momma“ INNOVATIONS AND MODELS
on THE STRUCTURE OF REFERENCE GROUP
BEHAVIOR

ThuiEforflNDogm ofM.A.

MICHIGAN STATE UNIVERSITY

Warren David Ten‘Houfen
19:63

 

ABSTRACT

METHODOLOGICAL INNOVATIONS AND MODELS ON THE STRUCTURE OF
REFERENCE GROUP BEHAVIOR

By‘Warren David TenHouten

The adolescent (and non-adolescent) perceives all his referal
points (parents, peers, society, siblings, spouse, and self) in their
desire and ability to provide opportunities to help him attain social,
occupational, financial, and intellectual skills, goals, and roles.

This perception may or may not be "objective," but it will be
organized in a well-ordered set of preferences. The adolescent, after
perceiving, then attempts to optimize his goal-attaining potential by
becoming disprOportionately involved with (oriented to) those referal
points that are perceived as having the desire and the ability to to
provide opportunities to help.

It is hypothesised that if referents are perceived as having
neither desire nor ability to help attain a goal, the adolescent will
be involved with that referent at a low level. This hypothesis is
supported, and it is shown that the proportion of adolescents involved
with referal points at any given level is a fixed fraction of the
proportion involved at the next highest level of involvement.

It is hypothesised that if referents are perceived as having
both the desire and the ability to help attain a goal, the adolescent
will be involved with that referent at a high level. This hypothesis

is supported, and it is shown that the proportion of adolescents

Abstract warren David TenHouten

involved with referal points at any given level is a fixed fraction of
the proportion involved at the next lowest level of involvement.

A third hypothesis is deduced from.the mathematical description
of the first two hypotheses. In the case where there is either desire
but not ability or net desire but ability, the proportion of adolescents
involved at increasing levels of involvement increases at a decreasing
rate, has a maximum, and then decreases at an increasing rate. This
hypothesis is supported.

Mbre generally, it is hypothesised that Stouffer's proposition
about intervening Opportunities is a special case of a more general and
sociological formulation. The proposition that adolescents at a given.
social distance (based on perceived desire and interest to help...) away
from any and all referal points is directly proportional to the perceived
ability to provide Opportunities at that social distance, and inversely
proportional to the sum of the intervening Opportunities at all lesser

social distances, is also substantiated.

l-IE'I‘HODOLOGICAL m‘IOVATIONS AND MODELS ON THE ST RU CTURE OF
REFEIENCE GHDUP BEHAVIOR

By
Warren David TenHouten

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER 01' ARES

Department of Sociologr and Anthropology

196 3

AC KN OWLED GI’EI‘ITS

The foundation for this endeavor was laid at the University
of Chicago. In particular, Dr. Robert Grain stimulated my interest
in mathematical models of social behavior.

My greatest indebtedness is to Dr. David Gottlieb for his
guidance and encouragement of my graduate work and on this thesis.
Throughout the course of this research, his ideas, encouragement,
criticism, and direction have been invaluable.

Informal seminars with Dr. David Gottlieb and Mr. Jon Reeves
contributed heavily to the first chapter.

Drs. William H. Form and William A. Faounce read the entire-
text, and made cogent criticisms and suggestions.

A special intellectual debt is owed to Drs. Louis Guttman
and the late Samuel A. Stouffer, which is best expressed by the
frequent references to their work.

Finally, in a category by herself, Diana TenHouten provided the
encouragement, patience, support, and significant suggestions without
which this thesis would never have been written.

***********

ii

CHAPTER
1 .

2.

3.

TABLE OF CONTENTS

A MODEL FOR A CROSS-CULTURAL STUDY OF ADOLESCENT
mnoao0.0.0.60-06...0-.-000t0O‘O’O'O-O0'

A. General Orientation o o O- o. o- o. o. 0- o o. g g. g

B. Assumptions for a Cross-Cultural Model. .. .. .i
C. An Applicable Methodology: The Metatheory of

Facets.........e......~....-.~..

Do Some Further Orientations o o- O~ o- 0- 0a o~ o- o o
E. The Facet Paradigm: A Cross-Cultural Model of

“0135081“: BahaViOr o o- o- o. O« o~ o. On 0o 0- O- O:
F. Definitions of Facets and Their Elements. .. ..

C. An Inter-Facet Mechanism: A Dynamic Aspect of
theMOdelOOOOOOOOOOoooooo-o

TWOBASICPmPOSITIONSOoooooo-Oooooo

A. A Differential Equations Model. .. .. .. .. .,
B. Data and Testing of Propositions. .. .. .. .~
C. A Methodological Evaluation of the Model.
D. A Difference. Equations Model. . .. .. .. .. .—

999?

A SOCIOLOGICAL REFORMUIATION OF STOUFFER'S THEORY OF
INTERVENING OPPORI'UNITIESO o o o O o O O O O O o

A. Propositional Schema. . . . .- .
B. Operationalization: of Variables
C. Data and Test of Proposition. .

CONCLUSIONS AND IMPLICATIONS . . . . .

c I '

Page

0 O O 0

ass a seen

0
O\
Kn

LIST OF TABLES

TABLE - CHAPTER 2

l.

2.

8.

TABLE
1.

2.

3.

Responses for Five Referents to Dichotomized Levels of
Desire and Ability to Help Attain an Occupational Goal,
andtoLevelofInVOlvement..o...oo..o.o...o

Level of Involvement with Parents, Peers, Siblings, Society,
and Spouse, by Dichotomized Response to Desire and Ability
to Help Attain an Occupational Goal. ..... . . ....... . . .

Level of Involvement with Parents, Peers, and Society, by
Dichotomized Response to Desire and Ability to Help Attain
an Occupational Goal: Probability Distribution . . . . . . .

Empirical and Theoretical Distributions of Level of Involve-
ment with Five Referents Perceived as Having No Desire and
No Ability to Help Attain an Occupational Goal.. . . ... . .

Chi-square values for'Different Choices of k in B

2

Empirical and Theoretical Distributions of Level of Involve—
ment with Five Referents Perceived as Having Both Desire and
Ability to Help Attain an Occupational Goal. . . . . . . . .

Empirical and Theoretical Distributions of Level of Involve—
ment with Five Referents Perceived as Having Either*Desire
but No Ability OP~NO Desire but Ability to Help Attain an
OccupationalGoal. O O O O O O O-Ov. .-O°O~O O O O O O O O O

.Comparisons of Chi-square and iKDiff.) for all Cases of

Perceived Desire and Ability to Help Attain an Occupational

GoalOOOOOOOOO~O-O»O.OOOO-O-OOO-O-O.OOO.O-OO

Page

O30

.33

.42

.143

.45

.47

- CHAPTER.3 Page

Percentage "Very'Intensely Interested" or "Some Interest" at
Each Rank of "Agreement" with Choice of Occupational Goal,
fOrFiveReferentSOOOOOOOOOOOOOOOOOOOOO.

Social Distance, 5, Social Distance Rank (1.9). and Number
of Cases, N, for Five Referal Points . ... . . . . . . . . .

Perceived Ability to Provide Opportunities to Help Attain an

Occupational Goal, Level of Involvement, and Social Distance,
for Five Referal Points: (N) and Probability Distribution. ..

iv

.53

.5”

~57

LIST OF TABLES - Continued

TABLE

4. Test of Hypothesis for all Referal

5.
6.
7.
8.
9.

Test
Test
Test
Test

Test

of Hypothesis for Parents . .

of Hypothesis for Peers . . ..

of Hypothesis for Siblings. .
of Hypothesis for Society . .

of Hypothesis for Spouse. . .

Points.

LIST OF FIGURES

FIGURE - CHAPTER.2 Page

2.

3.

Level of Involvement with Parents, Peers, Siblings, Society,
and Spouse, by Dichotomized Response to Desire and Ability
to Help Attain an occupational Goal. O O OOO O O-O-0.0~.-O-0-03u

Level of Involvement with Parents, Peers, and Society, by
Dichotomized Response to Desire and Ability to Help Attain
an Occupational Goal: Probability Distribution . .. .. .. .. . .. .36

Level of Involvement with Five Referents Perceived as Having
HO Desire and No Ability to Help Attain an Occupational Goal. #1

Level of Involvement with Five Referents Perceived as Having
Both Desire and Ability to Help Attain an Occupational Goal. .44

Level of Involvement with Five Referents Perceived as Having
Either’Desire but No Ability or No Desire but Ability to Help
Attain an occupational Goal. O O O O O 0-.vO-0«0!O O O O O O Ous

FIGURE - CHAPTER 3

l.

2.

Percentage Distribution of Orientation over Each Social Dist.
ance, for all Raferal POintS O O O O O-O O-O O O O O-O O O O O58

Percentage Distribution of Orientation over Each Social Dist-
ance, for Parents. . . . . . . . . . . . . . . . . . . . . . .60

Percentage Distribution of Orientation over Each Social Dist-
ance, for PeerS. O O O O O OOO O O O O O O O O O O O O O O O O61

Percentage Distribution of Orientation over Each Social Dist-
ance, for Siblings O O O-O O O~O O O O O O O O O O O O O O O O62

Percentage Distribution of Orientation over Each Social Dist-
ance, for SOCietyO O O O O O O O O O O O O O O O O O O O O O O63

Percentage Distribution of Orientation over Each Social Dist-
ance, for Spouse . . . . . . . . . . . . . . . . . . . . . . .64

This thesis began with an attempt to construct a model for a
cross-cultural study of adolescent behavior. The result of this
effort was a model much broader in scope that the original intent.

Some broad hypotheses were develOped out of this model. They
were empirically tested in a questionnaire administered to an under:
graduate sociology class. The hypOtheses were all supported by the
data, but the pOpulation contained a significant number of non-
adolescents.

The result is that this is a general model for reference
group behavior, with some general propositions about reference
group interaction.

No attempt is made here to explain the structural variations of
respondents in different social and demographic categories. ‘What is
done is the present a methodological and propositional framework for
examining adolescent-reference group interaction in different social
structural conditions. In this sense, this research is the early

groundwork for a cross—cultural study.

CHAPTER I
A MODEL FOR A CROSS-CULTURAL STUDY OF ADOLESCENT BEHAVIOR
A. General Orientation

There is much empirical justification for contending that
the literature about adolescence reveals some striking continuities
in the structure of adolescent behavior. There appear to be certain
elementary processes involved that go on in very different types
of societies. Further, this elementary social behavior may be
found in many contexts of sociological interaction, varying from
small groups to complex organizations.

What are the basic patterns of interaction between social
units such as the adolescent and his reference groups: that is,
what social organization is involved? A paper by Gottlieb and
Guttman begins with the general orientation that adolescents behave
much the same anywhere, in any socio-cultural context, in that
they will ask help from referents they perceive as desiring to help
them attain a goal, and will not ask help from other referents; and
they will turn to those that they perceive as desiring to help them,
and not to those that they do not.1 Any research or theoretical
analysis must begin with such predilictions for a certain way of

stating the problem to be investigated, for certain kinds of hypo-

 

lDavid Gottlieb and Louis Guttman, "Working Paper No. l,“
Mimeographed paper, November, 1962, pp. 1-3.

1

TH

 

theses, assumptions, and postulates. The values, judgments, and
the existing knowledge of the investigators are of course involved
in this. When the orientation reaches a certain level of specificity,

it can be described as a model.
B. Assumptions for a Cross-Glltural Model

Assumptions, by definition, are not directly under inves-
tigation. But they are characterized by the possibility that an
empirical study using a certain set of assumption might generate
findings that would demand that one or more be discarded or modified.

It is necessary that assumptions are not at variance with
postulates in the model: this condition will be rigorously met.

Janowitz defines an assumption as “...a proposition about a
relationship within objects of study or linking an object of study
to other variables, which is taken for granted in the investigation

of the immediate problem at hand.“2

This will serve as a working
definition for the following assumptions, which will be used
thralghalt the entire study.
I. The adolescent wishes to attain skills, goals, or roles.3
2. a. The adolescent perceives his parents, peers, society,
siblings, and himself.

b. The adolescent perceives these five in terms of a

specific skill, i._e_., as potential references.

 

2Morris Janowitz, ”Working Delimitation of the Subject Matter of Social
Organization Research, " Milneographed paper, University of amicago,
(Fa-ll, 1962). p0 2O

3 Skills, goals and roles are to be understood as ends, and will be
used interchangeably.

c. The adolescent perceives that these four referents
have differential power means (ability) and
intentions to help him toward his goal.

3. The five potential referents mentioned in 2. a. con-
stitute an exhaustive list, regardless of the
cultural setting.

1+. Goals may vary from society to society.

