Some Aspects of Contemporary Japan", Journal of Social Issues
24 (October 1968), p. 46.
Caryl P. Haskins, "Science and Policy for a New Decade", Foreign
Affairs 49 (January 1971), p. 239.
Levy, Modernization and the Structure of Societies , p. 103.
Moore and Feldman, Labor Commitment and Social Change , p. 59 and 60.
Wilbert E. Moore, Industrialization and Labor; Social Aspects of Economic
Development , (New York: Russell and Russell, 1965), p. 13.
Kerr et al, Industrialism and Industrial Man , p. 233.
Socialist bloc nations have been excluded from many studies of
comparative socio-economic development due to problems of comparability and
the lack of data. See Irma Adelman and Cynthia Taft Morris, Society, Politics
and Economic Development , (Baltimore: John Hopkins University Press, 1967),
p. 106; and Simon Kuznets, Modern Economic Growth , (New Haven: Yale
University Press, 1966), p. 508.
Nations which did not have a population of at least one million
and a GNP of at least $500 million in 1965 were excluded from the analysis
to insure comparability of national units.
One country which met the size minimums, Saudi Arabia, had to be
dropped due to a lack of data. For more detailed information see;
Stephen J. Kobrin, "Foreign Direct Investment, Industrialization and Social
Change: Acculturation and Modernization in Developing countries," (Ph.D.
dissertation, The University of Michigan, 1975) .
Examples of university data banks are; Charles Lewis Taylor and
Michael C. Hudson, World Handbook of Political and Social Indicators :
Second Edition , (New Haven: Yale University Press, 1972) and Arthur
S. Banks, Cross-Polity Timeseries Data , (Cambridge: MIT Press, 1971).
Qualitative Indicators are found in Adelman and Morris, Society, Politics
and Economic Development and Arthur S. Banks and Robert B. Textor , A Cross-
Polity Survey , (Cambridge: MIT Press, 1963).
Missing data was estimated by; 1) reference to other comparable
indicators or, 2) regression on other correlated variables. When a distri-
bution was highly skewed the variable was transformed logarithmically.
Variables were judged spurious — not in fact measuring what they purported
to measure — if they loaded randomly in factor analysis.
The raw indicators were all selected as measures of the elements of
social structure that are likely change with industrialization. Thus, they
all are components of, and should be correlated with, some generalized notion
of social development.
The source for this indicator is the United Nations Statistical Yearbook ,
(New York: United Nations, various years).
Power consumption per capita alone may be misleading as very resource-
intensive countries may consume relatively large amounts of power in an
isolated economic (or geographic) sector. The source for national accounts
data is the United Nations Yearbook of National Accounts Statistics , (New York:
The United Nations, various years).
Problems of accuracy and comparability are both practical and
conceptual. Developing countries, almost by definition, face a shortage
of administrators and technicians. It is not reasonable to expect that
collection procedures are standardized within countries, much less between
countries. However, the major problem may be conceptual. While international
organizations such as the United Nations are making progress, at this point
there is no reason to expect that each of the indicators of social structure
is defined the same way in each of the countries studied.
34 , ,
The three factors captured seventy-one percent of the variance
the original seventeen indicators had in common. Thus the use of factor
analysis for purposes of data reduction was relatively efficient as there
was a considerable gain in simplicity at a relatively low cost.
It is perhaps more common to obtain factor scores by a more complex
technique which maintains their statistical independence. However main-
taining the independence of factors would have resulted in problems in this
case. As the first factor (roles) is very highly correlated with industrial-
ization ( a simple R of .88), maintaining factor independence would have re-
sulted in the second and third factors being orthogonal to industrialization.
As this IS the result of a statistical artifact and would have interferred
with further analysis, the alternative method of deriving factor scores was
used. The simple correlation coefficients of mobility and roles with
industrialization are .85 and .70 respectively.
The countries were simply rank ordered on industrialization and the
resulting scale was divided into four equal parts.
Patterns of preindustrial social organization vary considerably
For example, while the extended family was the norm in India, nuclear
families formed the basis of society in many preindustrial Indian communities
m the Central American highlands. Similarly some areas which are now defined
as nations in the political sense have been relatively homogeneous culturally
since antiquity, while others have been highly factionalized for centuries.
Allen L. Edwards, Techniques of Attitude Scale Cons truction. (New York-
Appleton-Century-Crofts, 1957), p. 172.
39, . ,
A very interesting application of scaleogram analysis to a cross-
national research problem can be found in; Lee Sigelman, "Lerner's Model
of Modernization: A Reanalysis", The Journal of Developing Areas 8 (July 1974)
Warren S. Torgerson, Theory and Methods of Scalin g. (New York-
John Wiley and Sons, 1967), p. 307.
