SOCIAL STRUCTURE, BEHAVIOR, AND THE MEANING
CQMPQ MEN? OF SEW-SYMBOLS

Thesis for the Degree of M. A.
MlCHiGAN STA3'5 UNNERSITY-
Worth Cary Summers
@964

ABSTRACT
SOCIAL STRUCTURE, BEHAVIOR. AND THE MEANING COMPONENT
OF SELF-SYMBOLS

by Worth Cary Summers

The central focus of this thesis is the inter-
relationship between self-conception, social structural
involvement, and behavior. Symbolic interactionist theory
has asserted thatindividuals describe themselves in terms
of the group memberships which they feel identify them-
selves, the categories according to which their reference
groups identify them. and in accordance with norms associated
with their reference groups. Additionally. it is asserted
that these self-descriptions permit prediction of how an
individual will behave by indicating how he will define
situations. and the norms to which he will adhere.
Empirical results from the Twenty Statements Test have
attested to the validity and potential usefulness of this
view.

In an attempt to improve upon the kinds of self-
social structural-behavioral links that were forthcoming
from the Twenty Statements Test the Asset-Liability
Instrument (ALI) was developed. This instrument requires
respondents to list their most important assets and their

most important liabilities and thereby elicits evaluated

Worth Cary Summers

self-descriptions. It was reasoned that by thus specifically
requiring respondents to evaluate themselves, the value
according to which such self-assessments were made would
sharply delineate both the kind of involvement that the
respondent had in the social structure and his potential
behavior.

A sample of 372 students from introductory sociology
and social psychology classes were administered the ALI
and their responses were classified into nine content
categories. Hypotheses designed to determine the ability
of the ALI to provide the desired links between respondents,
their involvement in the social structure, and their
behavior were developed.

Problems encountered at a number of points in working
with the ALI and in interpreting results from it led to a
reconsideration and reformulation of the rationale which
had been used to justify its use. This reformulation
focuses upon the "relational" property of symbols including
self-symbols. In this view the meaning of all symbols
constitutes a class of forthcoming behaviors
between classes of actors and objects. It is on the basis
of the meanings of an actor's symbolic self-concept that
he is related to the environment. and according to which his
behavior is shaped and guided. Other symbols representing
the limits of the behavioral relationships which are appro—
priate to any given self-symbol are organized into normative

evaluative scales. A comparative relationship between the

Worth Cary Summers

symbolic representation of actual behavior and these limit-
ing symbols constitutes a self—evaluation.

Difficulties encountered in using the ALI are attri—
buted to the fact that rather than elucidating the connection
between an actor, the social structure, and behavior,
self-evaluations actually obscure it by virtue of their
added complexity. This fact, in addition to various other
technical problems associated with the ALI, are sufficient
to severely limit its utility as a means of yielding the
kind of information for which it was designed.

Recommendations are made for developing instruments
capable of exploiting the theoretical possibilities of the

reformulated position.

SOCIAL STRUCTURE, BEHAVIOR, AND THE MEANING

COMPONENT OF SELF-SYMBOLS

BY

Worth Cary Summers

A THESIS

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

MASTER OF ARTS

Department of Sociology and Anthropology

1964

ACKNOWLEDGEMENTS

There are a number of people to whom I owe a great
deal in connection with this thesis. I am and remain
especially indebted to Dr. Frederick B. Waisanen who served
as my thesis advisor until his departure for South America.
It was from him that I received a most stimulating and
challenging introduction to the ideas of Symbolic Inter-
actionist theory and it was he who suggested that I look
into the possibilities of developing an instrument utilizing
self-evaluations and who secured the sample used in the
present study. After Dr. Waisanen's departure Dr. Robert
L. Stewart kindly made room in a very busy schedule to
assume the time-consuming task of directing the thesis to
completion. The present form of the thesis owes a great
deal to his relevant and often painfully penetrating criticism
and suggestions. Fortunately for me, Dr. Stewart's skill
as a critic is matched by his patience. Thanks, too, are
due to Dr. Archibald 0. Haller for valuable suggestions
and criticisms, and to Dr. Carl Couch the third member of
the thesis committee. Needless to say, I alone am responsible
for any shortcomings in the present work.

It is unfortunate but inevitable that no expression
of gratitude would be adequate to convey my appreciation to

my wife Janet who has helped me in more ways than she knows.

ii

TABLE OF CONTENTS
Chapter
I. THE ASSET-LIABILITY INSTRUMENT . . . . . . .
Introduction: Purpose and Organization
of the Thesis . . . . . . . . . . . .

The Asset-Liability Instrument . . . . .

Coding the ALI
Scoring the ALI

Additional Measures and Indices . . . .
Hypothe ses O O O I O O O O O O O O O O 0
Description of the Sample

II. FINDINGS O O O O O O O O O O O O O O O O O 0

Method of Analysis . . . . . . . . . . .
Results 0 O O O O O O O O O O O O O O 0
Discussion and Critique . . . . . . . .

III. THEORETICAL REFORMULATION AND EVALUATION OF
THE ASSET-LIABILITY INSTRUMENT . . . . .

Introduction . . . . . . . . . . . . . .
The Theoretical Restatement . . . . . .
Evaluation of the Rationale Justifying
Self-Descriptions . . . . . . . . . .
Evaluation of the Asset—Liability
Instrument . . . . . . . . . . . . .
Implications for Further Research and
Development of the Theory . . . . . .
Summary . . . . . . . . . . . . . . . .

BIBLIOGMPHY O O O O O O O O O O O O O O O C O O O 0

iii

LIST OF TABLES
Table Page

1. Percentage distribution by people by number
of statements made in each asset and
liability category . . . . . . . . . . . . . 13

2. Percentage description of sample in terms of
sex, grade in school and socioeconomic
level 0 O O O O O O O O O O O O O O O O O O 27

3. Class in school and number of males making
social-skill asset and liability statements 33

4. Class in school and number of males making
normative asset and liability statements . . 33

5. Number of role and job skill asset and
liability statements by sex . . . . . . . . 37

6. Grade point average by number of intellectual,
scholastic asset statements . . . . . . . . 39

7. Grade point average by number of intellectual
scholastic liability statements . . . . . . 39

8. Number of Protestants and Catholics making 0 or
2 or more normative asset statements by
importance of religion . . . . . . . . . . . 41

9. Number of Protestants and Catholics making 0
or 2 or more normative liability statements
by importance of religion . . . . . . . . . 41

10. Relationship between number of social skill asset
and liability statements and organizational
membership . . . . . . . . . . . . . . . . . 43

ll. NUmber of individuals above and below the
median self-esteem score by socio-economic
level 0 O O O O O O O I O I O O O O O O O O 45

12. Self—esteem and perceived agreement with
significant others in self-evaluation . . . 46

iv

LIST OF FIGURES

Figure Page
1. The Asset-Liability Instrument . . . . . . . . 7
2. Question for important others and agreement

with important others . . . . . . . . . . . 18
3. Graphical representation of self-evaluation . . 65

CHAPTER I

THE ASSET—LIABILITY INSTRUMENT

Introduction

 

Belief that the self (a) reflects involvementin

the social structure, and (b) that it is an important
determinant of behavior, has resulted in continued efforts
to use measurements of the self as a technique for empirically
linking individuals to the social structure and for predict-
ing their behavior. The method that has been most often
employed for this purpose is the Twenty Statements Test
(TST)l in which respondents write up to twenty statements in
answer to the question "Who am I?" The theoretical rationale
underlying its development and justifying its use in empirical
research has been stated best by Kuhn:

There is no way for the individual to answer such

a question except by referring to himself as a

member of the groups that he feels identify him,

or to social categories which his reference groups

hold to be significant in identifying him, or

finally, to evaluations of himself which hinge on

norms held by his reference groups. Whichever of

these referents he chooses, his answer consists of

attitudinal statements, which are the best indexes
of what he will do in any situation. They point

 

1
(Thomas MCPartland).Manua1 for the Twenty Statements
Problem (revised, Department of Research, the Greater Kansas
City Mental HBalth Foundation, 1959).

to the norms that he will invoke to define a

situation, to determine the role that he will play,

and to supply the values that he will seek to

effectuate.2

Responses on the TST have been subjected to several

different scoring procedures and methods of content analysis;
results have confirmed an association between certain cate-
gories of content analysis and social structural involvement,
such as, religious affiliation,3 social class,4 age, sex,
and involvement in professional schools,5 and certain
behaviors, such as, differential readiness to evaluate
performance on the basis of cues from others,6 and behavior
of patients in a psychiatric ward.7 The development of the
Asset—Liability Instrument (ALI), which is to be discussed

and evaluated in this thesis, was stimulated by this theory

and by these promising results.

 

2C. Addison Hidkman and Manford Kuhn, Individuals,
Grogps, and Economic Behavior (New York: The Dryden Press,
1956): PP. 43-44.

3Manford H. Kuhn and Thomas S. McPartland, "An
Empirical Investigation of Self-Attitudes," American
Sociological Review, Vol. 19, No. l (1954), pp. 68-76.

4Thomas S. McPartland and John H. Cumming, ”Self-
Conception, Social Class and Mental Health," Human Organization,

5Manford Kuhn, "Self-Attitudes by Age, Sex, and
Professional Training,” Sociological Quarterly, vol. 1,
No. l (1960), pp. 39-55.

6Carl J. Couch, "Self-attitudes and Degree of
Agreement with Immediate Others," American Journal of
Sociology, V01. 63, No. 5 (1958), pp. 491-496.

7Thomas S. McPartland, John H. Cumming, and WYnona S.
Garretson, "Self-Conception and Ward Behavior in Two Psychiatric
Hespitals," Sociometry, Vol. 24, No. 2 (1961), pp. 111-124.

 

The ALI will be discussed in detail at a later point
in this chapter. Briefly, however, it asks respondents to
indicate their "most important assets (positive factors)"
and their "most important liabilities (negative factors)."
Behind the selection of this particular approach to the
measurement of the self was the belief that it would link
respondents to the social structure and to their behavior
more clearly and conclusively than had the TST. It was
reasoned that as with the TST, subjects would describe
themselves in terms that were associated with the reference
groups with which they were involved; but in addition, by
asking for assets and liabilities they would also assess
and describe themselves in accordance with the values
held by their reference groups. It was believed that these
values would (a) closely reflect their source in the social
structure and (b) would be an important influence on their
behavior.

Following Kuhn then, it was proposed that there was no
way in which an individual could answer the question posed
by the ALI--"What are your most important assets (positive
factors?)" and "What are your most important liabilities
(negative factors?"--except by:

1. "referring to himself as a member of the groups
that he feels identify him, or to social cate-
gories which his reference groups hold to be
significant in identifying him,

2. [and by simultaneously evaluating himself according

to the values derived from norms held by his
reference groups.]

3. "Whichever of these referents he chooses, his
answer consists of attitudinal statements, which
are the best indexes of what he will do in any
situation. They point to the norms that he will
invoke to define a situation, to determine the
role that he will play, and to supply the values
that he will seek to effectuate."

Accordingly, the original objective of this thesis was to
determine whether the ALI would, in fact, live up to these
expectations.