5. Nothing. is said here about the criteria used by the
adolescent in the selection of ends.

6. No statement is ever made about the "objectivity" (the
validity) of the adolescent's perceptions with
respect to goals or reference groups.

7. Perception precedes involvement. Perceptions can be made
without involvement, but involvement cannot occur
without perception..

8. The adolescent is the initiator in his involvement.

These will be referred to as “preliminary assumptions.a Other

assumptions will be added as the model is built, and these eight will

be re—examined.

C. An Applicable Methodology: The Metatheozy of Facets

A group of postulates is a group of statements which purports
to define an object of study. Here, the object of study is a m
_o_f_ interaction. In Guttman's terminology, a postulate can be a facet, or
an element of a facet--more technically, a statement about a facet. The

five facets will constitute the observational basis for developing and

1+

deriving a metatheoretical system, or 3993);. m theory is ultim-
ately defined in terms of its facets, 25.2... it must be concerned
with various sets of elements. The presentation of the five
facets in the next sections will rely heavily on “Facet Design" as
developed by Guttman.“

For now, a very brief discussion of Facet Design and Facet
Analysis (which together are defined as Facet Metatheory) is
sufficient. I

If V is a set (a collection of things) defined as the (hr-
tesian product (Cartesian space) of two sets A and B such that
V 3 AXB 3 AB, then A and B are facets of V. This is all that
“facet" means. Suppose that A is composed of elements a1. a2.
......, an (the number of elements may even be infinite, but that
is of no concern here), and B is composed of elements b1, b2,
......, bm’ Clearly V has n times m elements, where "times" means
ordinary arithmetic multiplication. This is the only sense in which
the word "product” is used in Cartesian product: it does not imply

manipulation, only the structuring of facets and their elements

 

in (here, sociological) space. There are n ways of choosing an
element from A and m ways of choosing an element from B, so

there are n times m possible profiles over A taken with B.

 

1""The technical concept Henoted here by 'facet' occurs widely

in the social and other sciences, as well as in mathematics, but
strangely has never been given a standard name. Christening the
concept by 'facet' was formally proposed by Louis Guttman, "An
Outline or Some New Methodology for Social Research,“ Public 012-;
inion marterly, 18 (Winter, 1951t-55), pp. 395-414, esp. p. 399.
rominent sociologists who have used the concept extensively in
their own work, albeit ratherinformally, include Pareto, von Viese,
Parsons, Sorokin, Dodd, among others,“ in Guttman, ”A Structural
Theory for Intergroup Beliefs and Action,” American Sociological
Review, 2”: (June, 1959), p. ljff.

 

For any element of V, call it v, v = aXb 9- ab, where it is under-
stood that a is an element of A and b is an element of B. v is
defined as a profile over V; a and b are defined as the co-ordinates
of V. The set V can be defined as the set of all possible profiles
v over A and 13.5 This example has only two facets, but of course
a Cartesian space can be composed of any number of facets.

In this model, there will be five facets, symbolized as a
Cartesian product V = PABCD, where

P = the, population of adolescents: a variable number of elements,

A = behavior of adolescent: two elements,

B 3 behavioral properties of referents : two elements ,

C = references: five elements, and

D 3 Ends (Skills, Goals, Roles).

'v will be referred to as the "Facet Paradigm". v-s- PABCD is
a Cartesian product, and v = pabcd is a profile over V. For any
set V, there are elements v0, v1..." vn. The set of all subsets
of V is called the power set, denoted by C? (V). The subset with
no elements is called the empty set and denoted ¢.

By considering only A. B, and C as variable facets, where P

and D are $5563 facets, UV" = ABC, contains fifty-three profiles,

that is fifty-three unique combinations of vCV*. Thus,

 

5For a fuller explication of the vocabulary of facets, see
Guttman, ”Notes on Terminology for Facet Theory, " reprinted from the
Proceed s of the Fifteenth International Co ress of Ps cholo
Amsterdam: North-Holland Publishing Company, 1959 . Pp. 130-132.
A good introduction to abstract systems is Daniel T. Finkbeiner, II,
Introduction to Matrices and Linear Transformations (San Francisco
and London: W. H. Fmeman and Company, 1950), pp. 1-18.

    
 

 

0B6 ' 1[mac 1* XAB+XAC + XBC + FA + Xa+ X0:

53 I (2X35) +8 (212) + (21(5) 4» (215) 4 2 + 2 + 5.

The task in this research project is to "explain" xPABCD'
This will require an hypothetical-deductive theoretical system
including hypotheses, assumptions, and postulates. This is Facet
AnalEis.6 Questionnaires will be administered (in a sense, tests)
to a population of adolescents, to test the level of validity of
components of propositional systems. The M of the entire
experiment is

X PABCD" Real scores, how the adolescent scores and interrelates

variables.

There is a slight distinction between a profile and a variable.
For example, alcl is both a profile and a variable, but clc2 is not a
profile because it contains two elements from one facet. So the
explanation of the elements in the model requires the addition of all
the variables that are not profiles. For the facets A, B, C. and D
there are 21+ such cases within facets, and many more between facets.

Facet designing consists of defining the variables of a given
study as Cartesian products of the elements of sets of more inclusive
concepts. The novely, and the usefulness, of Facet Metatheory is that
it formalizes the intuitive processes involved in constructing relevant

concepts and variables for a scientific study. This "spelling out" of facets

 

6Guttman, "What Lies Ahead for Factor Analysis?," Elucational and

Psychological Measurement, 18 (No. 3, 1958), pp. 506-511. Such an

explanatory system does not need to exhaust all interelations of all
variables.

decides which of these should be retained. A result of this for-
malization is that it clarifies which elements are common to different
variables and which ones are dissimilar among themselizes.7 For

example, given four variables 1. aiblcl such that they

20 alblcz
30 3.113202
40 a£02632

they form a perfect Guttman scale, we can say that the four variables
are ordered by similarity or contiguity. Foa's contiguity hypo-
thesis states that the empirical correlations between variables
increases with the semantical similarity of the elements of the fa-
cets in the profiles. For example, 1 and 2 differ only in the C
facet, whereas 1 and # differ in all three facets. Hence the con- .
tiguity hypothesis would predict that 1 and 2 will be more highly
correlated than 1 and h, in.making the "leap" from semantical struc-
ture to the statistical structure of observed (scored) data§'9

The utility of scaling variables is illustrated by'Guttman's refor-

mulation of a study by Bastide and van den Berghe.lo

 

7Uriel G. Fee, "The Scientific Development of the Isreal In-
stitute of Applied Social Research,“ Mimeographed paper, n.d.

‘ 8Uriel 6. Fee, "The Contiguity Principle in the Structure of
Interpersonal Relations", Human Relations, 11 (August, 1958) pp. 229-238.

9The distinction between semantical and statistical structures,
as different levels of systems of components, is developed by
Guttman in "The Principal Components of Scalable Attitudes", and
“A New Approach to Factor Analysis: The Radex“, both in Paul F.
Lazersfeld, ed., Mathematical Thinking in the Social Sciences
walencoe, Illinois: The Free Press, 195“).

10
Roger Bastide and Pierre van den Berghe, "Stereotypes, Norms
and Interracial Behavior in Sao Paulo, Brazil,I American Sociolo-
gical Review, 22 (December, 1957), pp. 689-694, which is discussed

A = Subject's behavior B = Rbferent C = Referent's Intergroup

 

Behavior
al 2 belief bl a subject's group c1 a comparative
a2 I overt action b2 8 subject himself c2 = interactive
Subuniverse Profile
1. Stereoti e b c
JP . a11.1
2. Norm. a b c
l 1 2
3. aHypothetlcal Interaction albzc2
4. P n Int ct' n b
erso al era 10 a2 2c2
An immediate result of this semantic analysis is to show that the
3

four variables are only some of the 2 = 8 combinations of these elements.
These four variables are ordered semantically: they are also

ordered sociologically. The meaning of the sociological order is the

degree of face-to—face contact with the Negroes, progressing from a weak

to a strong type of behavior of the subject vis-apvis Negroes. The

degree is strongest.in personal interaction, not so strong in hypothetical

interaction, less.yet in.norm, and weakest in stereotype.11 The facet

pattern shows, for example, that the nearest neighbor of stereotype is

norm: the differ only within facet C. Such reasoning shows the semantic

contiguity'of the attitudes.

 

in Guttman, 'A.Structura1 Theory for Intergroup Beliefs and Action,I
Anerican Sociological ggvi‘ew, 24, (June, 1959), pp. 3-4. ,

llGuttman, IThe Structuring of Sociological Spaces,‘ Technical
Note No. 3, Contract N. AF 61(052)-121, (December, 1961), p. 10.
This report is a portionrof the Proceeding§_of the Fourth Internationgl
Congress of Sociology, Stresa, Italy, (1959).

0n the non-empirical, conceptual level, perfect scales are
possible, as the example shows.

The next problem is how to u_§_e_ this semantic structure in order
to make predictions about the statistical relations among the parts
of the substructed semantic space. In this study, as with the
example, this can be attained by computing correlation matrices. This
gives considerable insight into the adequacy of the facets in a model.
If the contiguity hypothesis holds (Foa doesn't contend it always will,
even if the facets are adequate), there is some methodological justifi-
cation for rdtaining the substructed sociological space. In the example,
the contiguity hypothesis was supported, and the pattern of inter-
correlations formed a simplex, as predicted.

Considerable parsimony is attained by Facet Design in constructing
questionnaires, as all questions can be structured with a clear picture
of the variables involved.. The several hypotheses tested in Chapters 2
and 3 are based on four questions about five referents.

The following criteria seem quite fundamental to any research design.
Facet Metatheory certainly satisfies these criteria, and emerges as a
powerful methodological tool for the structuring of sociological spaces:

1. The need for clear dimensions.

2. The need to order variables, 3.5., from higher to lower. But
however they are ordered, variables should be articulate and
not ambiguous.

3. The need to work with variables instead of concepts.

4. The need for a neutral family name for any given variable.
Variables are frequently named only on one end of a spectrum,

2.3., "alienation“ means the negative of something that has
a positive, "commitment." Alienhtion and committment are

10

the negative and the positive extremes on a continuium of
"involvement." Here there are three concepts, but only one
of them is a variable. The family name is necessary, both
for comparisons within a family, and between families.

5. Variables may be close, and hide under'different labels, but
they should be distinct.

D. Some Further Orientations

Hill the hypotheses we generate from the model, and the model
itself, involve decisional processes on the part of the adolescent.
Simon suggests two alternatives for formulation of such a model: this
model will contain itself entirely to the first (see assumptions 6,
page 3):

l. Inquiring into the properties of the choosing organism, and

2. Inquiring into the enviormment of choice.12

It is the choice of using only the first alternative that gives
this study cross-cultural potentialities. Studies of adolescent
behavior based on the second alternative must wait on a great deal of
theory and substantively comparable data on the socio-economic structures
of different kinds of societies.

The adolescent's perceptions, particularly his information about
certain properties of potential or actual reference groups, is an
important behavioral property; !e are assuming,.it seems, that the
adolescent has knowledge about the relevant aSpects of his reference

groups in his social enviornment with respect to his goals. This knowledge may

 

l2Her'bert Simon, ”A Behavioral Mbdel of Rational Choice,! in
Simon, Mbdels of Man_(New York, London: John Wiley and Sons, Inc.,

1957), p. 242.

11

not be I'objective, but it will be organized in a relatively well-
ordered set of preferences. And we assume that the adolescent has
evaluative (or "ranking") skills that will enable him to "Isl-
culate" alternative choices of involvement (orientation) to optimize
his preferences.

Simon distinguishes three givens within which "rational"
choice is made:

1. The set of alternatives open to choice,

2. Relations that determine the goal-attainment as a function 13

of the alternative chosen,

3. A preference ordering of alternatives.1""

An implicit orientation to this study is the notion that
the adolescent is somehow Optimizing, 33.2., choosing the reference
group that has the intention and power to help him attain a par-
ticular skill, role, or goal. To the extent that he optimizes
reference groups (on the basis of his perceptions), the adolescent
is rational. We never hypothesize that the adolescent is 11m
rational, but the general notion is worth bearing in mind -- even
if only to see under what conditions the adolescent will behave

non-rationally. This begins to sound like Economic Man, but there

 

13A profile is a functional relationship between elements
from sets. A function is a binary relation, 3.2., a relation
whose Cartesian product is defined by two facets. A function is
identical to a mapping. See, 2.5., Robert R. Stoll, Sets, Logic,
and A951; omatic Theories, (In. I. See also Guttman, "Notes on the
Terminology of Facet Theory," cited in footnote 5.