Sigelman, "Lerner's Model of Modernization", p. 526.
Torgerson, Theory and Methods of Scaling , p. 318.
To avoid small classes due to the influence of extreme values the
two outlying observations, at each end of the scale, were discounted in
determining quartile cut-off points. They were, of course, included in all
Edwards, Techniques of Attitude Scale Construction , p. 186.
Thus the actual scale values for a given country are compared to the
ideal scale pattern which will minimize the number of errors. For example
if there are two ideal patterns , 121 and 222, an actual pattern of 111 would
contain one error. See, Torgerson, Theory and Methods of Scaling , p. 319.
Ibid, p. 323. However, as Sigelman notes, the CR is typically
defined for a dichotomous response pattern. If there are more than two
possible responses, the coefficient must be adjusted to take the greater
number of possible observations into account. Thus, the general formula is
CR = 1 - errors
where k is the number of columns, r the number of rows and n is the number
of categories ( or possible answers) for each entry. In the case at hand
there are three columns, fifty-nine rows and four categories for each
entry. Sigelman, "Lerner's Model of Modernization", p. 531.
Torgerson, Theory and Methods of Scaling , p. 323.
See note 46.
One of the major advantages of a scaleogram, as compared to a more
sophisticated statistical methodology, is that the case by case comparison
allows for identification and analysis of deviants.
In the West the processes of industrialization and social change were
intertwined; they developed in tandem over the course of several centuries.
In the current less developed countries, however, the situation is quite
different. The institutions of industrialization are transmitted from the
West and are often superimposed upon an existing traditional society. In the
context of the developing countries, industrialization can thus be regarded
as an exogenous variable, with its immediate effects more easily isolated
than was possible in the West.
It is difficult to justify excluding any case on an a posteriori
basis, especially if doing so improves the results. However, South Africa
is a very singular case and it is not unreasonable to argue that its
Government's policies force it out of the pale of observed relationships
linking industrialization and social change.
The clustering strategy is hierarchical, building from a single pair
which is the most homogeneous in terms of the selection criterion to a final
step in which one large cluster includes all of the cases. The variables
are converted to T scores and the selection criterion is then the average
of absolute differences in T scores. The iterative procedure works on a
pairwise basis, treating each cluster formed as a single individual by
averaging T scores. The algorithm was developed by Professor M.C. Johnson
of the School of Education of The University of Michigan.
See note 33.
Investigating historical phenomena cross-sectionally requires an
assumption that individual observations represent points on the longitudinal
path. This obviously does not present a very accurate conception of reality,
and while not destroying the usefullness of the analysis, it limits how far
one can take the results. See, Bruce M. Russett, "The Yale Political Data
Program: Experience and Prospects", in Comparing Nations: The Use of
Quantitat ive Data in Cross-National Research , eds. Richard L. Merritt and
Stein Rokkan (New Haven: Yale University Press, 1966), p. 104.
While there a major problems in terms of the availability of data
in extending the research to Include the socialist countries, it would not
be difficult to include the advanced capitalist countries. Although some
of the indicators utilized in this study are not available for the industrial-
ized countries (particularly those developed by Adelman and Morris) ,
sufficient data exists to permit quantification of the aspects of social
structure and industrialization.
MAR 2 4 1985
T-J5 143 w no.821- 75
Lorange. Peter/A framework for strateg
3 TDflO ODD bSb 21E
T-J5 143 w no.822- 75
Haehlinq von L/Optimizing claims fluct
725830 D*8KS Q.Q0.1.95&6...
3 TOfiO DOD bm blD
T-J5 143 w no.823- 75
Lihen, Gary L/The ADVISOR* project
725833 DfBKg 00019874
TDflD DD3 DDfl Q
T-J5 143 w no.824- 75
Schein, Edgar /Attitude change in the
3 TDflO 000
T-J5 143 w no.825- 75
Kobrin, Stephe/lndustrialization and v
72588P .P.S.BKS , ,. Q W 1 984.8
3 TOflD DOD bMS M3T
T-J5 143 w no.826- 75
Schein, Edgar /Managers at mid-career
3 TOaO DOD 747 T36
T-J5 143 w no.827- 75
Haehling von L/The probability distrib
726374 D*BKS 00019567
3 TOaO 000 bm b4M
25S-7fc _^^,g^ Y^cc^rd
13 .0.0 OG. t.7 s!a '-",. ""vi;,::^ -^ ^---^
HD28.M414 no.829- 76
Merton, Robert/The impact on option pr
726747 D*BKS 0QQ2PM5..
3 TOflO ODD bSb IDS