Hewever, this determination has been complicated
considerably by developments which occurred subsequent to
the collection of the data. Problems encountered in
developing the coding categories, in the selection of
hypotheses, and in attempting to analyze and interpret
the results led to a reconsideration and reformulation of
the theoretical rationale underlying and justifying the
development of the ALI. This reformulation has had two
important consequences which have greatly affected the
organization of this thesis. First, it has confirmed the
soundness of Kuhn's assertion regarding the importance of
self—descriptions. Second, it has, on the other hand,
demonstrated the invalidity of the basic assumption upon
which the ALI rests--that self-evaluations would yield
"better" results than a self-description,—-and furthermore
that this invalidity" is the source of much of the difficulty

encountered in using the ALI. While these problems can

be understood and explained in terms of the reformulated

 

position, they cannot be eliminated in this thesis for they

are a necessary result of the instrument itself and can
only be avoided by using different techniques.

As a result of these developments, the emphasis of
this thesis has shifted from an evaluation of the ALI based
on empirical results to an evaluation based upon a consider-
ation of the theoretical foundation upon which the instrument
rests. With respect to the organization of the thesis this
occurs in the following ways:

1. The discussion of the ALI and an interpretation

of the results of the ALI is carried out in

accordance with the original line of argument upon

which it was based, insofar as this is possible.
2. Difficulties encountered in attempting to do

this are indicated.

3. The problems encountered are then explained as
resulting from technical characteristics of the

ALI, and from inconsistencies between the

assumptions underlying the ALI and the reformulated

theoretical position.

In the remainder of this chapter the Asset-Liability
Instrument, the procedure for coding and scoring it, and the
hypotheses relating to it are discussed. In addition,
some of the problems encountered in working with it are
pointed out. In Chapter II, the results of the hypotheses
are presented along with a brief summary and discussion of

some of the technical and interpretive difficulties associated

with its use. In Chapter III, the theoretical reformulation
is presented, followed by an evaluation of the ALI in which

it is indicated how the inconsistencies between the ALI

and the reformulation led to the problems encountered in
working with and in interpreting the results of the instrument.
Finally, a brief discussion of the implications that the
theory and the flaws uncovered in the ALI have for the
development of suitable instruments and for further theoreti-

cal development is presented.

The Asset—Liability Instrument

The ALI is an adaptation of questions initially
employed as a technique for distinguishing between inner
and other directedness. Riesman8 asked respondents to indi-
cate their three best personality traits, and Littunen
and Gaier,9 and Gaier and Wambach10 had respondents list
three of their greatest personality assets, and three of
their greatest personality liabilities. Although the

results of these studies are irrelevant11 to the main

 

8David Riesman, Faces in the Crowd (New Haven, Conn.:
Yale University Press, 1950).

 

I 9Yrjo Littunen and Eugene L. Gaier, "Occupational
values and Mbdes of Conformity," Journal of Social Psychology,

10Eugene L. Gaier and Helen S. Wambach, "Self-
Evaluations of Personality Assets and Liabilities of Southern
White and Negro Students," Journal of Social Psychology,
Vol. 51, 1960, pp. 135-143.

11They are irrelevant largely because they are
uninterpretable, Riesman‘s because of the extreme informality
with which the responses were used, ani Littunen £3.éi
because of inappropriate application of statistical analysis.

purpose of this thesis, the kinds of answers which these
questions elicited suggested that a similar question
might lead to a significant differentiation of respondents
according to the evaluated self—descriptions. The ALI

appears below.

Figure l. The Asset-Liability Instrument.

1. In the space below please indicate what you think are
your most important ASSETS (positive factors).

 

 

 

 

 

 

 

 

 

 

11. In the space below, please indicate what you think are
your most important LIABILITIES (negative factors)

 

 

 

 

 

 

 

 

It can be seen that several characteristics were
built into the ALI in order to bring it into accord with
the theoretical statement with which this chapter was begun.
First, as has already been indicated, evaluations are explicit—
ly called for by asking for assets (positive factors) and
for liabilities (negative factors). Second, the instrument
was made more open-ended than the question employed by

Riesman, and Littunen, Gaier, and wambach, by providing for

eight, rather than three, responses in order to allow for
the possibility that, if given the opportunity, the subjects
might provide additional useful responses. Third, since

the kinds of self-responses that are made are of central
importance for the theory it was felt that by removing

the term "personality,' as used by the above mentioned
authors, the possibility of a systematic biasing of the

self-symbols would be removed.

Coding the ALI

Content Categories

Nine categories were developed into which the asset
and liability responses were separately coded by content
analysis. It is to be emphasized that the theory provided
no explicit basis for deciding a priori what properties
or characteristics the categories should exhibit. In a
general and rather vague way it was assumed that all of the
responses included within a category would represent some
underlying value or valued category of self-identification,
and that it would suggest, (a) a possible reference group
as its source, and (b) certain kinds of behaviors. The
basis for deciding on such a connection is undefined by the
theory; more about this will be said under the section
headed Hypotheses in Chapter I. Because of the lack of
explicit criteria for grouping certain kinds of responses

into categories, meanings were imputed to responses and

hence also to the different categories. More will be said
in Chapters II and III concerning the validity of imputing
meanings to self-symbols.

1. Physical. These are statements which deal
with the respondent's physical appearance, or other
attributes which suggest that the respondent views him-
self as a physical object. Statements dealing with physical
characteristics of health are also included.

Examples include: "strength," "looks," "coordination,"
"complexion," "too tall," etc.

2. Material or Economic. These statements reflect
differential economic advantage or disadvantage or possession
or lack of possession of material goods. Individuals are
Q9; included in this category.

Examples include: "few luxuries," "fail to make
ends meet," ”need of money,” "accumulated savings," but not
"my friends." Also included are statements indicating desire

for economic advantage, e.g., "desire for financial gain."

3. Significant Others. These statements presumably
indicate close primary relationships or an appreciation of
specific individuals.

For example, "my friends," "good wife," would be
scored as category 3, but, ”friendly" would be scored as
category 4—-social skills. Similarly, "my fiance" would
fall in category 3, but "engaged" would fall under

category 6--statements of Social Location or Categorical
Placement. Similarly, fhusband," “father," given by a

male respondent as an asset would be placed in category 6-—
Social Location, but if he had listed "my wife," or "son,"
these would be included in the present category.

4. Social Skills. These statements mention a
specific social ability or lack of ability and are regarded
as indications that the respondent perceives interaction and
social involvement in general to be important per se. All
statements in which people are mentioned are automatically
included provided they do not fall into category 3-—for
example, "ability to speak before a group." Not included
are some statementsjn which social relationships themselves
appear in a negative light. For example, "too easily
influenced by others," seems to be a statement of insuf-
ficient autonomy and is classified into category 5—-
Normative, Imperative, or Directive statements.

 

10

Examples include, "poised," "talk too much,"
extrovert," "interested in others," “sarcastic," "sense
of humor," "personality."

5. Normative, Imperative or Directive. These
statements seem to be generalized descriptions of ideal or
normative behavior frequently having the characteristics
of values or of moral or ethical imperatives. They are
regarded as representing valued, culturally carried,
generalized recipes for behavior applicable to many situations.
They are abstractions of complex ways of acting or behaving.

Examples include, "haven’t a set goal," "ability to
adjust," "not enthusiastic enough," "rebellious to authority,"
"desire," "liberal outlook," "good Republican," "little

prejudice." But statements such as, "feel insecure,"
"moody," are placed in category 8, a residual category which
includes markedly subjective and affect states.

6. Social Location or Categorical Placement. These
statements serve to place the respondent within some clearly
defined social category. They distinguish him from others
and establish his social position on the basis of certain
well defined roles or other classificatory criteria. In
contrast to category 5 statements, it is assumed that by
making self-assessments within this category a respondent
would tend to see himself in terms of categories of other
actors rather than in terms of the more general and pervasive
ideals or norms.

Examples include, "member of the middle class,"
"white," "ROTC Cadet," "married," "fraternity man,"
"male . ll

7. Job or Role Skills or Their Lack. These
statements express the respondent‘s possession of skills
or abilities requisite to performance within a specific role
or setting. Also included are capacities to enjoy various
activities.

Examples include, "manual skills," "athletic,"
"fast runner,“ "good musician,“ "good salesman."

8. Non—Specific or Idiosyncratic. These are state—
ments of mood or affect and general statements that are
neither imperative nor especially predictive or helpful in
defining subsequent behavior. They may indicate a lack of
facility in defining one's relationship to people or situations.

Examples include, "luck," "upbringing," "early
family environment," "faith in the future of America,"
"education," "backing of my parents and family." Also

included are statements of future events which may indicate

11

a lack of direction, e.g., "no job for the summer," "armed
service obligation," and abstract or general conditions
which may accrue to one, e.g., "opportunity to advance,"
"freedom as an American," "have traveled much," ”think

all people have some good."

9. Intellectual or Scholastic Characteristics.

This is a category which was created for the college sample.
Statements which are included would ordinarily have been
classified into one of the other categories. Included are
all references to characteristics which would directly
pertain to scholastic or intellectual competence.

Examples include, "smart," "have difficulty concen-
trating on my school work," ”I catch on to things quickly,"
"poor grades." '

There were occasions when certain responses could
have, with some justification, been placed in either of
two categories. In such instances, if the statement in
doubt appeared between two other statements, both of which
clearly fell into one of the disputed categories, then it was
included in that category: otherwise, the statement was

included in that alternative category which had the greatest

frequency or saliency for the respondent.

Coding Reliability

Two measures of the reliability of the coding
procedure are available: one, a measure of the reliability
over time using the same coder, and the other, a measure of
inter-coder reliability. After a lapse of 2-1/2 years
the original coder was able to code 94% of all of the
responses of 20 subjects into the same categories into which
they had originally been placed. On another sample of 20

subjects, using an independent coder, 87% of all of the

12

responses from 20 subjects were placed into the categories
into which they had been placed by the original coder.

The essential point to consider in connection with
the coding procedure, however, is not the apparent
"reasonableness" of the consistency-reliability figures but
the validity of the coding procedure; that is, the
correspondence between the meanings that the coded
responses have for the subjects and the meanings which
the various categories have to the researcher. A great
deal more will be said in reference to the validity of the
coding procedure in the course of the discussion of
meaning of the categories which occurs at the end of
Chapter II under Critique, and especially in Chapter III.

The frequency distribution of the respondents'
responses in each of the asset and liability categories
appears in Table l. The frequency distribution of each
category as it relates to the scoring and analysis of the
results will be discussed in this chapter under Scoring

the ALI, and in Chapter II under Method of Analysis.