1“Simon, 93. _c_:_i_t_.

 

 

12

are crucial differences. The rationality (or lack of it, as the
case may be) of an adolescent is hypothesiszed not "globally" or
"objectively," as with Economic Kan, but rather as restricted to
maximizing his goals on the basis of his own perceptions and

knowledge.

E. The Facet Paradigm: A Cross-Cultural Model of Adolescent

 

 

 

Behavior
. The following schema appears in the paper by Gottlieb and Guttman:l5

Adolescent; Reference Relation Object of Relation

(society) (need for)
(adolescent) ranks (parents) in their (status on him on (skills)

asking for
(Doors) help)
(himself) (ability to help)

Note: Facets are underlined, and elements are in parentheses.
"Ranks" is not a facet because it is what the adolescent
does in an interviewing situation,‘;.g., it is a statement about his
interaction within the interview situation rather than with is references.
we have assumed that the "...adolescent has evaluative (or 'ranking')
skills that will enable him.to'balculate' alternative choices...,"
(page 12). Also, the preliminary assumptions on page 3 indicate

that the sociological content of evaluation or "ranking" is differential.

 

15Cf. Vladimar Cervinka, reported by Stuart C. Dodd, "A.Dimen-
sional Theory of Groups," Sociometry, XI, (Februaryehay, 1948), pp.
100-107. Cervinka develOped a theory utilizing essentially the same
formulations as used in this paper: both formulations were developed
independentLy.

13

perception. This is equivalent to saying that the adolescent per-
ceives facets and elements of facets as variables. Also see
assumption 2. c.: "The adolescent perceives these reference

groups as having differential power means (ability) and intentions
to help him.toward his goal." In summary, ranking is evaluative
perception. Evaluation is behavior in an interview situation;
perception is behavior vis-a-vis reference groups.

There seems to be one other ambiguity. It is in the
Relation facet. (Need for) and (Ability to help) are behavioral
properties of the References, whereas (Status on asking for help)
is a property of the Adolescent.

So we have two properties for the Adolescent: 1. (Perception);
and 2. (Status on asking for help), which we will call involvement
or orientation. And'we also have two properties for the Reference:
1. (Need for), which we will call desire or intention; and 2. (Ab-p
ility to help).

Considerable space has been used to describe the substantive
aspects of the study by Bastide and van den Berghe which was
reformulated by Guttman (see pages 8-10; sources are described in
fottnote 9). There is a remarkable similarity to the facets
developed in Guttman's paper and in those just introduced in this

section. The comparison can be displayed in the following paradigm:

l4

 

 

 

 

 

 

 

Guttman's Analysis of Further Analysis of Gottlieb
Bastide and van den and Guttman's Mimeographed
Berghe ‘ Paper
Subject's Behavior Facet Adolescent's thavior

(belief) (element) (evaluative perception)

(overt action) (element) (orientation or involve-

. ment)

‘Referent's Intergroup Facet Behgvioral Properties of
thavior Referents

(comparative) (element) . (need for desire,

intention)

(interactive) (element) (ability to help)
Referent Facet Referent

(subject) (element) , (self)

(subject's group) (element) (subjects groupg: parents,

peers, society, siblings)

 

Note: Guttman's paper labeled the second and third facets differs
ently than they will be in this paper. He call the Referent "B" and
Referent's Integgroup Behavior I‘0"; here, B and C will be reversed in
the "Facet Paradigm" to follow. There is semantical commutativity
between these facets.

Guttman's analysis of Bastide and van den Berghe, and the
further analysis can both be summarized with the following statement:
Type of behavior of’g_§ubject vis-a—vis a type of intergroup behavior
of a type of referent (with respect to a certain goal of the subject.

This seems to be a generic statement about interaction.

15

The facets and elements in the model will be presented next,

In the Facet Paradigm. The terms will be presented now and defined,

in Section F. This is a necessary first step in model building, or

in research in general--to carefully define the variables that are

to be used. It is then necessary to develop postulates and

propositions relating variables to each other.

in Section G, and in.Chapters 2 and 3.

This will be attempted

Some mechanisms deduced from

the assumptions, and some general propositions will be advanced.

Tpg Facet Paradigp: V = PABCD

 

 

 

 

 

Facets Population Adolescent's Behavioral Referents Skills,'
pg Behavior Properties Goals,
Adolescents ‘pf Roles
Referents
to help
. vis-a-vis P attain
V I P A ' B C’ D
(p ) (perceives) (desire (self) (social)
0 a1 or cl d1
(p1) intention)
bl (parents) (occupa-
_ (p ) c tional)
(elements) 2 (becomes (power 2 d
. involved means or (peers) 2
. with or ability c (intel-
. oriented to) to provide 3 lectual)
. a oppora (society) d
(p ) 2 tunities) ch 3
n b (financial)
2 (siblings) d4
°5

 

An example of a profile over'V': PABCD: Let po represent lower

class adolescents in a given pOpulation, and choose a v'= poaibzcédz.

16

v I (The lower class adolescent's) (perceptions) vis-a-vis the
(ability to provide opportunities) of his (parents) to help-him
attain an (occupation);

V = A type of Adolescent's Behavior vis-a-vis a type of Interactive
Behavioral Property of a type of Referent to help him attain a type

of Skill.
F. gefinitions of Facets and Their Elements

P: The populjgtion of adolesceng. This refers to the sample we
draw. The work Rmiverse" will be reserved for the parameter of
adolescents- from which the sample is drawn. Adolescents will be
defined in terms of an age bracket from a particular universe, _e_.g.,
all males in Tokyo from 14 to 18 years. of age. In another social
setting, there might be another choice of ages defined as adolescents,
but this will present no great methodological problems.

Elements for this facet are obtained by characterizing a sample
by some criterion, such as father's occupatibn, or such as religious
preference.

Az'Adolescent's behavior. The two elements in this facet are
exhaustive with respect to goal-directed behavior.

a1 : (perception). Bastide and van den Berghe call this element
"belief." Evaluative perception is approximately the same, except that
we know the adolescent is able to evaluate by assumption. The content
of perception will be elaborated in the next section, "An Interfacet
Mechanism."

a2: (involvement, or orientation). This is interpersonal exchange,

17

and overt behavior, in contradistinction to al, which is an internally
contained, or covert, behavior. Involvement means the adolescent has
actually made a.gppigp on the basis of his perceptions (by hypothesis),
and is involved with his referents in certain ways. In a sense, he is
either "using" them or’not.

Involvement can vary from positive "committment," to a nearszero
"calculative," to negative "alienation."l6

B: Interactive properties of referenpp. This facet is in a sense
the set of perceived outcomes and actual outcomes of facet A. The
adolescent chooses from.A = (a1, a2) in certain ways. How he chooses
will be hypothetical, but choice must be somehow made from both a1 and a2.
So the two elements of B determine a functional relationship determining
the perceived outcomes and orientational outcomes. The variables alb1
and alb2 are perceived outcomes; the variablesazbl and azb2 are actual
outcomes, that is, outcomes of interaction. Hence there is a mapping of
the elements of A into the elements of B.

bl: (intention or'desire). Goldhamer and Shils define a person
as having "...power to the extent that he influences the behavior

. C O O 1
of others in accordance with his own intentions." 7

 

16This is closely related to Etzioni's use of "orientation," in
Amitia Etzioni, A.Co arative Analysis of Complex Organizations (Glencoe,
Illinois: The Free Press, 1951), Chapter 1, in which Etzioni develops a
terminology for a compliance relation between elites and lower
participants in complex organizations.

l7Herbert Goldhamer and Edward A. Shils, "types of Power and Status,"
Apprican Journal of Sociology, 45, (September, 1939), p. 171. Behavior is
defined as both overt and covert,‘irg., power refers to both elements of
adolescent's behaviors-perception and orientation.

18

It is in this cptative sense that "intention" will be used in
the model. It can also be viewed as the perceived social interactional
"distance" from referents.

In their first working paper, Gottlieb and Guttman used "desire"
for'this element. The two words do not mean quite the same thing. For
example, if an adolescent with high ability perceives perceives his
parents are able to help him attain a goal, they'may'gppppd to help
him because "Its his life, and we should help him attain his own goals,"
but may at the same time be disappointed in his choice of goal, and
lack psychological "desire" to help him, or lack "agreement" with his
choice of goal. Considerable caution must be used in naming this '
element "desire" or "agreement."

we also want to be consistent with Bastide and van den Berghe's
usage of this element as comparative relative to the adolescent's goal.
"Desire" connotes interaction outside of the goal-directed context; the
model is concerned with the adolescent's goals, not the goals of his
referents.

b2: Ability or power’means. Parson's definition of power illus-
trates another aspect of this facet, the interactive. (He says "Power-
is the ability'ppngip another actor carry out roles or norms (here,
ends) he supports."l8 (A referent can have power potentioal (ability)
without (intention) to help; and a referent can have (intention) without

(ability).

 

18Talcott Parsons, The Social System.(G1encoe, Illinois: The
Free Press, 1951), p. 121, in Etzioni, Ibid., p. 4. Emphasis and
parenthetical insert added. Etzioni uses "power means" in a sense
equivalent to (ability): the terms will be used interchangeably here.

1'9

C: Referents, or referal mints. By assumption 3 (page 3),
parents, peers, siblings, society, and self are an exhaustive list of
elements for the referent facet. Society is the largest category: it
makes the list exhaustive. In chapters to follow, however, (self)
will be left out and (spouse) will be added.

D: ikillsL goalsL and roleis, Homans makes an adequate definition
of goals: "A goal (end) is a part of the behavior to be explained, by
assuming that goal achievement is a reward which one's activities get
from another person, another group, or the environment."19 To this,
the (subject's group) element, the (subject himself) element must be added.
The only difference between Homans and this model is the length of the

chain of activities which are goal-attaining.
G. in Inter-Facet Mechanism: A Dynamic Aspect of the Model

The dynamic. postulates in this section are largely deduced from
the eight preliminary assumptions. In the mechanism, we will consider
only facets A and B, which are the behavioral facets.

In his-work on the Metatheory of Facets, Guttman again and again
substructs; facets in sociological space that are called stimulus facets
and response facets. This is of course more a consequence of Guttman's .
substantive interests than of Facet Theory itself, but at the same time
it is certainly true that Facet Theory lends itself readily to this kind

4

~ ~ ”George-C. Romans," Social. Behavior:’: Its Elementa oma (New
York, Burlingame::Harcourt, Brace & World, Inc., 19 , p-89.
Parenthetical insert added.

   

20

of analysis.

In this section, we will use the terminology of stimulus and
response, but we do not intend the terms to be interpreted in a
purely psychological sense; the mechanism is social-psychological,
or even sociological, in that interpersonal exchange is involved,.as:
well as perception.

The B facet can be considered a stimulus facet,,and the A facet
can be considered a response facet.. This is closely related to our
assumptions.

The adolescent is stimulated perceptually by the two elements of
B, that is, albl and albz' Following this perception (assumption 7,
page 3) the adolescent responds (becomes involved with) a referent by
initiating (by assumption 8,.page 3) an interactive process with the
referent, on the basis of his perception of B, that is, azci. The
adolescent, upon being stimulated cognatively, which is not interaction,
then responds (a2) by his involvement. His involvement is interaction.
For the purposes of this model, we will initially be concerned with
interactions initiated by the adolescent.

This sequence of events is usually neither rapid nor continuous.
For example, an adolescent may perceive the intentions and ability of
his parents to help him.attain an occupational goal for several years
befbre he becomes involved with them in attaining this goal.

To review the close relation of the mechanism to our assumptions,
it can be stated that the mechanism satisfies assumption 8, "Adolescent
is the initiator," and also assumption 7, "Perception precedes involve-

ment."

2.1

The properties of this mechanism can be formalized in a manner
used by Guttman.20

There are three facets involved:

(P I population of respondents, here the P facet,

S = stimuli variables, here the B facet, and

R = response categories, here the A facet.

We are considering a set P who are responding (R) to alternatives
which are stimuli (s). i

In general, P—b Rs(sé S), where Rs are possible categories for
stimuli = (a1, a2) and where the stimulus s (séS) = B facet.

PS—oR, 3.3., adolescent will respond to the stimuli.

The stimuli are faceted: S = 5152. So PSlSZ—DR. For any péP,
in conjunction: with 515.31 and $2652, there corresponds a particular
response.

In the inter-facet mechanism, interaction, can be reformulated to

mean. an exchange of stimuli between the adolescent and some sociological

unit. The exchange need not be continuous.

 

20Guttman,. "Introduction to Facet Design and Analysis," p. 132.

Q

 

.l '1‘: liiwi"‘l. Id!
.3 I "I.

CHAPTER 2

TWO BASIC PROPOSITIONS

A. A Differential Equations Model

Our general sociological orientation to the substantive topic
of adolescent behavior, and more generally, to reference group behavior,
presented in Chapter 1 is not to be confused with sociological £22932.
Broad postulates were presented, indicating the types of variables to be
accounted for. Facet analysis enabled us to reach a level of specificity
which can be described as a‘ggggl,for a cross-cultural study. The stage
has now been reached where relations between variables can be specified.1
The initial orientation that there are elementazy social processes
involved in this interactive behavior can be substantiated if basic
hypotheses can be developed, tested, and confirmed. If this can be done,
it will be possible to develop propositions that will hold up cross-
culturally.

In WbrkinggPapgr Number 1, Gottlieb and Guttman hypothesised

 

that adolescents will become involved with referents they perceive as
having the ability to help them attain goals, and will not become in-
volved with referents they perceive as not having the ability to help
them attain goals. hey also hypothesised that adolescents will become
involved with referents they perceive as having the desire to help them
attain goals, and will not become involved with referents they perceive

as not having the desire to help them attain goals.

 

lebert K. Herton, "The Bearing of Sociological Theory on
Empirical Research," in §pcial Theogy and Social Structure (Glencoe,
Illinois: The Free Press, 1957), pp. 86-89.

22

23

These "two basic propositions" can be stated more formally, in

the vocabulary and notation of our model:

 

1. If an adolescent (perceives) that a referent has neither the
(desire) nor the (ability) to help him attain a given goal, he will be
(involved) with that referent at a low level.

2. If an adolescent (perceives) that a referent has both the (desire)
and the (ability) to help him attain a given goal, he will be (involved)
with that referent at a high level.

 

we will now develop a system of differential equations to de-
scribe these propositions. By doing so, we are also able to $29222
a propositional statement about the additive union of the two ambiguous
cases: desire but no ability, and ability but no desire.

Let the following notation describe B a (B0, Bl’ B2):
(albl) = perceived desire to help, where

l - evaluated positively, and

0 I evaluated negatively;
(aibz) 3 perceived ability to help, where

1

evaluated positively, and

0 evaluated negatively;
B = (albl, aibz) a (perceived desire, perceived ability);

B0 = (O, O) = adolescent evaluates both the desire and the ability of
a referent to help him attain a goal negatively;

Bl I (O, l) = adolescent evaluates desire negatively but ability
a positively;

Blb = (l, 0) = adolescent evaluates desire positively but ability
negativexy;

Bl I (O, l) U (1, O) = all the B1a and all the Blb cases;

B2 = (l, l) = adolescent evaluated both desire and ability positively.

24

From.our'two propositions, we can make statements about

changes in 36 and in B with respect to changes in involvement,

2
where 3-: f(a2). ‘we are implicitly considering level of involvement
as some function of time. For all changes in both of these "congruent"
cases, we can ignore the negligable (if even possible) effect of the

instantly, or of changing from.B

2

adolescent changing from BC to B 2

to B0 instantly.

Using involvement as some function of time is sociologically
realistic, as we are dealing with a socialization process, of learning
to attain goals over a period of time and activity, And, by definition,
goals are not attained.

Let m equal the net probability of changing fromBo to B1 at
varying levels of involvement. By our porpositions, as involvement
increases, the probability of an adolescent being Bo decreases.

Hence, the net sign of mvis negative.

. Let k equal the net probability of changing fromB1 to B2 at
varying levels of involvement. By our'propositions, as involvement
increases, the probability of an adolescent being B2 increases.

we can represent the statements made so far with the following
system of differential equations: '

dB

(1) 48-11130,
da2

dB1

(2) -——.I mBo - kBe , and

leg-
MN

(3)

Q.
m
N
II
1...:

25

Our*data, to be presented in the next section, will give us
empirical epproximations to the graphs of these three equations.
By solving the equations for B, and differentiation of the equations,
liޤ., finding the second derivatives, we can determine the general
shapes of the curves for B.

(la). Solution of Bo:

 

dBo
a? ‘ 'me
2
dBo Boundary conditiong:
‘ -m(da )°
— I ' 0
B0 2 At a2 = 0, B0 = ce = c.
103686 = INK???) 4' 0.3
B = ce—m a2
0
(lb). Solution of B1:
dB
__E_= cme-m(a2) - kB ; Boundagy condition:
da 1
2
-m(a ) ° ' =£E~
dBl = cme 2 (daz) - kBl(da2); k-m

Using an integrating factor, ek<a22

ek(a2)dB * Blkek(a2) : mek(a2)e'm(82)da ;

l 2
: - -k
31 eem‘az’ - —— “2’.

(1c). Using a probability distrubution, for all a2,

BO+B1*B2=1.

(la'). Second derivative of B0:

dzB
° —m(dB ) = -m(-mB ) = m?B , which is greater than 0 for all a .
352" ° ° ° 2
2

26

(lb’). Second derivative of B1:

dZBl
(132 a d(mBo - RBI)

2 . mdB - kdBl
O

 

m(-mBo) - k(mBo - kBl)

2
-m Bo - kmBo +.k2Bl. If k is less than m, the lepe is

everywhere concave down. Bi. can have a critical value if m equals

k or is less than k. In general, the second derivative of B1 will

be negative, as k is negative in (l).

(lc*). Second derivative of B2:

dZB

2 = d(kBl) = k(dBl) = mazeo - kmBo - kZBl).

 

da2
In contradistinction to (l), k is positive in (2). Hence, the
second derivative of B2 will in general be positive.

On the basis of the deductive statements we have made about
the system of differential equations, we can determine the general
shapes of the B curves. This is the main advantage to using such a
model. The other'advantage has been that we have been able to deduce

a proposition about B from our original two propositions.

l

1.0

 

Bo
Proportion
of B .5
Bl
. 0 BL \.-
w High'

 

Level of Involvement (a2)

27

B. Data and Testing of Proposition;

We were able to test these propositions on the basis of a
questionnaire administered to 14-47 students in an undergraduate
sociology class at Michigan State University, in April 1963.
l+09 questionnaires were coded to test this model. Each respondent
was asked about five fererents, so the number of possible responses
was 5 x l+09 = 201+5. The actual number of codeable and applicable
responses was 1573.

The sample iszi‘a fairly representative cross-section of the
entire student body, though this is not the universe to which we
eventually hopecto generalize.

The relevant items were scaled with integers. Scaling is in
general dependent on the choice of mathematical model which states
differences» in response as a function of location on an underlying
dimension of scaleable data. The model consists of (1) level of
measurement stated (nominal, rank, etc.),. (2) the graph of the item
as a function of location on the underlying continuium, (3) the
judgment of the respondents, and (4) the mathematical pattern, by
which it is determined whether. the continuium is scaleable.

We defined the referent categories as follows for the
respondents as follows:

"You have-described your occupational goal in proceeding

questions. Now we would like to find out some things

about your relation to the peeple who may help you attain

your goal.

Consider. all the possibilities you have for getting help
to: attain your job goal:

28

(a) PARENT(S) or guardian

-(b) Your.)PEEIB (Those close to your own age that you interact
with

(c) Your SIBLING(S): Brother(s) and/or Sister(s), if applicable

(d) SOCIETY: Peeple in other relations to you, such as teaCher~
or adult friend, or significant peeple in society generally.
This is our catch-all category, so place everything that
won't go in other categories here.

(e) SPOUSE, if applicable."

"Desire to help" is Considered present if the respondent answers

"Strongly Agree" or "Agree" to the following question:

(Item 29) "Indicate for the following groups the extent, to which“
they agree or disagree with your choice of occupational goal."

Strongly Agree Agree Indifferent Disagree Strongly Disagge
Referent. + 2 + 1 O - l - 2

PATENTS
PEEFB
SIBLINGS
SOCIETY
SPOUSE‘
"Ability to help" is considered present if the respondent
answers "Great Ability" or "Some Ability" to the following question:

(Item 31) "To what degree do each of these groups have the ability
or power to help you attain your occupational goal?"

Great Ability Some Abillitl Little Ability No Ability
Referent 3 2 1 0
PARENTS
PEEIE
SIBLINGS
SOCIETY
SPOUSE

29

Four questions under Item 27 were constructed as: indices of
involvement with (orientation to) the five categories. of referents.
They dealt with (1) frequency of discussing goal, (2) frequency of
respondent going to the referent for advice or assistance, (3) how
often the referent offered advice or assistance, and (1+) how much
actual contribution has the referent made toward attaining the goal.

Item 28 is asked immediately after these four indices of
involvement: this question ordering made Item 28 meaningful to the
respondent. The question is asked as follows:

((Item 28) "Indicate, in general, the degree to which you interact

or aminvolved with: the following groups with respect to attaining
your occupational goal."

YOUR POSSIBLE GIDUPS O INVOL .

PARENTS 5 1+ 3 2 1 O
PEEIB 5 1+ 3 2 1 O
SIBLINGS 5 1+ 3 2 l 0
SOCIETY 5 ll» 3 2 1 O
SPOUSE 5 1+ 3 2 1 0

Very Highly Highly Moderately Little Very Little Not Involved
Igvolved ' Involved Involved ’ Involved . Involved ' At A11
0

5 T— 3 2 1

If a respondent'failed to answer gm; of Items 28, 29, or 31
for a referent, g his responses about that referent were thrown.
Any answers that were multiple, §.g., circle both ll- and 5 for parents
on Item 28, were also considered non-applicable. The total of l+72
non-applicable responses were distributed as follows: 11 for parents,

8 for peers, 77 for siblings, 25 for society, and 351 for spouse.
The results of the coded questionnaire are presented in Table 1.

30

Table 1.--Responses for Five Referents to Dichotomized Levels of
Desire and Ability to Help Attain an Occupational Goal, and to

Level of Involvement

 

 

 

 

Level of Referents B - (Desire to Help,gAbility1to Help)
Involvement (0, 0) (0, 1) (l, 0) (1, 1)
a D No Desire No Desire Desire Desire TOTALS
2 No Ability Ability No Ability Ability

Parents 10 3 l 21
Not Peers 5 55.14 1 6.6% 8 35.5% 1 2.5% 15 100.0%
Involved Siblings 41 (N : 67)3 (N = 8) 24 (N = 1+3) 0 ( =3) 68 (N = 1212
At All Society 10 1 4 1 16

Spouse 1 .0 O O 1

Parents 9 5 19 1 34
Very Peers 16 34.0% 3 13.1% 27 50.8% 1 1.6% 47 99.9%
Little Siblings 30 (N = 63) 10 (N - 21000 (N - 93) 0 (1:3) 80 (N = 1831
Involved Society 8 5 5 1 19

Spouse 0 1 2 0 3

Parents 4 7 l3 16 40
Little Peers 15 20.11% 3 8.8% 30 38.0% 22 32.8% 70 100.0%
Involved Siblings 22 (N a 51) 2 (N =22) 32 (N = 95) 13 (N=82)69 (N a 250:

Society 10 10 20 30 70

Spouse 0 O O l 1

Parents 7 4 17 82 110
Moderately Peers 22 8.0% 15 11.3% 46 21.6% 71 59.1%150 100.0%
Involved Siblings i"! (N = 37) 22 (N a 52) 24 (N s 100>33(N=273)83 (N - 4622

Society’ 4 11 12 82 109

Spouse 0 0 l 5 6

Parents 2 1.6% 5 6.6% 5 13.9% 75 77.9% 87 100.0%
Highly Peers 2 (N = 5) 5 (N: 20)16 (N :42) 60(N=236)83 (N = 3032
Involved Siblings O O 8 12 20

Society 1 10 9 84 101+

Spouse 0 0 4 5 9

Parents 0 2 6 98 106
Very Peers 0 0 0% 1 3.1% 5 5.1% 26 91.8% 32 100.0%
Highly Siblings 0 (N = 0) o (N : 8) 1 (N'13) ll(N=233)12 (N = 2541
Involved Society 0 5 1 60 66

Spouse 0 0 0 38 38

Parents 32 26 67 273

Peers 60 28 132 181
TOTALS Siblings 97 (N = 233)L12 (N=l34)129 (11-386) 69 (N=83o)

Society' 33 37 51 258

Spouse 1 l 7 49

 

31

Table 1 indicates an unusual distribution for two referents,
siblings and spouses.

Siblings are not highly involved with the respondents. 217
siblings are categorized having little, very little, or no involve-
ment, whereas only 115 are moderately, highly, or'very highly
involved. 'The nonpsibling distribution here is 345 at the lowest
three levels, and 904 at the highest three.

Siblings also fall disproportionately into the (desire, no
ability) case, particularly at the three low levels of involvement.