Scoring the ALI

 

Three alternative procedures for scoring the ALI

were considered--the Medal, Proportional, and the Absolute.
In the Modal technique each individual would be assigned

to the category into which the majority of his responses were
coded. In the Proportional technique each individual would

be given a score for each response category according to the

14

 

.mnommumo muflaanmea some new ummmm some now mum u 2*

00H m. m. H.H m.mH m.mh oaummaosom
HmsuomaamucH

OOH m. o.¢ o.ma m.mm o.mm can
ImuocmmompH
oamaommmlcoz

ooa m. m. o.m m.om maaflxm maom
so now

00H m. o.m m.mm coeumooq
ameoom
ooa m. m. v.m o.v H.m ¢.ma m.mm m.bm mbfiumauoz
ooH m.a m.HH m.mm m.mm maaaxm
Hafiuom

00H m. m. H.H w.nm mumsuo

DQMUHMflcmHm

13

 

 

OOH m. m. m.m e.m O.sm oHeocoom
Ho amauwumz
OOH m. O.e s.OH O.Os HmOHmsem
s Hmuoe O s O m e m m H O
mucmEmumum muHHHQwflH mo HmQEsz

OOH 0.0 e.ee O.se UHummHoeom
HmsuomHHmucH

. . . . . UHHmuocsmomOH
OOH m m H m s sH H on onHommmuaoz
OOH m.H O.e O.mH m.ms mHHme
Odom so now

OOH m. m. m.H O.e 0.0H m.mm coHumooH
HMflUOm
OOH m. m. m. m. m.e O.m m.HH m.sH 0.0~ H.Om m>Hum2uoz
OOH m. m.H H.H e.O 0.0H e.Om 0.0m mHHme
HMfloom

OOH m. O.H O.m O.s m.HH m.ms mumeuo
DCMUAMHcmfim

OOH m. m. O.H N.N m.s m.sm UHsocoom
Ho HMHkumE
OOH m. m. m. m. m.O H.Om m.mO Hmonsem
s Hmuoe OH O m s O m e m m H O mmHHommumo
mucmfiwumpm ummmm mo Hmnfisz ucmusoo

 

 

Ucm ummmm comm CH meme mucmfimumum mo Menage >9 mamomm mo soflusnfluumwp mmmucmoumm

e.>uommumo huHHHQMHH

.H magma

15

ratio of the number of his responses in that category to
his total response frequency. In the Absolute technique
some arbitrary number of responses in a given category
would be defined as distinguishing between individuals
that have or have not responded to that category in a way
that is meaningful for the analysis.

Although the Mbdal technique is simple and has
frequently been employed with the TST, responses on the
ALI did not fall clearly into modal categories and so this
approach was not used. A final choice between the
Proportional or the Absolute technique would ultimately
need to be based on theoretical and empirical considerations.
Since this thesis is not intended as an exploration of
various scoring techniques, and since neither the original
theory as stated by Kuhn nor the reformulation of it in
Chapter III provided any basis for deciding between one or
another of these alternatives, the decision to use the
Absolute technique was based upon the fact that of the two
it raised fewer difficulties than the Proportional technique.
The main difficulty in using the Absolute technique is the
establishment of cutting points; the basis on which this
was done in this thesis is to be found at the beginning
of Chapter III.

The Proportional scoring method was not used for
several reasons. It is possible to have anywhere from 0
to 8 responses in any one of the nine categories for

assets or for liabilities; because of the great number of

16

categories in relation to the total number of responses
the maximum was around 4. Furthermore, for each category
the median frequency was always either at 0 or 1 responses.
Since the median number of Egggl responses was 5 for
Assets and 3 for Liabilities, any score based upon a pro-
portion would ultimately rest upon very small numbers in
both the numerator and denominator. This would have the
consequence of making the Proportional score highly sensitive
to variations of one or two responses and hence, sensitive
also to any unreliability or invalidity of the coding
procedure. Uhder these conditions, using such proportions
as a metric score would be highly misleading.

One bit of evidence gives weight to the validity
of using the Absolute scoring technique. In the course
of the analysis of each of the hypotheses the sample was
dichotomized according to those making fewer and more than
the median number of total responses for assets and for
liabilities. In every case there was no difference in
the results for those who made more than the median and
for those who made fewer than the median number of total
statements. Thus, it would appear that the absolute pre-
sence or absence of a response in a category was more signifi-
cant than its frequency relative to the total number of
statements.

Assets and Liability statements were analyzed
separately in this thesis because it was not known if they

could be meaningfully combined. For example, it was not

17

known if a Social Skill asset statement was equivalent to
or opposite from a Social Skill liability statement.
Furthermore, it was not known how they would behave with

respect to different independent variables.

Additional Measures and Indices

The purpose of the ALI is to link an actor to the
social structure and to behavior by virtue of his self—
assessments; therefore, several indices and measures of
reference group involvement and of behavior were obtained.
These are described and defined below prior to their use
in the hypotheses. The extent to which the following
measures are appropriate to the problem is considered
in Chapters II and III.

Important Others and Agreement
with Important Others

The following question was used to elicit the sub—
ject's reference groups and his perception of the extent
to which they agreed with his own self-assessment. It
was originally intended that the "others" would, like the
ALI responses, be classifiable into categories, e.g.,
primary, secondary, etc.; however, it was not possible
to work out a satisfactory set of categories. Nevertheless,
the number of "others" mentioned, as well as the amount
of Agreement, do figure in the Mean Agreement Score. The
scoring of the Agreement and Mean Agreement portion of the

following question are given below.

Figure 2.

18

with important others.

Question for important others and agreement

All of us have certain "other" (individuals and/or groups)

who are particularly important to us.

Please list some

of your important others in the spaces at left below.

For each "other" listed,

indicate whether there would most

likely be agreement or disagreement with the assets and
liabilities you have listed above.

Others Who are
Important to You

 

 

 

 

 

 

 

 

Would this "other" generally agree or
disagree with you regarding most
of the assets and liabilities you
have listed
appropriate

Would

Would

(Would

Would
Would
Would
Would

Would

Agreement and Mean Agreement

agree
agree
agree
agree
agree
agree
agree

agree

Score

above?
response.

can't
I
can t
I
can t
t
can t
I
can t
I
can t
I
can t

can't

Circle the

say
say
say
say
say
say
say

say

may
may
may
may
may
may
may

may

disagree
disagree
disagree
disagree
disagree
disagree
disagree

disagree

In the Important Other and Agreement with Important

other question respondents were asked to indicate if each

important other which they listed "would agree," "can't

say,‘

liabilities statements which they made.

or "may disagree," with most of the assets and

Each respondent

was given an agreement score by scoring "would agree" as

2, "can't say" as l, and "may disagree? as 0 and then

l9

summing across all "important others." A Mean Agreement
Score was then computed by dividing the Agreement Score
by the total number of important others listed. This last
score served as the measure of perceived agreement with
significant others and was independent of variations in the

number of significant others listed by the subjects.

Index of Self—Esteem

.This index was obtained by dividing the total number
of asset statements by the total number of liability state-
ments. The rationale for this is based upon the assumption
that more asset than liability statements are an indication
of a more favorable than unfavorable self-evaluation. In
order to avoid certain metric difficulties involved in
utilizing ratios as scores the index of self-esteem and
the mean agreement score were divided at the median

when they were used in the testing of hypotheses.

Importance of Religion

Respondents were asked "how important is your
religion in the way in which you live your life?"; five
alternatives were possible from "One of the most important
influences in the way that I live my life," to "One of the
most unimportant influences in the way that I live my
life." Responses were dichotomized by combining "one of the

most important . . ." and "a fairly important . . . into

20

an "Important" category, and "Neither important or unim—
ll

portant . . . "Fairly unimportant . . ." and "One of the

most unimportant . . ." into a "Not important" category.

Socio-economic Level

Respondents were asked to indicate their fathers'
occupations; five socio-economic levels were computed using
the NOrth-Hatt index of occupational prestige and are based
upon a breakdown employed previously by Davis g; $1.12
The Elite group contains such occupations as major pro-
fessionals, presidents of medium large firms, management
in large firms; for the Middle-Middle, secondary teachers,
wholesale salesmen and middle managers are representative:
Working Class Elite include plumbers, carpenters, owners
of small retail stores, white collar supervisors, and
farmers; Respectable Working Class is characterized by
postmen, barbers, mechanics, busdrivers, clerks in retail

stores, and machine operators; Low Status by garbage

collectors, janitors, and truck drivers.

Hypotheses

 

Not all of the possible combinations of content
categories of the ALI and independent measures of involve-

ment with reference groups or behavior are specifically

 

12James A. Davis and David Gottlieb, Jan Hejda,
Carolyn Huson, Joe L. Spaeth, Stipends and Spouses/The
Finances of American Arts and Science Graduate Students
(Chicago: University of Chicago Press, 1962).

21

accounted for in the following hypotheses. Eclectic as
they are, these specific hypotheses were chosen because,
when taken as a whole, they seemed to provide a reasonable
test of whether or not the ALI constituted a valid means

of linking, (a) the self, (b) reference groups, and

(c) behavior. There was no intention that the hypotheses
should contribute to any substantive sociological or sociaL-
psychological knowledge.

While all of the following hypotheses seem to be
more or less "reasonable" in light of Kuhn's statement and
the meanings which have been imputed to the response
categories, reasonableness is a far cry from theoretical
specification. The position represented by Kuhn's state-
ment seems capable of asserting only that the symbols in
terms of which individuals represent themselves will ig
some unspecified way have their links or connection to ggggp

and to behavior revealed ipso facto. This serious theoreti—

 

cal hiatus between the self-symbol, social structural
involvement, and behavior is bridged in the case of these
hypotheses only by intuition, empathy, and some vague
sense of appropriateness. Nor is there any help along
these lines to be had from the theoretical reformulation
in Chapter III. The nature of the difficulty is more
clearly seen, however, and it is possible to offer some

suggestions for theoretical development in this area.

Hypothesis I. According to Clark the "collegiate

subculture," characterized by an orientation that is

22

predominantly concerned with social activities, "has been

a strong if not the dominant governor of student life ever
since the 1890's. . . . Today, the fraternities and
sororities set the patterns of the collegiate way of life."13
This view of one of the predominant value orientations of
college students, in conjunction with the foregoing inter—
pretation of the meaning of the Social Skill category,
leads us to assume that over time, the effect of the
"collegiate subculture" would befatjincrease the likeli-
hood that subjects would come to value and to evaluate
themselves according to standards that were consistent with
the kinds of responses which were categorized into the

Social Skills category. Hence Hypothesis Ia.

Ia. The number of Social Skill asset and liability
statements will increase with grade in school.

At the same time that socialization is presumably
occurring within the "collegiate subculture" the effects
of socialization from the family and church are presumably
weakened and decrease in importance. The Nbrmative
category is presumed to represent values and criteria for
behavior derivedirom a familial or religious context.
Hence Hypothesis Ib.

Ib. The number of Nermative asset and liability
statements will decrease with grade in school.

 

13Burton R. Clark, Educating the Expert Society
(San Francisco, Calif.: Chandler Publishing Company,
1962), p. 203.

23

Hypothesis II. The traditional role for the male
in our society is to assume responsibility for securing a
livelihood; therefore, we expect males to be more likely
to see and evaluate themselves in terms of these roles.

II. Males will make more Role and Job asset

and liability statements than will females.

Hypothesis III. Either because the definition and
evaluation of one's self in terms of Scholastic or Intel-
lectual attributes leads to higher grades or because higher
grades lead to such self-assessments, saliency of
Intellectual or Scholastic statements should be associated
with objective evidence of grades.14

IIIa. Higher gradegpoint averages will be

associated with those individuals who
make Intellectual or Scholastic asset

s 8 se 0 t ose who make
none.

IIIb. Lower grade point averages will be

associated with those individuals who make
Intellectual or Scholastic liability

statements as opposed to those who make
none.

Hypothesis IV. If religion is perceived to be an

 

important influence on the way in which one lives his life,
then we would expect individuals to view and evaluate them—
selves in terms of standards associated with that religion.

Such evaluations presumably would yield responses on the

 

l4Wilbur B. Brookover, Ann Patterson and Shailer
Thomas, Self-Concept of Ability_and School Achievement,
Final Report of Cooperative Research Project No. 845,
Office of Research and Publications, Michigan State
University, 1962.

24

ALI that would fall into the Normative category. Hence,

IVa. Those who perceive their religion to be
important in the way_in which they_live

their lives will make more Normative

asset and liability statements than will

thgse er thm religign IS at little gr hQ

importance.