The definition given for sibling accounts for a lot of this:
no criteria are given for ordinal position, sex, or age. For
siblings a lgt_younger than the respondent, it is highly improbable
that they have developed much potential to help the referent attain
an occupational goal, even though they might approve of the goal,
and desire to help. The category is meaningless in terms of the
respondent's three year old sister. But in terms of the definition
for sibling, she should have been included. Of the non-applicable
and uncodeable responses excluding spouses, 64% were of the sibling
category. If this was entirely'due to not having a sib, then 77 of
the 409 respondents did not have a sib. This is highly improbable.
The conclusion is that respondents answered non-applicable for’much
younger sibs, even though the letter of the instructions indicated
they should have been included. This conclusion is supported by
remarks written in the margin by several respondents.

Married respondents are highly involved with their spouses,
and perceive that they have a lot of desire and ability to help..

32

It is not the distribution of spouses in Table 1 that is misleading;
it is the sample itself: spouses are overerepresented relative to a
typical adolescent population. The sample included a lot of non-
adolescents, and these respondents are far more apt to be married.
Spouses were included to make the reference facet exhaustive.
Ehaustiveness was attained at the price of a distorted sample.

There is also a weakness in the scaling for level of involve—
ment revealed by the data in Table 1. There appears to be only a
gigght.psychological distinction between being "Highly Involved"
and "very Highly Involved" for (1, 1) relative to the;psychological
distinction between "Mbderately Involved" and "Highly Involved."
The percentage increase for the former is 19%, but only 14% for the
latter. There is apparently a greater psychological space between
moderate and high than between high and very high, though the
percentage distributions at each involvement level Show that very
high is more extreme than high. The difficulty is that bgt§_high
and very high are extreme responses: both are highly correlated with
(l, 1).

Equation (1c*) predicts that 82 will accelerate as involvement

increases. But, contrary to prediction, B ’has an inflection point,

2
and then decellerates. The deeelleration of B2 in turn creates an
unusual Bl curve. Equation (lb*) predicts D1 will, in general,
decellerate as involvement increases, though this is not necessary.
Bl has an inflection point, and then accelerates.

The psychological mid-point of the involvement scale seems to be

"Little Involvement," But the scale has an even number of alternatives,

33

and therefore has no integral midpoint. On the basis of the psycho-
logical reality of the scale, the use of two extreme positive
responses over-represents that end of the continuium.

0n the basis of these weaknesses-in the scale, there is adequate
justification for grouping the responses "Highly Involved" and "Very
Highly Invoved," I

To test our propositions, a probability distribution will replace
the percentage distribution in Table 1. A five-point involvement scale
will be used. And (0, 1) and (l, 0) will be combined for the 131 values.
Table 2 and corresponding Figure 2 will test our propositions for all
five referents, and Table 3 and corresponding Figure 3 will re-test,

with siblings and spouses deleted.

Table 2.--Leve1 of Involvement with Parents, Peers, Siblings, Society,
and Spouse, by Dichotomized Response to Desire And Ability to Help
Attain an Occupational Goal"I

 

Level of Involvement B I (Desire to Help, Ability to Help)
With Five Referents (O, 1) and TOTALS
<mo>=i<Lo>=g n.n=B

 

 

 

2

—-.=:

Not Involved at All .554 .421 .025 1.000
Very Little Involved .3u4 ' .639 .016 .999
Little Involved .204 .468 .328 1.000
Moderately Involved .080 .329 .591 1.000
Highly Involved or I '

Very Highly Involved . 009 .149 .842 1.000

I"Source: Table 1.

.-

"V

31,.
Figure 2. Level of Involvement with Parents, Peers, Siblings, Society,
and Spouse, by Dichotomiged Response to Desire and Ability to Help
Attain an Occupational 0081*

1.0
.9
.8 " 32“
Probability .7
of Involve- .6 *
ment for B .5.
.11:- .
03 . "' *
.2 o
.1
00" «- 1;. B (e
0 1 2 3 l; °
(low) Leml of Involvement (high)

.
*
m
H
A
'I'
W

Source: Table 2.

Bo behaves as predicted. The first derivative is negative;
and the second, positive. ~
B1 and B 2 behave gy_i_t_e_ well, but Bl almost has an inflection
point at "Little Involved"-at the psychological midpoint, and B2
almost has an inflection point at "Little Involved.;~" In each case,
there is almost no concavity at involvement levels 2, 3, and 4: the
functions become nearly linear.

For BZ’ the N at 1 and 2 is 3 responses. So the slight
decline from 0 to 1 can be ignored. With a larger sample, this
would probably not occur.

Another look at Table 1 reveals a highly plausable explanation
for the deviance in Bl and B2 at the higher involvement levels.

The answer is to be found in the El category. Bl is composed of
two cases, (0, 1) and (l, 0). We have earlier referred to these as the

"ambiguous" cases. The connotation is premature.

35

In Table 1, the distribution of (1, 0) a (desire, no ability)
is .355, .508, .380, .216, .099 for five values of involvement, from
lowest to highest. This is a well-behaved concave-down curve.
Desire but no ability is perhaps logically ambiguous, but it is
clearly not psychologically ambiguous.

In Table l, the distribution of (O, 1) n (no desire, ability)
is .066, .131, .088, .113, .050 for five values of involvement, from
lowest to highest. This is a poorlyebehaved curve. It slumps at
the psyChological midpoint, "Little Involvedz" both adjacent cases
are larger. No desire but ability is perhaps logically;ambiguous,
and it is also psychologically ambiguous. Here is the case where the
referent is perceived as somehow not wanting to g§2.his potential
to help the respondent attain his goal. The responses are polarized
at the extremes, either toward psychological rejection and withdrawl
from involvement, or to a level of high involvement. we have
implicitly based our'two propositions on perceptual optimization and
rationality;-but where the stimulus facet is perceived as ambivalent,
this rationality is lacking,.and the respondent is confused about
how involved he ought to be, to attain his goal.

we can now return to the problem of siblings (the inadequate
question) and spouses (the distorted sample), and ask what effect
they have had on the shapes of the B curves. ‘We will re—test our
propositions, with siblings and spouses deleted. The results are

presented in Table 3 and Figure 3.

36

Table 3.--Leve1 of Involvement with Parents, Peers, and Society, by

Dichotomized Response to Desire and Ability to Help Attain an
Occupational Goal: Probability Distribution’"

 

Lovel of Involvement

B a (Desire to Help, Ability to Help)

 

 

 

With Three Referents (6,1) and TOTALS
- “- (0, 0) . Bo (1, 0): Bl (1,1) = 82

Not Involved at All .481 .462 .058 1.001
Very Little IUVOlved 0367 .717 .033 0999
Little Involved .161 .461 .378 1.000
Moderately Involved .088 .281 .631 1.000
Highly Involved or

Very Highly Involved .010 .147 .843 1.000

 

I"Source: Table 1

Figure 3. Level of Involvement with Parents, Peers, and Society, by

Dichotomized Response to Desire and Ability to Help Attain an

Occupational Goal"

1.0

Probability .7
of Involve- .6
ment for B .5

( low)

Source: Table 3.

‘l

Lovel of Involvement

2

*B (*)
.BiC.)

(high)

37

Deleting siblings and spouses has had almost no effect on
the structure of the rhlationship between B and a2. Relative to
the distribution of involvement for‘the other three referents,
siblings are undereinvolved and spouses are overuinvolved. The
total effect of throwing both out is that they tend to cancel

each other. In Section D, we will consider all five referents.

C. A Methodological Evaluation of the Ebdel
‘We have assumed there exists a differentiable function
a2 3 f(I), I I 0, 1, 2, 3, 4,
where a2 is the level of involvement and I represents the integral
values for involvement we obtained from responses to Item 28.

There are three functional relationships between a2 and
B = (Bo, B1, 132),. such that for each of the five values of I there
corresponds one and only one value for each case of B. This is
not a continuous function. And we must be careful about assuming
continuity for the following two reasons:

1. ‘We initially asked each respondent to make six discrim-
inations for level of involvement with each referent. Research on
scaling has indicated that peOple cannot meaningfully amke much
more than seven discriminations on scaled items. In this case,
it was necessary to reduce the number of discriminations from six
to five in the analysis of the data. But if our‘differential
equations are meaningful, we must have continuous functions, for
all differentiable functions must be continuous. The continuity

criterion means we must have infinite discriminatory power on the

38

part of the respondents for involvement. This assumption is not
justified.

2. Any sample we draw is finite. As the ratio of discrim-
inations over a sample of size N grows larger, the three-
dimensional histogram (the balues in both tables and figures) will
fluctuate irregularly, and then collapse.. And if this ratio
becomes improper, there must necessarily be discontinuities.

Such an improper ratio is exactly what happens in the limiting
process that defines differentiation.

In spite of these logical and methodological shortcomings,
differential equations are a useful tool in generalizing about,
and making deductions about, a set of interrelated propositional
statements. .

It is necessary to be extremely carefulezabout making unreal
assumptions. The _s_3bstantive assumptions have been presented in
Section B, Chapter 1. We will attempt throughout to keep our
gthodologgcal assumptions explicit, minimal, and realistic.

With this in‘ mind, we will next re-work our propositions in
terms of difference equations,. limiting ourselves entirely to the
integral values of involvement. This can be done with no loss of

generality.

D. A Difference Equations Model
By constructing difference equations analogous to our
differential equations, we can easily solve for m and k. This

will also enable us to determine the "goodness of fit" for the B

39

curves Bo and B2. This will lead to the theoretical distribution
of the B1 curve.

Deviation in B2 has already been established, which is explained
by the psychological reality of 81a . (0, 1). So the model will fit
B2 worse than B1. In this sense, the model is inadequate, but it
is concomitantly adequate in that it will make possible a rigorous
measurement of the degree of inadaquacy. Also, other kinds of
samples can be used in future research to determine the generality
of the model. Samples from less advantaged groups and strata than
college students may ShOW'leSS over-involvement in B2.

Define Pn as the probability of an adolescent in the B0 case
beingvinvolved at level n, where n I 0, 1, 2, 3, 4. ‘we know from
our'data, presented in Table 2, that PC I .554.

Proposition (1) can be reworded to mean that the probability
of B0 being involved decreases as n increases,‘i,g., if'APn : (Pn+1> - Pn
is negative, we know the direction of change. we also need to know the
magnitude of change. In our’differential.equation model, we have
expressed the relation that the probability of involvement is a fraction,
call it m, of the next highest level of involvement:

Pl I mPo,
P2 I mPl,

Pn=mPn_1,n'0. 1! 20 3! 1“"

In general, Pn + 1 = mPn. Subtracting Pn from both Sides,
we obtain Pn * 1 - Pn = mPn - Pn' Therefore,

4O
APn = (l - m)(-Pn).
In words, this means the actual decrease in probability of involve-
ment at increasing levels of n, APn, is a fixed fraction (1 - m) of
the W possible decrease, which is (IPn).
Finally Bl = mBC-j,
82 = mBl.= szo, and in general,
(4) an I m'nBo, where n = 0, l, 2, 3,, 4,1
Formula (4) emables us to test our proposition. (.554) = (.344)
which implys m I .6. The theoretical and empirical values and percentage
distributions are presented in Table 4 and Figure 4. Goodness of fit is

measured by X2, which is computed from the percentage distributions.

Table 4.--mepirica1 and Theoretical Distributions of Level of Involve—
ment with Five Referents Perceived as Having No Desire and No Ability
to Help Attain an Occupational 0031*

 

Leml of Involvement Expirical Theoretical
with Five Referents Distributions Distributions

Prob. (%) Prob. (fl)

 

Not Involved at All .555 (46.5) .554 (10.6)
Very Little Involved .344 (28.9). .332 (26.0)
Little Involved .204 (17.1) .199 (15.6)
Moderately Involved .080 ( 6.7) .119 ( 9.3)

Highly Involved or .009 ( .8) .071 (5,6)
Uery Highly Involved

 

Total (100.0) (100.1)
2
x = 5.05

 

'Source: Table 2.

 

2Samuel Goldberg, Introduction to Difference Equations (New York
London::John Wiley & Sons, Inc., 1961, pp. 2-4.

41

Figure 4. Level of Involvement with Five Referents Perceived as
Having No Desire and No Ability to Help Attain an Occupational Goal*

90.0 ($) Theoretical (*)
<90 Emmi-real <.> '

Percentage
Involved
for BO

0

.‘

I—‘NW O‘s‘Q
00003000

_I. ,L,

 

0 1. ~ 2
(low)~ Level of Involvement (high)

I"Source: Table 4.

The fit is good. The theoretical curve, however, does not come
as close to vanishing at 4>as the empirical value. Our first hypothesis
is confirmed, and the mathematic description--Formu1a 4u-is adaquate to
obtain a highly significant I? of 5.05.

A.difference equation.can also be obtained for E2 I (Desire,.
Ability) by much the same procedure used to obtain Formula 4 for Bo.
But it is necessary to change the argument to a certain extent, for at

the two lowest levels of involvement for B the sample N is 3. And the~

2:
percentage involved at the second level (1) is lower than the percentage
for the second level (2). This problem of sample size is obviated by an
obverse argument from.highest to lowest.

Our second hypothesis is that the probability of involvement
increases as involvement increases for B2° we can equivalently say
that the probability of B2 decreases as involvement decreases.

were specifically, we hypothesise that the probability of

involvement is a fraction, k,.of the probability of the next highest

42

level of involvement .

(5) Pn-l = mPno

P1 = mPZ,
P0 = mPl, where n = 4, 3, 2, 1, O.
subtraction of PH from both sides of Formula (5) yeilds

Pn--l

-APn = (1 - n)Pn.

In words, this means the decrease in probability of involvement at

-Pn=mPn-Pn, and

decreasing levels of involvement, - APn, is a fixed fraction (1 - m)
of the maximum possible decrease, which is Pn."

In. general, the empirical curve is not very concave up,. as
predicted; in particular, involvement is deviantly high at "Moderately
Involved," having a probability value of .591. The solution for k is
.7: for this 1:, the curve does not come close to vanishing, as does
the empirical curve. The theoretical percentage distributions have
the valuesu7.3%, 26.1%, 14.3%, 7.9%, and 4.3% from.highest involvement
to lowest..- This is not an adaquate description of B2.

We can, however, determine the k that has the Egg}; goodness of
fit, as measured by chi-square. In every case, it was necessary to

group the two lowest percentages at the two lowest levels of involve-

ment, so that 12 will be applicable. We obtain a curvilinear distribution:

Table 5.--Chi—square Values'for Different choices of K in B2

 

 

Values of k3? .5 0514' I55 I56 06 I7
Chipsquare: 10.82 10 .71 10.09 11.22 15.07 22.33

 

43

As k decreases, the theoretical curve is lowered, and chi-square
decreases. The increase in error from underestimating the higher
level values for involvement is more than offset by the decrease in
error from overestimating the lower level Values: for involvement. A
minimum 1:2 of 10.09 is reached at k = .55. At smaller k,.the error of
underestimation of the higher levels of involvement becomes greater
that the reduction in error of, overestimation of the lower levels.

we will use k = .55, as it produces the best goodness of fit.

The results are presented in Table 6 and Figure 5.

Table 6.-Empirica1 and Theoretical Distributions of Level of Involve-
ment with Five Referents Perceived as Having Both Desire and Ability
to Help Attain an Occupational Goal"

 

Level of Involvement Erupirical Theoretical
with Five Referents Distributions Distributions

Prob. (%) Prob. (2)

A

Not Involved at All .025 ( 1.4) .077 (i 4.3)
Very Little Involved .016 ( .9) .140 (‘ 7.9)
Little Involved .328 (18.2) .255 (14.3)
Moderately Involved .591 (32.8) .463 (26.1)

Highly Involved or .842 (46.7) .842 (47.3)
Very Highly Involved

 

Total (100.0) (1 99.9)
:2 .-.- 10.09

 

*Source: Table 2.

44

Figure 5. Level of Involvement with Five Referents Perceived as
Having Both Desire and Ability to Help Attain an Occupational Goal*

100.0 (fi) Theoretical (*)
90.0 _
80.0 (%> Empirical (.)
70.0
Percentage 60.0
Involved 50.0 *.
for B 40.0 .
2 30.0
20.0 J
10.0- ‘ ‘ .
0.” . .
0 If ' 2 3 :54
(low) Level of Involvement (high)

*9.

*Source: Table 6.

The relation between B2 and involvement is more nearly linear
thhn curvilinear. The linear correlation r is .96. A.larger sample
might raise the percentage of "Very Little Involved" somewhat closer
to the predicted level, and 'improve" the model. This small sample
N at the loweinvelvement end of the continuium makes it difficult to
generalize about the universe of addlescents. Although the goodness
of fit of the gage; may be in doubt, the hypothesis is validated.

Using the best fit k I .55 for Bz, and m.= .6 for Bo’ we are
able to deduce the theoretical values for El = 100.0 - £80 - %B',

n = l, 2, 3, 1+, 5. The results are presented in Table 7 and Figure 6.

45

Table 7.-Empirical and Theoretical.Distributions of Level of
Involvement with Five Referents Perceived as Having Either'Desire
But No Ability or*No Desire but Ability to Help Attain an
Occupational Goal"I

 

 

 

Level of Involvement Empirical Theoretical
With Five Referents Distributions Distributions"
Prob. (%) Prob. C?)
Not Involved at All .421 (21.0) .369 (18.9).
Very Little Involved .637 (31.8) .528 (27.1)
Little Involved .468 (23.4) .546 (28.0)
moderately Involved .329 (16.4) .418 (21.0)
Highly Involved or .149 ( 7.4) .087 ( 4.5)
very Highly Involved
Total (100.0) ( 99.9)
22 = 1.98

 

*Source: Tables 2, h, and 6.

Figure 6. Level of Involvement with Five Referents Perceived as
Having Either'Desire but No Ability or No Desire but Ability'to
Help Attain an Occupational Goal*

Percentage
Involved

for E1

100.0

(%),Theoretical (*)

(fl) Empirical (.)

0*

 

(low)

l’Source: Table 7.

Level

2

of Involvement

4‘"

(high)

46

Chi—square I % (empirical - theoretical )2/ theoretical
n . 0 n n n

has two crucial weaknesses as a statistical description of goodness
of fit:

1. The predicted values are a mean average of 3.69% off the
empirical values. X2 divides the squared difference by the
theoretical values. Holding the difference (empirical .. theoretical)
constant, we see that the contribution to X2 at a given 11 is largest
where the theoretical value:is largest. For example, (571’: - 1%)2/ 5% :
.80,.but (50% - 48,4)2/ 50% = .014. In chi-square, equal deviations are
weighted more heavily for theoretical values of smaller magnitude than
of larger magnitude.

2. Our first objection would be substantively irrelevant if
it could be expected that the differences (empirical - theoretical)
be smaller. for smaller theoretical values. But if we can determine
any criterion for size of difference, it is sample N.

Sample N. is extremely small (N = 5) for the Bo case at n a 5.
The 12 component at this n is 3.76, which is 71+.57o of I2 I 5.05.

Sample N is also extremly small (N 8 3) for the B 2 case at

n=0andn=l. Thex2

is 80.9% of 12 = 10.09.

components of these two ns are 8.16, which

In summary, it is not the case that we can expect more exact
predictions where the values are small, for by definition of the
B-axis (percentage of Bi)" the sample size is small when E is small.
On the contrary, the small N necessarily associated with small B

will create more chance variation, and less exact predictions.

47

To compensate for these two difficulties with 12, another
simple statistic can be imployed--the average of the sum of the
differences: 4
2(Dii‘f.) = 11:0 l(% theoretical“. 7% empirical“), .

n.
n = O, l, 2, 3, 4. Comparisons ofx2 and iCDiff.) are presented in

the following table:

Tahle 8.--Comparisons of Chi-square and iKDiff.) for All Cases of
Perceived Desire and Ability to Help Attain an Occupational Goal

 

(Desire, Ability) measure of Goodness of Fit

 

l = yes __

0 I no Chi-square .X(Diff.)3
W
Bo - (0. 0) 5.05 2.94%
Bl . (0. 1) and (1. o) 1.98 3.86%
132 s (1, 1), 10.09 4.28%
Average 5-68 3-59%-

 

Our two objections are confirmed. Only B1, with an i‘wiff.)
slightly above average, has a very small 12. The explanation is
that Bi is the only B case that has a large sample size at every

level of involvement.

CHAPTER.3

A SOCIOLOGICAL mommow OF STOUFFER'S mmoar OF
nxrssvmmm OPPORTUNITIES

A. Propositional Schema

The movement of people in space is a fundamental aspect of
sociological inquiry. Since the classic statement by Ravenstein,l
research has overwhelmingly shown a close relationship between
mobility and distance. Ravenstein's basic proposition that "Most
people go a short distance; few people go a long distance," has been
validated at a high level.

A standard approach to the study of this two-variable relation-
ship is the “push-pull" schema: what factors pust migrants from their
place of origin, and what pulls them to their destination. This intro-
a third variable, which is independent, and is used in demographich
research and population studies to explain the relationship between
mobility and distance, which are dependent. This third variable is the
opportunities structure: at the place of destination, and at the place
of origin.

Stouffer found empirical and theoretical inadequacies in the

push-pull explanations, and developed a refinement Of the opportlmities

1E. G. Ravenstein, "The Laws of Migration, " Journal of the Royal
Statistical Society, 48, (June, 1885), pp. 167-235: 52, (June, 1889),
Pp.2§+1-305, in Samuel A. Stouffer, Social Rasearch to Test Ideas
(Glencoe, Illinois: The Free Press, 1962). P. 69.

48

49

variable.

He said there is no necessagy relationship between mobility

and distance, and introduces the concept of intervening opportunities.

He hypothesised that

|‘...the number of persons going a given distance is directly
proportional to the number of Opportunities at that distance
and inversely preportional to the number of intervening

opportunities . '2

Stouffer was able to relate empirical data very closely to

theoretical expectations based on this hypothesis.

The proposition

seems to fundamentally generic to a demographic approach, that it merits

an attempt to generalize it to sociological analysis.

If we can develop

a proposition that is sociolo ical, Stouffer's proposition will potentially

become a special case of a general sociological law.

Analogies between demographic migration and sociological orien-

tation can be organized in the following paradigm:

 

DEMOGRAPHIC MIGRATION

1. Migration to a physical point:
movement over physical space.

2. Perceived physical distance
properties of the point of destination,
with accessibility defined in terms of
"physical friction of space“ and
measured by criteria such as time and
distance.

3. Perceived Opportunities to attain
goals at potential points of destina-
tion.

SOCIOLOGIGAL ORIENTATION

1. Orientation (involvement)
with a referal point: movement
over social space.

2. Perceived social distance
properties of the referal point

of destination, with accessibility
defineddin terms of "social fric-
tion of space" and measured by
criteria such as desire and interest.

3. Perceived Opportunities to

attain goals at potential referal
points, i,g., the perceived

ability to help of the referal points.

 

'1

* 1Stouffer, Ibid, p. 71. This article was first published as

"Intervening Opportunities: A Theory Relating Mbbility and Distance,"
American §pciological Review, 5, (December, 1940), pp. 845-867.

50

Such a generalization of Stouffer's basic proposition allows it
to be reformulated entirely within a context of the elements in the
Facet Paradigm developed in Chapter 1.

Stouffer found the usual push-pull analysis relating mobility
to Opportunities to be an oversimplification. Our hypothesis relating
orientation to "...ability to provide opportunities..." (page 15) to be
an oversimplification of the same order. A more subtle, and more general,
kind of hypothesis is found in the theory of intervening opportunities.
Stouffer has demonstrated this in a demographic context; in this chapter,
an att mpt will be made to demonstrate that the preposition is extensile
to a sociological orientation.

Proposition: The preportion of adolescents oriented to referal

 

points a certain social distance away is directly proportional to the
perceived ability to provide opportunities of all referal points at
this social distance, and inversely proportional to the abilities of
all intervening referal points at all closer social distances.

This can be represented by a difference equation:

Ararat
lg XE,where

Preportion of adolescents oriented to referal points a social

5|?

distance S away (actually, a band, with the closer edge of 8 being
5 - -§- 43 away and the far edge S + %AS). In the model, this is
azci forAP, and b1 forAS.

fi}%.: Opportunities to provide help of all referal points at a social

distance band S away. In the model, this is a1b2°

51

IS = 1 3:, s 2 l, 2, 3,...,,n. The intervening Opportunities, 1.2.,
the sum of the ability to provide opportunities for all.
referal points.closerrthan 5. Note that at the closest
social distance, 1, there are 22 intervening Opportunities.

K = a constant of proportionality.