 

Hypothesis V. Membership in many organizations
should be associated with a tendency to value interpersonal
relations and the possession of certain associated skills.

V. Those who belong to more organizations
while in school will make more Social Skill
asset and liahility statements than will
those individuals who belopg to few or no
organizations.

Hypothesis VI. There is ample evidence15 that

 

individuals at the lower economic levels of our society
are exposed to fewer of the opportunities and possess
fewer of the skills requisite for adequate performance in
educational, social, and economic pursuits.. Such a
condition is likely to produce experiences which cause
these individuals to evaluate themselves relatively
unfavorably. Furthermore, differences between socio-
economic levels are likely to be more marked when they are
within a context which places a premium on certain skills
that tend to be distributed in favor of those at higher

economic levels.

 

15

of Status to Experience, Perception and Value,
Journal of Sociology, Vol. 66 (1960), pp. 1-31.

Cf.. Alex Inkeles, "Industrial Man: The Relation
“ American

25

VIa. Self—esteem as measured by_the ratio of
asset to liability statements will increase
with socio-economic level.

Hypothesis VII. Assuming that individuals attempt
to maximize relatively favorable evaluations of themselves,
that these evaluations are largely based upon cues from
significant others, and that such cues will more likely
be forthcoming when an actor shares the values of these
significant others, then, when the evaluations of signifi-
cant others are perceived to correspond to one's own self-
evaluations, there should be a maximizing of favorable
responses from them, and correspondingly, there should
be high self-esteem.

VII. There will be agppsitive correlation be-
tween self-esteem and mean agreement score.

This list of hypotheses originally included several
hypotheses which related categories of the ALI and categories
of Important Others. Hewever, these had to be abandoned
when the proposed classification of Important Others proved

to be totally unworkable.

Description of the Sample

The questionnaire was administered to classes of
beginning social psychology students at four institutions—-
112 were from Michigan State University, 112 from Northern
Michigan University, 107 from Michigan Tech., and 40 were

from Suomi College. Table 2 presents a summary description

26

of the sample in terms of sex, class in school and socio-
economic level.

The fact that this sample was not randomly drawn
will not be of concern in this thesis since there will be
no attempt to generalize any results to a population.
Furthermore, the mere presence of connections between an
actor, social structure, and behavior, which are expected

to be revealed by the self, are assumed to obtain regard-

less of what subsamples of any population might be observed,

although the kinds of connections will vary from sample to

sample.

27

 

 

 

 

 

mum KOCH KOCH KOCH KOCH KOCH KCOH Hmuoa
CNH mm Ow O.Nm m.Cm m.mm mm mHmEmm
mmm ms gm v.5O m.mm m.hm me mHmz
3. .NA N anv wanna H Cuzv 3.53 m vuzv doing Hm anv QQKA b muzv OBIKACCHZC
Hmuoe mHQmCHmuHmomC =wuHHm: zmHOUHS zmuHHm :mmMHO =m5umum xmm
uoz prpHZ: mmMHU OCHMHOK 30H:
OCHHHOS: mHQmuommmmM=
Hm>oq UHEOCoomloHoom
th KOCH KOCH KOCH KOCH KOCH Hmuoa
ONH C m.ON m.mm O.bm m.O¢ mHmEmm
Nmm COH m.m> m.CO m.mO m.mm mHmz
HsHuzO HOOuzO lemnzv HOOHuzC Hemuzv
Hmuoe mmumspmuw mHOHCmm mHOHCCC mmHOEOCmom Cmfismmum xwm
Hoonom CH mpmno
.Hm>wH UHEOCoom
.m mHQma

IOHoom pCm HOOCUm CH OUMHC

:xwm mo mEHmu CH wHQEmm m0 CoHumHnommp mmmquUHmm

CHAPTER II

FINDINGS

In this chapter the results from the hypotheses will be
presented and discussed; in addition, a brief critique of the
ALI will be made in terms of these results and hi terms of
certain of the methodological and interpretive difficulties
associated with the instrument itself. A theoretical

critique of the instrument will be postponed until Chapter III.

Method of Analysis

 

Having decided to use the Absolute method of scoring
for the ALI, it is necessary to determine those cutting
points of the distribution that will yield groups which
differ in the property represented by the response category
in question. With self-esteem, mean agreement score, and
gradepoint average it was possible to use the median as
a point for dichotomizing the distributions--a procedure
that has the double advantage of being consistent from
application to application and of being the most power-
ful point at which to dichotomize these distributions for the
Chi Square test. Unfortunately, the task was more compli—
cated for the frequency distributions of ALI categories.

As can be seen from Table l in Chapter I, these distributions

28

29

are markedly skewed in the direction of many responses for
each of the categories and the modal frequency was always
either 0 or 1.

As a result of the restricted variance and skewed-
ness of these distributions and because the total number of
persons at the median comprised a rather large proportion
of the total number of persons in the sample, the use of the
median as a cutting point posed some special problems in the
analysis. Three possible ways of arriving at cutting points
for the frequency distributions of the various ALI categories
were considered. All respondents who fell in the group
in which the median also fell could be:

1. Randomly assigned to the groups on either side of
the median. This procedure was rejected because
reassigning the large number of cases in the median
group would tend to wash out any difference that
might exist between the groups above and below the
median.

2. Dropped from the analysis. In which case,

a) when the median fell at 0 it would be clearly
impossible to analyze the results, there no
longer being any group below the median.

b) When the median fell at l the analysis could
proceed with a 0 or 2 or more response split.

3. Retained as a separate group. In which case,

30

a) when the median fell at O the analysis could
proceed with a 0, or 1 or more, response split.
b) when the median fell at l the analysis could
proceed with a three-way split of 0, l and
2 or more responses.

Of methods "2" and "3" the preference for the
analysis was for method "2"-—dropping the median group from
the analysis--for two reasons. First, the groups which
remained would be "extreme" and more likely to reveal
differences in relation to control variables. Second, it
is believed that method "2" tends to reduce any effects of
unreliable coding. Thus, it is more likely that gas of the
responses of an individual might be miscoded than that twp
might be and it is therefore possible that the "1 response"
category has a disproportionate share of individuals who are
there by virtue of coding error.

Nevertheless, method "3" was used under the two
following conditions: (1) if discarding the "1 response"
category drastically reduced the number of individuals falling
in the "2 or more" response group thus resulting in the
distribution being too skewed for a reliable application
of the Chi Square test or (2), if the result of retaining
the "1 response" category was clearly the same as the result
obtained by omitting it. Under the second condition it
was retained in the analysis to emphasize the fact that the

relationship was strong enough to be revealed by one response.

31

This last contingency was applied to only one case in the
analysis, Hypothesis IVa. It should be noted that in every
case the result of omitting the "1 response" category was
such that it strengthened an already existing relationship
or had no effect when a relationship was not present. In
no case where this category was omitted would its inclusion
have changed the direction of any obtained relationships;
therefore, the procedure provided a "fair" test of the

hypotheses.

Results

Hypotheses Ia and Ib proposed that:

The number of Social Skill (SS) asset and liability
statements will increase with grade in school.

The number of Normative asset and liability state—
ments will decrease with grade in school.

It will be recalled that a complimentary increase
and decrease of these two categories of evaluation was
expected as a result of the assumption that they were
reciprocally related modes of evaluation. That is,
students would come to reflect, through their SS statements,
what was assumed to be a dominant orientation of student
reference groups--an emphasis upon social activities, personal
attractiveness, and acceptance by others. It was also
assumed that normative moralistic criteria of behavior supposed-
ly derived from familial and religious reference groups,

and supposedly also related to the Normative response category,

32

would become less relevant or appropriate in the new
context.

The results, presented in Table 3 and 4, do not
support such a simple interpretation. Only males were
included in this analysis since it was found that females
made significantly more SS statements than did males, thus
introducing the possibility that the scoring by the Absolute
technique would be affected by different numbers of females
at the various class levels. Although females did not differ
from males in this category, for the sake of uniformity in
the discussion of results females were also removed from
the analysis of Hypothesis Ib, since the two hypotheses
are related. When males and females were analyzed separately
for Hypothesis Ia there were some slight departures from
the overall pattern of relationships1 which had originally
obtained with the combined group. While these were not
great and the overall pattern remained the same, males
and females differed sufficiently to introduce an unwanted
interpretive complexity at this stage of the analysis.

The tables reveal that, in general, males increase
the number of Normative ahg SS evaluations which they make
as grade level increases. For, whereas there are fewer
than the expected number of freshmen making "many" asset
or liability statements for both categories of response,

seniors are overrepresented in all of these categories. It

 

lThe departures merely consisted of differences in the
relative magnitudes but not direction of the discrepancy be—
tween actual and expected frequencies at the sophomore and
junior grade levels for the combined group.

=.mHmmHmCd mo COCHOEO CH Mm wHCH CuH3 OUCMOHOUUM CH pmumHCm mm3 uCHom OCHHHCOHH

.mOHUCwCOmHm pmuommxm UCmmmHmmH mmmmCquHmm CH mmHCmHme

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ONO. A C A OO. O .O.O Oe.O u NH OOO. A C A HO. O .H.O OO.NH u NH
OO be OOH Os OO Hapoe
em HON.OHO OH HON.OHO ON OO AN0.0HO O HOO.OHO KN OCOHOOO
we HOA.ONO HN HHN.NNO HN OO AOO.NNO ON HHH.HNO 5N muoHOOO
Oe HOO.ONO HN HOO.HNO ON Oe HO0.0NO NN HOO.ONO ON mmuosoemom
OO ANO.ONO ON HOO.NHO HH OO HOO.OHO ON ANO.OHO OH Omeemmum
Hmuoe mCOZ OHOE no N Hmuoa mCoz OHOE no N Hoosum
CH mmMHo
mquEmumum huHHHQwHH mquEmume ummm¢
% .mquEmeum muHHHQmHH pCm ummmm m>HHmEHOC mCwaE memE m0 HmHECC pCm HOOCUm CH mmMHU .g OHQMB
OO. A C A OH. O .0.0 Oe.s u NH OOO. A C A HO. O .0.0 HO.NH u NH
eON OOH OO OOH NO OO Hmpoe
OO HOH.ONO ON HOO.ONO ON OO ANH.ONO OH HOO.OHV NN muoHcmm
Ne Ass.NeO we HON.ONO ON OO HON.eNO ON ANN.OHO AH muoHasO
OO AOO.OOO em HHO.ONO em Ow HOH.ONO 5N ANO.ONO NN mmnosoemom
es HOH.ONO HO HOO.OHO OH OO AOO.OHO ON «ANO.OHO N cmsemmum
Hoosum
Hmuoe mCoz OHOE Ho H HMDOB mCoz OHOE no N CH mmMHo
*emquEmHmum muHHHQMHH mquEmumum ummmm
.mqufimumum muHHHQMHH pCm Hemmm HHmeIHMHUOm mCmeE memE mo HOQECC UCm HOOCom CH.mmMHU .m mHQMB

34

is concluded that males as freshmen do not tend to evaluate
themselves according to either SS or Normative values but
that they do so evaluate themselves by the time they are
seniors.