S = social distance. In the model, thiS'lS alb1°

B. Operationalization of Variables

1. Social Distance
Thisvariable is concerned with the (orientation) element of the
referal point's behavioral properties, as. (perceived) by the adolescent.
Guttman has demonstrated that all scaleable attitudes have many
(technically infinitely many) principal components.1 The psychological.
interpretation given to the first component is the direction of the
attitude.2 This has been measured by Item 28 in the questionnaire
(on page 28) which was designed to determine if referal points are
perceived as favorable,,neutral, or unfavorable to the adolescents choice

of occupational goal. This is the first index of social distance to be

 

lPrincipal components '...are also known in mathematics and physics
by various other names such as principal axes, latent vectors, eigen.
vectors, or eigenfunctions,“ in Guttman, "The Principal Components of
Scalable Attitudes, " in Paul F. Lazarsfeld, Mathematical Thinking in the
Social Sciences (Glencoe, Illinois: The Free Press, 1954), pp. 428 ff;
see also pp. 224-226. A formal mathematical treatment of principal
components is presented in Guttman, "The Principal Components of Scale
Analysis,' in Samuel A. Stouffer et. al.,.Measurement and Prediction Vbl.
4: :fiStudies in Social Psychology in _World War Ilerinceton, New Jersey:
Princeton University Press, 19 50), pp. 313-361.

 

6 2G'uttman, "The Principal Components of Scalable Attitudes," pp. 219-
22 .

52

used in So

The second principal component of the (intention) element is

the intensity component .

'If (a) scalable attitude has a meaningful zero-point
(point of indifference), then as peOple have rank farther and
farther to the right of it, they should become more and more
positive, and hence more and more intense. Similarly, as ranks
get farther and farther to the left of the zero-point, they
should become more and more negative, and hence also indicate
more intensity. Intensity accordingly should have a U. or J-
shaped relation. with the underlying rank orders *3

The following question was designed to measure the second
principal component of the (intention) element:

(Item 30) 'HOw much interest do these groups take in your attaining
your occupational goal?

 

 

 

Very Intensely ~ . . No
Interested ' Some Interest Little Interest Interest
3 2 1 0
Referent-
PARENTS
PEEFB
SIBIINGS
SOCIETY
SPOUSE

We hypothesise that perceived "Agreement“ with the adolescents
choice of goal (in a context of helping him) is the first principal
component of the scalable (intentions) attitudeuwhich is direction, and
that perceived 'Intemst' in helping the adolescent attain his goal is

the second principal component-which is intensity. The test of this

.......

 

3cmttman, Ibid., p. 229.

53

hypothesis is simply to determine if the relation is curvilinear, or
ummw. '

Table l.--Percentage "Very Intensely'Interesteda or "Some Interest" at
Each Rank of.'Agreement" with Choice of Occupational Goal, for Five
Referents -

 

Percentage at- Rank.of FAgreement' with Choice of Coal

'Very Insensely

Interested“ or Strongly' Agree Indifferent Disagree Strongly
“Some Interest. Agree Disagree

 

97.3% A 88.6% 44.9% 60.0% 42.8%

 

Total 686 740 336 40 7

 

The hypothesis is in general validated. The deviant value is
42.8%, but N = 7, which is extremly smalL-too small to reject the
hypOthesis.

Social distance will be defined on the basis of the first two
principal components of (intentions) as follows:

S I Agreement ? Interest.

Each'component of S will be equidistantly scaled from 0 (the point of
minimum distance) to 3 (the point of maximum.distance).' The range of
social distance is thus scaled from O to 6. There must be at least one
tie on any scale: where agreement is nearest and interest farthest, and
where agreement is farthest and interest nearest. Tied combinations
will be added together as a single social distance. Each distance will
be given a rank number, 1, 2,...,n, but these ranks are not necessarily

equidistant..

| l '1' 1' | IIII||.||1‘ 1111

54

The UIShaped relation, and the general tendency toward
perceived high agreement and high intensity presents a sampling
problem for S. As the respondents are unevenly distributed, there
are several cases with few entrys. The original design for S was a
twentyzpoint scale (nineteen unique points). This had to be rejected
because eight of the nineteen distances had less than ten cases, which
weighted the individual case too heavily. This problem was overcome
by dichotomizing the four-point scale for “Interest,“ as in Table l.

The resultant matrix for S is presented in the following table:

Table 2.--Social Distance, 8, Soical Distance Rank (1.9), and Number
of Cases, N, For Five Referal Points-

 

 

Level of Level of Agreement
Interest
Strongly Agree Indifferent Disagree Strongly
Agree Disagree
DISTANCE . . '_0 .75 1.50 2.25 3.00
Very In- :
tense 0 0 (l) .75 (2) 1.50 (3) 2.25 (4) 3.00 (5)
or Some N I 668 N = 656 N I 151 N I 24 N = 3
Little
or None 3 3.00 (5). 3.75 (6) 4.50 (7) 5.75 (8) 6.00 (9)
N = 18 N = 84 N I 185 N = 16 N = 4

 

The first measure of social distance, devised by Bogardusn in 1925,

was a Racial Distance Scale. A set of seven items was used to measure

 

uEmory S. Bogardus, @Measuring Social Distance,I Journal of Applied
W. 9. (1925). pp. 299-308.

55

degrees of social aggeptibility, or social distance. The Bogordus
Scale ordinally measures the direction and intensity in a way that
the components cannot be separated out. Krech and Crutchfield
cogently criticise this scale: 'Social distance is a complex qualit ,
related in the most intimate way to the ego standard of the indiv.
idual, his concepts of prestige in the...group..." and "...social
distance from an object may in some cases be markedly independent of
the general affectivity of the object.'5 This is an attempt to iso.
late1the components, as they are in Table 2. Since Kramer's6
elaboration in 1949, the concept is applicable to other individuals

in various personal and social relationships.

2. Opportunitieg

At each social distance, an index of the perceived ability
to provide Opportunities to help the adoleSCent attain an occupational
end can.be established. This is based on Item 31 (page 28)::

‘A1 a 3(probability of "great ability") + 2(probability of

AS I‘some ability“) + 1(probability Of "little ability").

XS can be deduced from this.

30 Orientation
There is now enough information to predict the distribution; of

adolescents involved overs. An independent measure of involvement can

 

5David Krech and Richard s. Crutchfield, gpeory and Problems of
Sgcial Psychology (New York: l-IcGraw-Hill Book Co., 1948), pp. 222.

63. N. Kramer, "Dimensions of Prejudice," Journal of Psychology,
27. (1949). PP- 389-451.

56

be constructed, to test the proposition by giving an I‘actual'
distribution of involvement over S. This is based on Item 28
(page 29).
Real AP/ As = 2(probability of “very highly involved") .
1(probability of "highly involved").

The index of.oportunities measures the whole range of
perceived ability; Orientation will be given a more restricted
interpretation, as only the two highest levels are considered.

Section C will test the proposition, for all referents, and

for each referent separately.

C.

57

Data and Test of Proppsition

Table 3.-—Perceived Ability to Provide Opportunities to Help Attain an
Occupational Goal, Level of Involvement, and Social Distance, for Five
Referal Points: (N) and Probability Distribution

 

 

 

 

- (least) Social Distance (greatest)
OPPORP-
UNITIES 1 2 3 4 5 6 7 8 ». 9
Greate- (271) (146) (13) ( 4) ( 3) ( 4) (1o) ( 3) ( 0).
Ability .406 .222 .099 .174 .167 .048 .054 .188 .000
Some, (241) (289) (47) ( 6) ( 6) (19) (33) ( 7) ( 0)
Ability .361 .440 .311 .261 .333 .226 .178 .438 .000
Little (107) (174) ('49) ( 8) (' 4) (35) ('55) ( 2) ( 3)
Ability .160 .265 .324 .348 .222 .417 .297 .125 .075
N0, (49) (47) ('40) ( 5) ( 5) ('22) (87) (. 4) ( 1)
Abihty .073 .072 .265 .217 .278 .262 .470 .250 -025
TOTAIS (668) (656) (151) (23) (18) (84) (185) (.16) ( 4)
1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000.
orgmmurow
Vgery (208)(60)(2)(1)(1)(3)(3)(000)(0)
High Inv. 311 091 .013 .044 .056 .036 .016 .000
High (178) (147) ( 18) ( 4) ( 2) ( 4) ( 4) ( 3) ( o)
Inv. . 5 .224 .119 .174 .111. .048 .022 .188 .000
rod. (160) (237) (48) (14) ( 3) (15) (16) ( 2) ( 1)
Inv. .240 .351 .318 .609 .278 .178 .086 .125 .250
Little (70) (113) (33) ( l) < 5) ('29) (35) (' 5) ( 1)
IDV- .105 .172 .218 .044 .278 .345 .189 .312 .250
gnarl» (370473) (’39) ( 2) < 3) (25) (1+8) (2) (1)
V- .055 .111 .258 .087 .167 .298 .259 .125 .250
Not (15) (26) (ll) (. 1) ( 2) ( 9) (.79) ( 4) ( 1)
Inv. .072 .040 .073 .044 .111 .107 .427 .250 .250

 

|l|l|llll|lJlll.ln'ull"

58

Table 4.--Test of Hypothesis for all Referal Points*

 

uh-..-

 

Hypo. (%)Hypo. (%) Real Real
1 2.100 ....- ..... 2.100 49.9% 40.1% .888
2 1.811 2.100 .476 .862 20. 5% 18 . 3% .406
3 1.243 3.911 .256 .318 7.6% 6.5% .145
4 1. 392 5.154 .194 .270 6.4% 11.8% .262
5 1.384 p 6.546 .153 .212 5.0% 10.1% .223
6 1.013 7.930 .126 .128 3.0% 5.4% .120
7 .815 8.943 .112 .091 2.2% 2.4% .0 54
8 1.565 9.758 .102 .160 3.8% 5.3% .118
9 .075 11.323 .088 .066 1.6% 0.0% .000

 

muff.) s 3.12

 

*Source: Table 3.

Figure 1. Percentage Distribution of Orientation over Each Social
Distance, for all Referal Points‘

50.0...
(%) Theoretical (*)
40.0.
(%) Empirical (.)
Percentage 30.0 '

 

Oriented
20.0 f
10.0 .. ,'
' * t o ... o
0.0 I ‘e I j;
l 2 3 4 5 6* 7 8
(least) Social Distance (greatest)

*Source: Table 4.

59

The data in Table 3 can be broken down by each referal point:
parent, peers, siblings, society, and spouse. Computations can then
be done for each referal point, to test the hypothesis.

Sampling problems are anticipated, so it is necessary at the
outset to establish certain "ground rules" for grouping adjacent social
distances:

Any S with N less than 5 is to be grouped with the nearest
lesser’distance with N of 5 or greater.

To avoid distortion of the scale of social distance, no distribution
at any S is to be moved more than two distancé ranks.

Empty cases will be left out of the computations.

The social distance of grouped Ss is the mean of the S of each
included rank,‘gflg., if 8 and 9 are grouped, the distance of the values
combined is 8.50.

They hypothetical and real distributions are presented in the

following tables and corresponding figures.

Table 5.--Test of Hypothesis for Parents

1.))

Predicted % '\ \
Distribution 52.2% 19.8% 7.8% 5.6% 6

l
s

.0% 1.9% )2.2% 2.8% 11.8%
g .

1 ' L g 7
Real % l E I
Distribution 42.0% 0.7%: 8.4%{11.6%12.9%3 0.0%;0.0% 4.3% 0.0%

 

 

p? r r I

7
muff.) - 3.44

 

60

Figure 2. Percentage Distribution of Orientation over Each Social
Distance, for Parents*

60.0
50.0*- (%) Theoretical (*)
40.0- (%) Empirical (.)