The fact that there is a significant increase in
the number of individuals evaluating themselves by
Normative standards rather than the anticipated decrease
requires a modification of the interpretation of the meaning
of this category. One possible interpretation of this result
is that standards of behavior and conduct are acquired con-
currently with school attendance. Hence the category may
represent abstract, general criteria for behavior that have
been derived from the greater exposure to ideas and diverse
kinds of people encountered in school, and not from the
accumulated teachings of home and church as had been
previously supposed. Evidence from Hypothesis IVa reported
at a later point, which shows that the Normative category
is not associated with stated importance of religion,
further supports this interpretation.

There is a parallel increase in self-assessments
for hgth assets Egg liabilities for the Normative and SS
categories and this would support the View that these
evaluations are being made in terms of certain standard
values, since the incorporation of values should reveal
itself both in favorable Egg in unfavorable evaluations.

There remains the problem of accounting for the

correspondence between the theoretical and actual frequencies

35

of evaluation in these two categories for assets and liabili-
ties at the sophomore and junior grade levels. While there
is insufficient information to determine what sort of pro—
cesses may be involved, three possibilities can be con-
sidered. First, it is possible that the Chi Square test
is insufficiently sensitive to changes; computing the mean
proportion of statements for each class might have been more
sensitive, although as indicated previously, such an
approach would have raised some formidable difficulties.
Second, the apparent leveling off of the rate of change in
evaluations might be the result of some complex interaction
between differential rates of change in self-evaluations
for asset and liability statements. Thus, if self-evaluations
according to some set of values would result primarily in
liability assessments, then it is possible that there would
be less of a tendency to evaluate oneself in terms of
such values until such time as asset statements would be
forthcoming as well. Finally, it is possible that a change
from no or few evaluative statements in one category to many
statements in that category involves a period of transition
during which time the values are learned only imperfectly,
and, hence, during which evaluations in terms of these
values might have been improperly coded.

The foregoing interpretations suffer from the fact
that they assume chronological processes when the data are
not longitudinal. Consequently the possibility that these

findings are artifactual cannot be entirely ruled out. In

36

addition, this interpretation would not be complete until
all of the other categories of self-evaluation have been
analyzed in a similar manner——a task which was too extensive
to be undertaken in this thesis.

One source of artifactual contamination can be ruled
out. On the possibility that variations in the frequency
of Normative and of SS statements could be explained on the
basis of the total response frequency, a "best" median
split of the total number of asset and the total number of
liability statements (including all categories of classifi-
cation) was run against grade level in school. No signifi—
cant differences were found either for total number of asset
statements or for total number of liability statements.

The possibility that the total number of responses
might be associated with any of the asset or liability
categories examined was similarly tested for all subsequent
hypotheses. This possibility was rejected in all cases.

In general, then, the results are interpreted as
indicating an association between one‘s mode of evaluation
on the ALI and a certain connection with the social
structure. For males, increasing involvement in college,
as represented by increasing grade level, was associated
with greater tendency to evaluate oneself according to
Social Skill and Normative categories. The Normative
category is reinterpreted as possibly indicating the presence

of abstract norms of behavior which are not necessarily

37

confined to early socialization, but which may result
from contact with other reference groups (in the college

perhaps) and which may increase with wider experience.

Hypothesis II proposed that:

Males will make more role and job asset and
liability statements than will females.

The data in Table 5 below do . not permit the

acceptance of the hypothesis.

Table 5. Number of role and job skill asset and liability
statements by sex.

 

 

Role and Job Skill Statements

 

Assets Liabilities

 

 

 

 

 

Sex 1 or more None Total 1 or more None Total
Males 55 197 252 6 246 252
Females 24 96 120 7 113 120

Total 79 293 372 13 359 372

X2 n.s. X2 n.s.

 

The most likely interpretation of these results is
that the meaning of the Job or Role Skill category upon
which the hypothesis was derived is incorrect. A re-
examination of the responses which were coded into this
category do, indeed, confirm this;2 the kind of behavioral

and social structural context to which these responses refer

 

2Some examples include, "good at sports," "manual
skills," "good musician,” etc.

38

are at once so specific and diverse that they cannot be
regarded as relating specifically to sex roles as had been
thought. In short, in view of the kinds of responses which
were actually included into this category the original
hypothesis appears to be manifestly bad. This raises, in
addition, the question of just what it is that all of these
responses have in common which would justify their inclusion
into the same category. Knowing that they all refer to some
specific skill in a specific kind of setting hardly seems

a sufficient basis for inferring an additional general
property which they exhibit in common. If there is in

fact such a general property it is certainly not given
either from the theory or from the nature of the responses.
It may well be that for subjects who are more directly
involved in the actual business of earning a living, or who
are closer to it than is a sample comprised predominantly

of lower division college students, this particular category
might become more dominant and reveal the expected pattern.
In any case, there is no relationship between self-assessments
of job or role skill and differentiation according to sex

roles.

Hypotheses 111a and IIIb proposed respectively that:

Higher grade point averages will be associated
with those individuals who make Intellectual or
Scholastic asset statements as opposed to those
who make none.

Lower grade point averages will be associated with
those individuals who make Intellectual or Scholastic
liability statements as opposed to those who make none.

The data, presented in tables 6 and 7 permit us

39

to accept both of these hypotheses.

Table 6.

Grade point average by number of Intellectual,
Scholastic asset statements.

 

 

Number of intellectual

Grade Point Average*

 

 

 

 

scholastic asset Above Below
statements median median Total
2 or more 18 10 28
(14.48) (13.52)
One 91 68 159
(82.23) (76.77)
None 72 91 163
(89.29) (78.71)
Total 181 169 350
X2 = 7.416, df = 2,p<.025
Table 7. Grade point average by number of Intellectual

Scholastic liability statements.

 

 

Grade Point Average
Number of intellectual

 

 

 

scholastic liability Above Below
statements median median Total
One or more 25 47 72
(37.23) (34.77)
None 156 122 278
(143.77) (134.23)
Total 181 169 350

 

X2 = 10.47, df = 1, p < .005
*Because grade point average was originally compiled
using grouped data it was impossible to compute the average
grade point for the different number of statements made. As
a result, a less powerful best median break was utilized.
The median grade point average was 2.50.

40

In contrast to the preceding result in Hypothesis
II the Intellectual or Scholastic statements do seem to
share a common property which is relevant to a context
from which such statements might have been drawn or to a
certain kind of behavior. Thus, they all refer to mental
ability and all respondents are involved in a context in
which evidence to this effect is forthcoming.

Although the hypotheses are born out, it is a moot
point as to whether performance (and, more generally, whether
behavior) is the result of self-conception or whether the
self-conception is the result of performance. Symbolic
interaction theory would accept both as true together, and

in the final chapter a rationale for this view is elaborated.

Hypothesis IV proposed that,

Those who perceive their religion to be important

in the way in which they live their lives will

make more Normative asset and liability statements

than will those for whom religion is of little or

no importance.

The data are presented in Tables 8 and 9 and do not
permit the hypothesis to be accepted. Catholics and
Protestants were analyzed separately on the chance that
parochial instruction might have final to more marked effects
for Catholics, although there were no measures of such
separate instruction.

Neither for Catholics nor for Protestants is there
a significant departure from expected frequencies on either

assets or liabilities. This finding, in addition to that

presented in the previous discussion of Hypothesis Ib,

41

Table 8. Number of Protestants and Catholics making 0 and
2 or more NOrmative asset statements by importance
of religion.

 

 

Number of Normative Asset Statements

 

 

 

 

 

Importance of Catholics Protestants
Religion 0 2 or more Total 0 2 or more Total
Important 29 38 67 37 65 102
Not Important 5 5 10 21 29 50
Total 34 43 77 58 94
X2 n.s. X2 n.s.

 

Table 9. Number of Protestants and Catholics making 0 and
2 or more Normative liability statements by
importance of religion.

 

 

Number of Normative Liability Statements

 

 

 

 

 

Catholics Protestants
Importance of
Religion 0 2 or more Total 0 2 or more Total
Important 33 36 69 61 47 108
Net Important 7 7 14 25 33 58
Total 40 43 83 86 80 166
X2 .n.s. X2 n.s.

 

requires a modification in our understanding of the Normative
category of evaluation. Responses which fell into this
category were believed to be the result of self-evaluations
made in terms of abstract, moralistic values. It was felt
that such values would most likely to imparted within a
context providing for religious instruction, and that

those who conceived of themselves as having their lives

42

governed by their religion would reveal the internalization
of such precepts as Normative, Imperative, or Directive
statements on a self—assessment.

It now appears that such evaluative statements
are independent of religious commitment (as we have measured
it) and are more readily interpretable as the product of
such experiences as college, which may foster the formation
of some sort of personal credo or ideology of behavior
according to which they act and evaluate themselves.
HOwever, the data do not permit us to test the extent to

which such values actually do influence behavior.

Hypothesis V proposed that,
Those who belong to more organizations in school
will make more Social Skill asset and liability

statements than those individuals who belong to
few or no organizations.

Respondents at all schools except Michigan State,3
were asked "How many organizations, clubs, or groups have
you belonged to while attending college?" Since answers
varied with the year in school, respondents were divided
into "many" and "few" membership groups separately for each
class in school so that these groups came closest to the
best median break for that class. All classes were then
combined for the analysis.

From Table 10 it is evident that the hypothesis is

only partially supported. Only Social Skill asset evaluative

statements are associated with membership in more organizations.

 

3This question was added to the questionnaire after
its initial administrat1on to Michigan State subjects.

43

Table 10. Relationship between number of Social Skill
asset and liability statements and organizational
membership.*

 

 

Skill in Social Relationship Statements

 

 

 

 

 

Member-

ship in Assets Liabilities

Organi-

zations 2 or more None Total 2 or more None Total

Many 47 38 85 19 66 85
(39.19) (45.81) (21.25) (63.75)

Few 30 52 82 23 60 82
(37.81) (44.19) (21.25) (62.25)

Total 77 90 167 42 126 167
X2 = 5.88 d.f. 1 p (.02 X2 n.s.

 

*Graduate students were excluded and data for MSU
were unavailable.
The present evidence might suggest that individuals may only
evaluate themselves in a manner appropriate to their
involvement within the social structure when such an
evaluation results in a relatively favorable self-
conception. Yet, such an interpretation would be contrary
to other findings already reported, e.g., Hypotheses I and
III, that if individuals are evaluating themselves according
to certain values they will evaluate themselves by these
values 29th favorably apg unfavorably. An interpretation
more consistent with the preceding remarks might be that
individuals may be inhibited from becoming involved in the
kinds of situations where experiences with reference groups
may confirm or reinforce an unfavorable evaluation of
themselves. A sufficient number of such cases would tend

to wash out any trend for favorable and unfavorable evaluations

44

to be made simultaneously in a given category.

Hypothesis V then gives only partial support to the
view that the ALI links the self to reference groups and to
possible behavior by virtue of SS evaluative statements.

A more meaningful test of the category would have been to
ask respondents to list the groups to which they belonged,
the amount and nature of their participation in each,

and the importance to them of the views of these groups.

The establishment of such clear cut reference groups and
degree of behavioral involvement would at least have
clarified the interpretation of the results of the above
findings. What ats_the effects of a disproportionate amount
of liability statements over asset statements with respect

to the seeking out or the avoidance of interaction, etc.?

Hypothesis VI proposed that,

Self-esteem as measured by the ratio of asset to

liability statements will increase with socio—

economic level.

The assumption lying behind this hypothesis was
that the higher the socioeconomic level the greater would
be the opportunities to acquire the skills and to be exposed
to experiences that would lead to successes and hence to
favorable evaluations in a school situation.

Table 11 indicates that the hypothesis cannot be
accepted. Since evidence will subsequently be presented in

connection with hypothesis VII that supports the validity

of the index of self-esteem we can tentatively rule that

45

out as a factor which might have yielded these results.

Table 11. Number of individuals above and below the median
self-esteem score by socio—economic level.

 

 

Self—esteem

 

 

 

Above Below
median median Total
Elite 27 34 61
Middle-Middle 19 22 41
Working Class Elite 87 63 150

Respectable Working
Class 17 20 37
Low Status 35 32 67
Total 185 171 356
X2 n.s.

 

One explanation for these findings is that individuals
with low self-esteem and low socioeconomic levels do not go
to college, but if this were generally the case then we
would expect the low status group to be relatively higher
in self—esteem than the other groups-—which it is not.
It may be that such socioeconomic level has been about
equally able to find a context in which suitable levels of
self-esteem could be maintained; research on school sub—

cultures suggests that this could very well be the case.4

 

4David Gottlieb and Benjamin Hodgkins, "College
Student Subcultures: Their Structure and Characteristics
in Relation to Student Attitude Change," The School Review,
Vol. 71: PP. 266-289.

 

46

Hypothesis VII proposed that,

There will be a positive correlation between self-
esteem and perceived agreement with significant
others on self-evaluations.

Results are presented in Table 12 and permit acceptance

of the hypothesis.

Table 12. Self-esteem and perceived agreement with signifi-
cant others on self-evaluation.

 

 

Mean Agreement

with Significant Self-esteem

 

 

Others High Low Total
High 94 73 167
Low 69 94 163

Total 163 167 330

X2 = 6.42, df. 1, p < .01

 

Acceptance of the reasoning underlying the hypothesis
is not such a simple matter. On the one hand, there are
indications that individuals evaluate themselves in accordance
with the actual and perceived evaluations of others5 while,
on the other hand, there is evidence that there is distortion
or selective perception involved in estimates of the
evaluations of one's self by others particularly for those

with high self-esteem.6 Thus, the data not only support

 

5See for example, S. Frank NUyamoto and Sanford
Dornbusch, "A Test of Interactionist Hypotheses of Self-
Conception," American Journal of Sociology, Vol. 61 (1956),

6Ezra Statland, Stanley Thorely, Edwin Thomas,

Arthur R. Cohen, and Alvin Zander, "The Effects of Group
Expectations and Self—Esteem upon Self-Evaluations," J. of

47

the assumptions of the hypothesis, but also support a
hypothesis which has some very different implications for
the interpretation of human behavior. For example, the
implications of an assumption that people perceive them—
selves favorably because they perceive nearly everything
about themselves favorably, including the reactions of
others, are quite different from those leading from the
assumption that people, as a result of behaving in accordance
with the expectations of others, receive cues confirming
self—evaluations. Although there is no way to resolve
this with the present data, it is possible to suggest a
rapprochement.

There seems to be no reason why the two processes
could not be operating simultaneously. Thus, we can con-
ceive of individuals as sharing the values of others with
whom they are in contact so that their behavior is favorably
confirmed by these others. Such individuals would be likely
to assume that most other individuals in groups with whom
they were in contact would also evaluate them as they
evaluated themselves--leading to distortions of the
imputed evaluation of others toward oneself. Those who had
a rather low self-esteem would, on the other hand, be con—

cerned with, and hence more vulnerable and sensitive to, the

 

Abnormal and Social Psychology, Vol. 54 (1957), pp. 55-63; and
Arthur R. Cohen, "Some Implications of Self-Esteem for

Social Influence," in Hovland and Janis (eds.), Personality
and Persuasibility (New Haven: Yale University Press, 1959),
pp. 102-109.

 

 

48

objective clues from others with whom they happened to be

in contact.

Discussion and Critique

On the basis of the results and the discussion so
far, six major criticisms may be leveled against the ALI.
We will first consider three which are directly related
to an interpretation of the foregoing results, and then
three more which are more directly relevant to some of the
technical characteristics of the ALI as a research device.
It will be noted that some of these criticisms have been
explicitly anticipated in preceding remarks while others
have not. Of the nine hypotheses and sub-hypotheses discussed
above Hypotheses IIIa and b, and VII were completely con-
firmed, Hypotheses Ia, Ib, and V were partially confirmed
if the reinterpretation and speculation with respect to them
are accepted, and Hypotheses II, IV, and VI were disconfirmed.
In Chapter I it was pointed out that the hypotheses
which had been selected were varied enough so that the results,
when taken as a whole, should indicate whether or not the
ALI did link self-conceptions to involvements in the social
structure and to behavior. However, in view of the mixed
support that the hypotheses have received it is difficult
to give an unequivocal overall assessment of the effective—
ness and validity of the instrument. If one accepts as
valid those reinterpretations which were given to the

meanings of the content categories and the other speculative

49

explanations offered in the course of the analysis, then,
in a very general way, it does appear that the instrument
reflects certain kinds of involvement in the social structure
and is related to certain kinds of behavior. Just what kinds
of involvement and just how this is reflected is difficult
if not impossible to specify for four interrelated reasons.
1. The extent and nature of the involvement within
the social structure is measured quite crudely by control
variables such as, socio—economic level, number of organi-
zational memberships, grade level in school, etc. This is
true especially in view of the fact that this is a crucially
important point in the link which we are endeavoring to
establish between actors, social—structure, and behavior via
the symbols of self-representation. From the standpoint
of Kuhn‘s statement, such involvement is the context trgm
whigh_derive the symbols used in the self-assessment, and
to which the symbols are to link the actor, and in which
behavior occurs.
2. The basic assumption behind the ALI was that by

eliciting self-representations that were evaluations,

 

various connections between actors, social—structural involve-
ment and behavior would be clearly revealed. That they are
not clearly revealed is apparent from the two preceding

chapters; the basic reason for this is that evaluations

 

by their very nature obscure rather than clarify these
connections. The theoretical formulation developed in

the following chapter will supply the reasoning behind this

apparent contradiction.

50

3. Related to "2" is the fact that meanings have been
imputed to responses on the ALI with virtually no knowledge
of the accuracy with which this has been done. Presumably,
of course, the validity of the coding procedure was to be
determined from the results of the hypotheses, the hypotheses
being derived in part from the imputed meanings of responses
in the various categories. The circularity which follows
from trying to establish the validity of the categories at
the same time that the instrument is being validated is
readily apparent.

4. The weaknesses discussed in "2" and "3" are made
more serious and make more serious the following criticism.
As noted in Chapter I no precise theoretical rationale is
available for linking particular categories of self-
assessment to either involvement in the social structure,
to reference groups, or to behavior. That is, while the
theory is explicit in asserting a connection between self—
and social structural involvement, and behavior at an abstract
level, it is notably vague with respect to particular
instances of these connections. This weakness is largely
responsible for the difficulty that was had in showing
that the hypotheses logically followed from the meaning
attributed to a category.

When these four factors are viewed in combination,
there remains little of an established standpoint from
which interpretive judgments can be made. We are placed

in the position of saying with some conviction that

51

"something is going on," but are at a loss to define and
evaluate precisely what that may be. This is best seen

by contrasting the results forthcoming when, (a) these four
criticisms are minimized, i.e., when the content category
from the ALI and the control variable were relatively con-
crete and clearly defined, as in the case of Hypothesis II
which related grade point averages to mental or scholastic
evaluative statements: and (b) when these same criticisms
are most applicable--that is, when the meanings of the
content category and the control variable are vague, as in
Hypothesis V which related Social Skill statements to organi-
zational membership.

But what is the likelihood that these criticisms
could be eliminated by any or all of the following measures:
1. A more precise specification and measurement of

involvement in the social structure (or of reference
groups), or of behavior?
2. Supplementary questions which would clarify the
basis upon which the self-evaluations were made?
3. Supplementary questions which would clarify the
meanings of the responses for the subjects?
4. An elaboration of the theory which would permit
more precise links between ALI responses, reference
groups, and behavior?
Without elaborating at this point, let us only suggest that
for reasons to be presented in Chapter III only the first

of these remedial measures could be successful.

52

Three additional criticisms of the ALI are concerned
mainly with the "mechanical" difficulties associated
with it as a research technique.

5. There are many possible ways of constructing
categories for the ALI responses as is suggested by "2,"
"3," and "4" above, and unless alternative methods are
developed and tested there is no way of knowing which would
yield the best results. There is a possibility that results
vary with the obviousness with which a given response can
be associated with some category on the ALI and the
obviousness with which such a category can be associated
with behavior or social structural involvements. Thus,
Intellectual or Scholastic statements are rather clear cut
as to what they mean and their extension to a context of
scholastic achievement, e.g., grades, obvious. This seems
to occur at just that point where the category is also most
narrowly defined and of least general applicability.
Disturbingly, this may indicate that the ALI works best
where the links between actor, social-structure, and
behavior are most obvious and trivial; that is, at just
that point where it is needed least. For surely the results
obtained in Hypothesis III could have been gotten far
more easily by a straight-forward question to the effect
of "wa much scholastic ability do you have?" There is
more than just a strong hint of vacuousness in finding

obvious connections between obvious measures.

53

6. The present system of classification yields 18
separate categories for assets and liabilities and unless
this number could be drastically cut the system is an
extremely cumbersome and unwieldly one with which to work.
It is not yet clear whether one category is properly inter-
preted as remaining the same for assets and for liability
statements. No attempt has yet been made to determine
whether treating these separately or in combination would
be the best research startegy. Results from Hypothesis V
suggest that until this kind of an investigation would be
undertaken any combining of asset and liability categories
would be premature.

7. As previously noted in Chapter I it is somewhat
problematical whether the scoring procedure should be of
the Proportional or Absolute method. No attempt has been
made to test the effectiveness of one approach over another,
let alone explore the consequences of the previously noted
troublesome characteristics of each.

In view of the fact that from the standpoint of
the position to be developed in Chapter III criticisms "2"
through "4" will be shown to be attributable to a judgmental
flaw in the rationale behind the development of the ALI, the
very great effort that would be needed to resolve "5," "6,"
and "7" would be best spent developing techniques more
consistent with the theoretical positions presented in

the following chapter.

CHAPTER III

Introduction

It was noted in Chapter I that attention has been
directed toward studying and measuring the self in the
belief that it provides a uniquely important link between
an actor's social-structural involvement and his behavior.

It was proposed that a measure of the self which included

an explicit self—evaluation would yield results that would
be particularly useful in this respect. Hewever, a number
of significant problems (listed at the end of Chapter II)
were found to be associated with the ALI and particularly
with interpretations of the results from it. These dif-
ficulties precipitated a re-examination of the rationale
behind the assumption that the self, as measured by techniques
like the TST and the ALI, was as important as originally
believed. The theoretical position arising from this
re-examination is developed in the first part of the present
chapter. It attempts to reassert the social and behavioral
importance of the symbolically represented self by

focusing on what is here called the relationality or the
relational prpperty of symbols. This position is then

used as a basis for reaffirming the theoretical and empirical

importance of the symbolic self, and for an analysis and

54

55

explanation of the problems encountered in using the ALI.
Some suggestions for further development of the theoretical
position, and for the construction of instruments that are

consistent with this line of thinking are also given.