Percentage
Oriented 30.0

 

 

 

 

20.0 &
10.0 0 ’
u: g ,,
0.0L__' 6 a ' ._i
1 2 3 4 5 7 8 9
(least) Social Distance (greatest)
Source: Table 5.
Table 6.--Test of Hypothesis for Peers
3 1 2 3 4 5 6 (7 8 9)
Predicted %
Distribution 44.0 26.1 10.0 6.8 5.8 4.9 2.4
Real %
Distribution 40.6 18.5 10.2 7.9 10.6 9.4 2.8

 

Taliff.) = 3.60

 

61

Figure 3. Percentage Distribution of Orientation over Each Social
Distance, for Peers*

50.0
(%) Theoretical (*)
40.0?
(5) Empirical (.)
Percentage 30.0

 

 

 

 

Oriented *
20.0 .
10.0 . O .
I g .
000 I
l 2 3 4 5 6w 8
(least) Social Distance (greatest)
Source: Table 6.
Table 7.--Test of Hypothesis for Siblings
S l 2 3 (4 5) 6 (7 8 9)
Predicted %
Distribution 61.8 18.8 7.8 6.3 2.5 2.7
Real %
DiStribution 7602 2004 3014’ 0.0 0.0 0.0

 

i(Diff.) : 5.31

 

62

Figure 4. Percentage Distribution of Orientation over Each Social
Distance, for Siblings*

 

80.0
. (%) Theoretical (*)
70.0
(%)Enpirical(.)
60.0*
Percentage 50°C
Oriented
40.0
30.0
20.0 t
10.0
* I
0.0 ° 1‘ 1 I
1 2 3 455- 5 8

least Social Di t c t
*Source: Table ). ) 5 an 9 (grea est)

Table 8.--Test of Hypothesis for Society

 

 

 

S l 2 (3 4) 5 6 (7 8 9)
Predicted %
Distribution 57.4 21.2 8.9 4.7 4.3 3.4
Real %
Distribution 40.4 26.7 12.8 5.2 7.3 7.5

 

Imiff.) = 5.73

 

63

Figure 5. Percentage Distribution of Oriehtation over Each Social
Distance, For Society*

 

 

60.0
i
50.0 (5’3) Theoretical (*4)
40.0. (%) Empirical (.)
Percentage
Oriented 30.0
20.0 .
10.0 ; ‘ 1 .
0.0 ' "'
1 2 3.5 5 6 8
(least) Social Distance (greatest)
I"Source: Table 8.
Table 9.--Test of Hypothesis for Spouse
8 1 (2 3) 4 5 6 7 8 9

 

Predicted %
Distribution 74.9 25.1

 

Real%
Distribution 64.0 36.0

 

2(Diff.) s 10.9

 

64

Figure 6. Percentage Distribution of Orientation over Each
Social Distance, for Spouse*

80.0

70.0. (%) Theoretical (*)

60.0 (%) Empirical (.)

Percentage 50.0
Oriented
40.0

30.0
20.0
10.0
0.0
l 2.5
(least) (greatest)

Social
Distance

CHAPTER.4
CONCLUSIONS AND II-ZPIICATIONS

Our general orientation that adolescents show a great deal of
continuity between their perceived opportunities from referents and
their actual involvement with referents is in general substantiated,
on the basis of the data analyzed.

Further analysis needs to be directed at a breakdown of the
population, to see which kinds of respondents (male, female, parent-
orinnted, peer-oriented, urban, non-urban,...) behave as predicted, and
which do not. It is hypothesised that every category will exhibit
consistency and cognative rationality. This may or may not hold up.
Certainly it can be anticipated that there will be systematic diff—
erences both within and between societies, and within and between
referents.

The models in the second chapter show that for a certain
population, which is both adolescent and non-adolescent, and which is
considered with and without consideration of siblings and spouses,
the propositions brought forward. are validated. Here, again, the next
question is whether this behavior will hold up cross-culturally.

The model permits future research to find different distributions of
involvement for different perceptions of desire and ability. A break-
down. should also be made by each referent. For example, male and
female adolescents can be expected to show differential involvement

with parents that are perceived as providing the same opportunities, as

65

66

the social role expectation differ between sexes in all societies.

A descriptive model has been presented, to hypothesise what
variables are relevant to a study of adolescent behavior, and to see
in what ways these variables are interrelated. A cloéer look at the
social structure, within which these processes are hypothesised to
take place, is needed to explain why the variables in the model are
related as they are.. Even if the hypotheses and models could describe
adolescents in‘all cultural settings, they would not constitute a
scientific explanation. much of the description presented here is
methodological, but the explanation of the interrelationships must be
theoretical and sociological.

This paper is a preliminary model, and an orientation to the
study of adolescent behavior involving a fixed number of variables.

The usage of Guttman's facet design and analysis has been of consider;
able utility in developing the model and the Facet Paradigm. The value
of this metholology has been far greater than is apparent from the paper
V as written, as it was highly suggestive in choosing what variables are
relevant, and in what ways they are sematically interrelated, and
interrelated by our assumptions and postulates.

But we are essentially finished with Facet Theory and the algebra
of abstract systems. It has given a start. But defining variables and
making assumptions about them is not the stuff of sociological knowledge:
it is onlyrprolegomenous. Where the paradigm has led us to here bounds
the difference between definitional systems and hypothetical-deductive

systems. The philosophical implications of this can be illustrated by the

6?

theories of Parsons-Smelse r and of Homans.l This paper attempts to

begin to do what Homans did in The Human Group.2 So a breif exegesis of

 

the methodologies of both will be instructive..

Parsons calls his system of interrelated concepts a theoretical
model. It is a model, but it is not a theoretical model. His combining
of pattern variables lggk_likg, but on such a high level of abstraction,
they become virtually tautological and definitional. Like Parsons, thiS-
paper'has deveIOped a definitional system of variables and postulates:
but it is never claimed that this constitutes a theory. It does not
explain in any philosophical sense.

In the first Homans book, The Human Group, a lot of propositions
of the type “x varies as y" are deve10ped. This is heavily inductive,
building up from empirical case studies. The model developed in Chapter
1 enables us to follow a similar procedure. By reviewing the literature
on adolescence, we will attempt to generate--by an inductive process--

prepositions interrelating variables that appear in the model. Hopefully,

 

lSee Carl G. Hempel and Paul Oppenheim, "The Logic of Scientific
Explanation,” Parts I and II, in Feigl and Bradbeck, Readings in the
Philosophy of Science. Also see Ernest Nagel, The Structure of Science:
Problems in the Logic of Scientific Explanation (New York & Burlingame:
Harcourt, Brace & world, Inc., 1961), Preface and Chapters 1 and 2.

The methodologies are contrasted sharply in their respective theories
of human exchange: See Talcott Parsons and Neil J. Smelser, Economy and
Society (Glencoe, Illinois: The Free Press, 1956), and see George C. Homans,
Social Behavior: Its Elementagy_Forms.

2(New York: Harcourt, Brace, and Company, 1950). The parallels
are substantive as well as methodological, for Homans deals with refer;
ence groups and social structure. See Merton, "Continuities in the
Theory of Reference Groups and Social Structure," in Social Theory and
Social Structure, p. 285.

68

after the propositions deve10ped in this paper, and from a review of the
literature, are empirically tested in a cross-cultural setting, it will
be possible to borrow or invent more general (anatomical) propositions,
which, in conjunction with a conceptual schema, will lead to some kind of
a.thgg§y. From such a theory, the described propositions can be deduced.
This is precisely what Homans did in Social Behavior. But at this stage,
as with Homans in The Human Gropp, we are about to engage in "The first
process, the process of building up from the empirical to the more general
...,' with the ultimate purpose of "...building back down from the general
to the empirical. The first is an act of creation, which has no rules of
procedure that will unsure you success; the second has definite rules, the
rules of logic."3

This technique is debatable, but it can be defended on the grounds
that the process itself contributes to sociological knowledge, as prop-
ositions are useful, where definitions alone are not.

One other concept, in Chapter 3, needs a note of explanation.
As indicated, social distance is used quite differently here than in the
usual way. Perhaps the concept is too stretched here. But it is used in
a nominal sense, to defigg the equal weig.ting of two variables, which,
in conjunction with a third, the Opportunity structure for each case of
he sum of the other two, predict level of involvement.

Social distance, as a concept, lead this research to a theoretical

formulation of a relationship between three variables, to explain a fourth.

 

BSocial Behavior, p. 10.

 

69

80 social distance was useful in forming this hypothesis, but the test
of the hypothesis is not contingent on the concept. For, as stated,
it is nominally defined.

If the reader objects to calling the sum of agreement and interest
(direction and intensity of perceived intentions) social distance, he is
encouraged to call it whatever he likes. And if he objects, he is
objecting to the statement that the proposition is a generalization of
Stouffer's theory of intervening opportunities; but he is not objecting

to the h'.othesis nor to the testing of the hvoothesis.
. , Q «.4.

BIBLIOGRAPHY

A-.§22ka

Etzioni, Amatia. A Comparative Analysis of Complex Organizations.
Glencoe: The Free Press, 1961.

Finkbeiner, Daniel T. II. Introduction to Matrices and Linear
Transformations. San Francisco and London: W. H. Freeman and
Company, 1960.

Goldberg, Samuel. Introduction to Difference Equations. New York,
London: John Wiley & Sons, Inc., 1961.

Homans, George C. The Human Group. New York: Harcourt, Brace, and
Company, 1950.

. Social Behavior: Its Elementary Forms. New York,
Burlingame: Harcourt, Brace & World, Inc., 1961.

Krech, David, and Crutchfield, Richard S. Theory and Problems of
Social Psychology. New York: IvicGraw-Hill Book Co., 19148.

Nagel, Ernest. The Structure of Science: Problems in the Logic of

Scientific Explanation. New York & Burlingame: Harcourt,
Brace & World, Inc., 1961..

Parsons, Talcott. The figgjal System. Glencoe: The Free Press, 1951.

, and Smelser, Neil J ., Economyr and Society. lencoe:
The Free Press, 1956.

 

Stoll, Robert R... Setsp LogicJ and At-ziomatic Theories. San Francisco
and London: W. H. Freeman and Company, 1961..

Elm-9.1.92

Bastide, Roger, and van den Berghe, Pierre. "Stereotypes, Norms and
Interracial Behavior in Sao Paulo, Brazil," American Socio-
logical Review, 22. (December 1957), 689-694.

Bogardus, Emory S. "Measuring Social Distance," Journal of Applied
Sociologz. 9 (1925). 299-308.

 

7O

71

Cervinka, Vladimar. "A Dimensional Theory'of Groups,“ Sociometry, XI
(February—Hay 1948), 100-107.

Foa, Uriel G. "The Contiguity Principle in the Structure of Inter;
personal Relations," Human Religions, 11 (August 1958), 229-238.

Guttman, Louis. "The Principal Components of Scale Analysis," in
Samuel A. Stouffer et. gl., Studies in Social Psychology in
‘World‘War II, vol. 5: measurement and Prediction. Princeton:
Princeton University Press, 1950, 312~36l.

. "The Principle Components of Scalable Attitudes," in
Paul F. Lazarsfeld, ed., Mathematical Thinkipg in the Social
Sciences. Glencoe: The Free Press, 195R, 216;257.

. "An outline of Some New methodology for Social Research,"
Public Opinion Quarterly, 18 ( Winter 19 51+. 55), 39 5.414.

 

'. "What Lies Ahead for Factor Analysis?,' Educational and
PsyChological Research, 18 (No. 3 1958), 506-511..

. "A Structural Theory for Intergroup Beliefs and Action,"
American Sociological Review, 24 (June 1959), 13.

 

. "Notes on Terminology for Facet Theory,‘ Proceedings
of the Fifteenth International Congress of Psychology. Amsterdam:
North-Holland Publishing Company, 1959, 130-132.

 

‘. "The Structuring of Sociological Spaces," Technical Note
No. 3, Contract H. AF 61(052)-121 (December 1961), p. 10. This
report is a portion of the Proceedings of the Fourth International
Congress of Sociology. Stresa, Italy (1959).

Goldhamer, Herbert, and Shils, Edward A. “Types of Power and Status,n
American Journal of Sociology, #5 (September 1939), 171.

Hempel, Carl G., and Oppenheim, Paul. "The Logic of Scientific
Explanation,“ in Feigl and Bradbeck, Readings in the Philosophy
of Science, Parts I and II.

Kramer, E. N. "Dimensions of Prejudice," Journal of Psychology, 27

(1949). 389-451.

Merton, Robert K. "The Bearing of Sociological Theory on Empirical
Research,“ in Social Theory and Social Structure. Glencoe: The
Free Press, 1957, 86—89.

'. 'Continuities in the Theory of Reference Groups and Social
Structure," in Social Theory and Social Structure, 285.

72

Ravenstein, E. G. "The Laws of Migration," Journal of the Royal
Statistical Society, 48 (June 1885), 167-235, in Samuel A.
touffer, Social Research to Test Ideag. Glencoe: The Free
Press, 1962, 69.

Simon, Herbert. "A Behavioral Hedel of Rational Choice,I in Models

of Man. New Yerk, London: John'Wiley and Sons, Inc., 1957,
251- 26 0 .

C. Unpublished Materials

Foa, Uriel g. "The Scientific Development of the Isreal Institute
of Applied Social Rasearch,' Himeographed paper, n. d.

Gottlieb, David, and Guttman, Louis. "working Paper No. 1," Himeo—
graphed paper, (November 1962), p. 1—3.

Janowitz, Norris, "working Delimitation of the Subject matter of Social
Organization Research," Himeographed paper, (Fall 1962), p. 2.

r:."_'- b

 

MICHIGAN STATE LINN, ERSIT LIBR- W1 IE3

If I.’ ll} ‘
'I"I “ ‘
III lllll‘lll I

03175 0718

 

 

‘