The Theoretical Restatement

In Mead’s terminology symbols have the capacity to
arouse attitudes—-tendencies to behave or imagery of contact--
toward the object symbolized:l"an organization of attitudes
with reference to what we term objects is what constitutes
for us the meaning of things."2 Things are meaningful, then,
in terms of (a) an actor's behavioral tendencies, or (b) imagery
of behavior with respect to some aspect of the environment.
Relationality simply refers to the existence of such
behavioral relationships between a class of actors and a
class of objects.3 These behavioral relationships are
thought of as being a complex seggence of actor-object
relationships. Consequently, the full meaning of a symbol
includes all of the potential behavioral relationships

between an actor and objects which might occur in a sequence

 

1George H. Mead, The Philosophy of the Act (Chicago,
Ill.: The University of Chicago Press, 1938), p. 222.

2George H. Mead, MindL Self and Society (Chicago,
Ill.: The University of Chicago Press, 1934): P. 125.

3Throughout this paper it will be convenient to
understand "object" as referring to any person, behavior,
thing, or situation, or to any combination of these which
have been isolated as units and given names.

56

or in different situations, either as actual behavior or
as this behavior can be known through reflective thought.
The relationships inherent in symbols serve as "recipes"
or "formulae" accordingyto which behavior is governed,
guided, and shaped. We are not here asserting a doctrine of
of symbolic determinism, for "unpacking" the entflgE\meaning
inherent in a symbol would reveal that behavior could
materialize in a variety of alternative ways contingent
upon who an actor might be and on what might transpire
during the course of behavior, and yet the behavior would
remain consistent with the relationships inherent in the
meaning of that symbol.4
A formidable difficulty arises in explicating the
concept of relationality: the thought patterns and the
language in which we are necessarily forced to communicate
implies a unique existence to physical objects. "Object"
and "actor" are somehow felt to be isolated things, indepen-

dent of and more substantial than any relationship which

they might assume. waever, it is against precisely this

 

4It might be argued that a symbol which included many
contingencies, which permitted numerous behaviors in a
variety of circumstances, all of which were consistent with
the meaning of the symbol, is prima faci evidence that the
symbol is so vague as to be without meaning. That this is
not necessarily or even generally the case can best be
illustrated by an analogy to chess. In chess the meaning of
the various pieces can be conveyed precisely and exhaustively
in terms of rules which specify the relationships which can
obtain between the chessmen and the squares of the board.
The game itself, however, like all human behavior, is
infinitely varied and complex.

57

notion that we are arguing, for "object" and "actor" are
mutually implied by one another in terms of behavioral
relationships. In conventional speech it is apparently a
grammatical or conceptual convenience to focus upon one
portion of actual relationships and give names to actors,
objects and relationships as if there were no connection
between them.5 The fact that symbols are used in this manner
should not be allowed to obscure the fact that in order to
understand the meaning of any symbol, regardless of whether

it mamas an actor, an object, or a relationship, it is
necessary to refer symbolically to each of the others.

For example, it is clear that the meaning of symbols which

name relationships, e.g., "trip," "coctail party," "conference,"
"fight," or "work," can only be conveyed to others by reference
to other symbols, thereby indicating what some class or sub—

class of actors do in relationship to some class or sub-

class of objects. We can only partially cope with these

 

5This may only be a handicap of Indo-European language.
For example, contrast the usual mode of speaking of behavior
in English with the Navaho. "The Navaho speaks of 'actors'
and 'goals' (the terms are inappropriate to Navaho), not
as performers of actions or as ones upon whom actions are
performed, as in English, but as entities linked to actions
already defined in part as pertaining especially to classes
of beings. . . [In speaking the Navaho links] individuals
to actions and movements distinguished, not only as actions
and movements, but as well in terms of the entities in
action or movement. This division of nature into classes
of entity in action or movement is the universe that is
given; the behavior of human beings or of any being indivi—
duated from the mass is customarily reported by assignment to
one or other of these given divisions." Harry Hoijer, "The
Sapir—Whorf Hypothesis," in Readings in Anthropology, Vol. I,
(New YOrk: Thomas Y. Crowell, 1959), pp. 219-231.

58

semantic limitations by stressing that when "object,"
"actor," or "relationship" are used in this essay they
should be understood to be an incomplete part of the tri-
polar relationship--they are only grammatically, and hgt
analytically, separate.

The shared meanings of a group's symbols both reflect
and lead to patterned behavior on the part of the members
of that group. When the behavioral relationships inherent
within the meaning of a symbol are thus shared and when they
constitute a common organizing principal of behavior toward
the environment for a number of individuals, then that
symbol can be said to be a hgrm. Frequently, relatively
stable sets of behavioral relationships between classes of
actors and classes of objects acquire names; when this
occurs they can be regarded as rglgs, e.g., father, student,
president, or as situations, e.g., dinner, examination,
conference.

Inasmuch as the extensions6 of a symbol are to
classes or relationships, actors, or objects, it is possible
to conceive of symbols, including norms and roles and
situations, as differing in the specificity or generality
of the classes of which they are true. For example, the

extension of the symbol "house" is more inclusive with

 

6The concept of extensions is borrowed from logic
where it is used to indicate the universe of which terms
or predicates are true.

59

respect to kinds of actors and relationships than is the
symbol "dog house." And, in general, symbols differ in the
degree to which the behavior which they imply is specified
or uniquely determined for a given universe of actors.

In this thesis the theoretical importance of the
concept of relationality lies in its application to particular
kinds of symbols designated as self-concepts. If an actor
can be said to have meaning for himself in the same way
that any other symbol has meaning for him then all that has
been said with respect to the relationality of a symbol
applies as well when it is oneself that is symbolized.

Once again the vagaries of our language force the impression
that the self must be a concrete object; however, response
protocols from instruments like the TST suggest that the
self is more appropriately regarded as a cluster of symbols

each of which defines complex relationships between a

 

class of actors and a class of objects in particular societal
contexts. From the foregoing discussion it should now be
clear that very different empirical implications follow

from theoretical commitments to self-concept conceived in
terms of meanings and relationships than follow from a
self—concept which is seen as an entity with the attendant
connotations of isolation and uniqueness. In the evaluation
of the ALI which appears at a later point in this chapter
these implicatiOns become directly relevant.

Some examples of responses on the Twenty Statements

Test make the relational property of the self more apparent.

60

As noted above, symbols will vary with respect to the generality
of the classes of actors, objects, and relationships which

are their extension. This factor may be observed in the
self-definitions of the TST protocols. For example,

responses such as "one with red hair," "21 years old," seem

to be so general with respect to the classes of actors or
objects of which they might be true that the behavioral
relationships which would be implied by these symbols would

be largely indeterminate, that is, they do not have specific

or fixed meanings. On the other hand, "friend," "wife,"

"doctor,' and "night club performer,‘ seem to be increasingly
specific with respect to classes of actors and classes of
objects and consequently, increasingly specific with respect
to the behavioral relationships which could exist between
them. Such responses suggest shared meanings and relation—
ships; that is, they imply norms and roles. Thus, in these
responses there is an implied social context in which the
meanings of these symbols are shared, in which they emerge,
and in which they govern behavior. Such self-designations,
besides specifying, i.e., naming, a class of actors alsg
imply socially defined ways of behaving toward specific
categories of individuals, objects, or groups, in certain
specific kinds of situations. In short, an individual's
potential behavior is glyph in the meanings of the symbols

by which he chooses to designate himself.7

 

7This position on the self appears to be quite

close to that advanced by Sherif and Sherif. These authors
conceive of the self as a subsystem of interrelated attitudes

61

With the foregoing in mind, the self can be defined
as the symbolic representation of a set of relatively
stable relationships to the environment which an actor
experiences as "being" himself-—the self is the sum of
the meanings which he has for himself. One reason for
believing that the self-concept is an important area of
concern for social-psychology is simply that it is, in the
present view, the locus of the meanings whereby an actor
establishes behavioral relationships with his environment.
These symbols should reveal, then, the different ways in
which individuals are engaged with other persons, groups
or events. Furthermore, a knowledge of the symbols by which
an individual represents himself to himself should yield
important information about the context in which he has
established these behavioral relationships. From the
properties of the social-structural context additional
inferences about the actor and his behavioral relationships
might be drawn. For, if the objects, i.e., the people,
situations, and things, in relation to which the actor de-
rives his meanings can in turn be seen to change or to be
organized in certain ways, then much more of the likely

behavior of the actor could be known and explained. For

 

which an actor has toward himself, others, groups and
institutions, and values and "which define and regulate

his relatedness to them in concrete situations." Muzafer
Sherif and Carolyn W. Sherif, An Outline of Social Psychology
(New York: Harper and Brothers, 1956), p. 581.

62

example, it is quite possible that self-conceptions of the
kind yielded by the TST can be arranged according to the
degree to which the relationships which they suggest are
components of formal or informal structures; knowing this
would be to know much of the structure and organization of
the objects to which an actor would or would not be
addressing himself. The meanings attached to self-symbols
differ then in the degree to which they are fixed in
formally constituted group processes, in the degree to
which they imply particular settings or contexts, and in the
degree to which they permit the establishment of behavioral
relationships in different kinds of situations.

It is assumed that the extent and form of the be—
havioral relationships which may constitute the meaning of
any particular self-concept for an actor are derived from
the norms, values, and beliefs of groups which are important
to him. These may vary in their lack of contradiction,
the accuracy with which they are perceived, the degree of
consensus with which they are held, and the strength of
any sanctions involved in violation or compliance with
them, as well as in numerous other ways. Important as these
factors might be in determining the nature of the relation—
ship between a self-conception of some particular sort and
actual behavior, they cannot be dealt with in this essay.

Also to be excluded is any discussion of the precise
mechanisms involved in the acquisition of self-conceptions.

For this thesis it suffices to make the following three

63

related assumptions relevant to this point: First, self-
concepts are formed from the store of symbols available
in the groups with which actors are in contact. These
symbols represent any of the behavioral relationships which
have acquired meanings in these groups. Second, individuals
behave toward an actor according to the meanings which he
has for them, and, ultimately, according to the relationships
which he affects with them or according to the identity they
have assigned to him. Third, individuals acquire meaning
for themselves as "selves" from the nature of the relation-
ships which they establish with their environment, parti-
cularly from cues in the actions and behavior of other
individuals around them. Fourth, the greater the amount of
time spent in some location within the social structure the
greater will be the likelihood of forming a self-concept
from the symbols at that location in accordance with
'assumption three, in particular, the greater is the likeli-
hood that an aspect of the self-concept will be linked to
a normative reference group associated with that location.

A fifth assumption summarizes and condenses much
of what has already been said with respect to the self and
actual behavior: to some important extent the meanings
which the world has for an actor derive from experienced
relationships to it by virtue of the relational property of
his self—symbols--since behavior will be consistent with

the meanings it will also tend to be consistent with the self.

64

This last assumption suggests the existence of an
evaluative dimension to the self. Judgments as to the
consistenoy of actual behavior with behavior implied by the
meanings of a self—concept implicitly involve a comparison
by the actor of his actual behavior against certain criteria
or standard relationships—-in short, it entails an evaluation
according to relevant normative standards by means of symbolic
behavior.

It would not be inconsistent with the view that the
experience of social reality is largely a process of experienced
relationships to assert that for social groups certain
actor-object relationships serve as standards and represent
different degrees of, for example, goodness, value, sacred-
ness, or appropriateness. These patterns of ordered relation—
ships may be thought of as shared internalized evaluative
scales, while values, as distinct from an evaluation, may
then be regarded as relatively favorable locations on such
a scale. The symbols "good," "immoral,” "funny,” ”honest,"
or "shocking" which result from an evaluation convey meaning
to others; that is, they are significant symbols and rgprg-
sent the relativeppositions between whatever relationships
that have been evaluated and those behavioral relationships
that may comprise comparativetpoints on the shared scale of
reference.

The actor's experience of the outcomes of these
evaluative acts with respect to his self-concept constitutes

his self-esteem. Just as the self was regarded as a set of

65

relationships so also can self-esteem be regarded as a
set of evaluations of these relationships. Consequently,
it is possible to have as many specific self-esteems as
there are somewhat distinct relationships comprising the
self. It is possible, too, that there is a generalized
self-esteem as the composite of all such evaluations.
Self-esteem at any one time, then, might consist of any
number of experienced discrepancies or convergencies between
actual and ideal relationships.

A graphical but over-simplified representation of
self-evaluations and self-esteem may be helpful in conveying

these concpets and their connection to self-symbols.

Figure 3. Graphical representation of self—evaluation.

Ideal Relationship

‘(/'~‘\b Normative Evaluative

Scale

Actor

 

 

IIHIHH

Actual or Experienced
Relationship

In Figure 3, three relationships relevant to the meaning of

a given self-symbol, S, are represented by straight lines.

Ri represents an ideal behavioral relationship as given by

the meaning of S, Ra is an experienced behavioral relationship
relevant to S, and R.e is the symbolically experienced
discrepancy between the ideal and the actual relationships.
This discrepancy is experienced as a level of self-esteem,

and the comparison of the two relationships Ri and Ra is

66

an evaluation. The meaning of this and other possible
discrepancies consist of relationships between the ordered
relationships which comprise points on the normative evalua-
tive scale appropriate to S. From this example it can be
seen that self—evaluations have a more complex meaning
structure than do non-evaluative self-symbols; a self-
evaluation is in fact a relationship of relationships.

The significance of this distinction between Self-symbols
and self—evaluations will be pointed out later in this
chapter in connection with the critique of the ALI.

One theoretical convenience of thinking of the self
in terms of itsesteem—dimension is that it suggests that
changes in self-concept can possibly be interpreted in
terms of fairly specific propositions of level of aspiration
theory. Should this be true then it would have implications
for the interpretation of the self as a meaning. For,
changes in the relative positions of an ideal and an actual
self, by shifts of either one along some scale, would also,
by definition, result in changed relationships between the
actor and his symbolic environment.

In speaking ofnormative comparative scales we are
simply indicating that such scales can be reconstructed from

8

the behavior of individuals. It does not necessarily imply

 

8The recent work of Clyde Coombs describes theoretical
and methodological techniques of great promise for investi-
gating the existence and the characteristics of such scales.
See Clyde H. Coombs, A Theory of Data (New York: John Wiley
& Sons, Inc., 1964).

67

any awareness of scales as such on the part of those so
behaving. That individuals can choose between alternatives
with some consistency is sufficient justification to attempt
to use scales as an explanatory device.

The concept of reference group offers one possibility
for tying the relational property of the self, along with
the interpretation of norms as scales, back into the social
structure. Groups to which an individual compares himself—-
comparative reference groups~qnuigroups from Which an indi-
vidual has adopted norms or values—-normative reference
groups--have been elaborated by Merton and Rossi.9 Within
the framework which is being employed, normative reference
groups can be regarded as those groups which provide the
scales governing the appropriateness of various relation—
ships which actors affect with their environments, and which
are a major factor in the development and maintenance of the
self-concept. Comparative reference groups can be viewed
as occupying important points of comparison along one kind
of normative evaluative scale.

In summary then, location and involvement in the
social structure affect an actor's self-concept through

reference groups, which in turn provide the behavioral or

 

9Robert K. Merton and Alice S. Rossi, "Contributions
to the Theory of Reference Group Behavior,” in Robert K.
Merton, Social Theory and Social Structure (revised and
enlarged; Glencoe, Illinois: The Free Press, 1957), Chapter
VII.

68

symbolic settings and conditions under which an actor
establishes stable relationships with his environment,
which then become symbolized as a self-concept. They
provide normative depth to the behavior associated with any
self, and they may provide important points of reference in
relation to which the self is evaluated. Second, self-
concepts influence behavior since, to a large extent, they
select and give meaning to certain portions of the environ-
ment in relation to which behavior occurs.

Evaluation of the Rationale
Justifying,Self-Descriptions

It can now be seen how this theory reconfirms Kuhn's
justification of the assertion that self-descriptions provide
lihhs_between an actor, the social structure, and behavior.
The particular symbols in terms of which an actor conceives
of himself have the virtue of (a) being a portion of the
shared symbols of some group, (b) designating a class of
actors, (c) designating classes of objects, and (d) designat-

ing a class of relationships between "(b)" and "(c)" which

 

define both “(b)" and "(c)" in terms of one another in a
context of on-going behavior. Implicitly then, given the
symbols in terms of which an actor defines himself and the
meanings which these symbols have for the actor, we should
also have (1) a significant portion of the behavior in which
that actor is likely to engage, and (2) some understanding

of the nature and scope of his social—psychological location

69

in the social structure both inferred from the fact that
these symbols will reflect their source in the unique
symbolic universe shared by some groups in the social-
structure. The links arise then from the context from which
the symbols were derived and are "given" by relationality

comprising their meaning.

Evaluation of the Asset-Liability Instrument

It is possible also to show that some of the funda-
mental difficulties associated with the ALI are explainable
from the standpoint of this theoretical position. Since
the significant criticisms were most relevant to this issue,
and because in what follows we will attempt to show that
they are sufficiently damaging to warrant rejecting the
ALI as a technique for linking self to reference groups
and to behavior, only these three criticisms will be con-
sidered here.

It will be recalled that in essence the criticisms

labeled "2," "3," and "4" centered around the following
points:
"2." Self-evaluations obscured the very links between

actors, social-structural involvement, and be-
havior that they were supposed to reveal,

"3." Lack of knowledge of the meanings to be attri-
buted to the responses which would justify their
categorization,

"4." Absence of explicit theoretical basis for linking

a response on the ALI to social structure or to
behavior.

70

It will also be recalled that the fundamental assumptions

justifying the ALI were that by merging self-description

 

and self-evaluation
(1) the values according to which the self-assessments
would be made would closely reflect their source
in the social structure (as had the categories
of self-description in the TST) and

(2) the values would be an important influence on
behavior.

It was assumed that on the basis of these two considerations
the ALI would be an effective means of linking the self to
reference groups and to behavior.

Now the difficulties noted in criticisms "2" through
"4" can be seen to directly negate assumption "(1)" above;
that is, on just those points where the ALI was expected
to be the most effective it has been found to be the most
defective. An explanation of this curious result from the
standpoint of the present perspective follows.»

Criticism "2": In Chapter I, no clear distinction
was drawn between categories of self—description and self-
evaluations as to the possible consequences that each might
have in establishing links between actors, the social-
structure, and behavior. It was felt that the symbols would
somehow reveal both values and categories and would, there-
fore, yield more obvious relationships, and that the
responses themselves would somehow reveal their connection
to the social structure and demonstrate their implications
for behavior. waever, it can now be seen that the symbols
on the ALI are far more complex than non-evaluative self—

descriptions, for self—evaluations are relationships gt

71

relationships. For, whereas symbols of self-description
are relationships of classes of actors, and objects, an
evaulation of such symbols introduce an additional relation-
ship--the one between the symbol and another symbol on an
evaluative scale.

From the standpoint of the present theory the symbols
elicited by the ALI have, by virtue of their complexity, so
obscured the kinds of links which were supposed to be forth-

10

coming from them that the ALI is rendered virtually useless

for accomplishing the task for which it was designed.

‘ Criticism "3": Closely related to "2," and partly
resulting from it, is the absence of certainty associated
with the meanings which were imputed to various responses
on the ALI. waever, in accordance with the present
chapter, it should be apparent that it is just these meanings
that are essential to the task of relating the selftp
reference groups and behavior--to which end the ALI was
constructed. Without these meanings there are no relation—
ships and, consequently, no links except those that can

be derived solely on the basis of an uncertain and unreliable

empathy.

 

10This is true with the possible exception of some
obvious and trivial instances such as the previously noted
connection between grades and Intellectual or Scholastic
asset and liability statements.

72

Criticism "4": Absence of a theoretical basis for

 

relating specific categories of response on the ALI to specific
instances of reference group involvement or behavior did

not entirely frustrate all attempts at deriving testable
hypotheses. But, if it is true that the self is important
because it links modes of symbolic representation to specific
contexts of social processes and to activity, then it is
particularly serious that it is at just this point that there
is no theoretical rationale according to which these links
are predictable. It is at this point too, that the present
theoretical statement is silent and in greatest need of
development; more will be said of this later.

Since the difficulties giving rise to these three
criticisms of the ALI appear to reinforce the obscurity of
the very relationship which the ALI was designed to reveal,
and since criticism "2" appears to be a sufficient condition
preventing the elimination of "3" and "4" and is by the very
nature of the ALI not capable of being eliminated, we can
only conclude that, from the standpoint of the theoretical
statement in this chapter, the Asset-Liability Instrument
does not accomplish what was intended for it.

Implications for Further Research and
Development of the Theory

Two of the most significant problems for empirical
research posed by the position developed in this chapter are
(a) finding suitable techniques for gathering these self—

symbols, and (b) finding techniques for getting at the

73

meanings inherent within these symbols. If the ALI is
inadequate for doing this, what sort of techniques would be
acceptable and empirically feasible?

First, in any such technique the fundamental error
of the ALI, which gave rise to criticism "2", must be
avoided. Specifically this means that self-descriptions
and self-evaluation should be independently measured. These
two components of the self can prdbably be much more pre-
cisely and fruitfully studied by instruments designed to
mesh with the theoretical properties of evaluations and self-
representation. Combining these into one instrument
only obscures both.ll’ 12

Second, the gathering of self-symbols could best be
done by an instrument with a minimum amount of structure
in order to avoid leading subjects to represent themselves
in terms of symbols which they otherwise might not use.
This requirement seems to be admirably met by the TST,
consequently our attention will be directed to the problem

of discovering the meanings of symbols derived from this

or some similar technique.

 

11This depends, of course, on the presence of a
theory which will give sudh information.

12For an example of promising research which
combines such separate measures see, F. B. Waisanen, "Self-
Attitudes and Performance Expectations," Sociological
Quarterly, V01. 3 (1962), pp. 208-219.

74

Third, it would be as unnecessary as it would be
difficult to completely catalog all of the meanings of all
of the symbols gathered from some population of respondents.
From the standpoint of research which would utilize these
meanings, the problem would reduce to getting just enough
material to fit the requirements of a particular problem.13
Any number of techniques of questionnaire construction
currently employed in social-psychology would probably be
adaptable to this end. Hewever, it would be premature to
suggest specific techniques until one had acquired a ”feel"
for the ways in which respondents indicate the nature of the
relationships entailed by their self-symbols. This might
best be done in an interview situation. Specific techniques
should follow only after a close look has been had at the
relational structure of self-symbols as they appear to the
respondent.

Fourth, the measurement of these relationships (at
such time as suitable techniques are developed), should
proceed concurrently with independent efforts to assess
the respondent's involvement in and commitment to reference
groups and the context of his behavior, i.e., the objects
toward which relationships are established. It is at this

point that a wedding of the theoretical position presented

 

13This would be true provided, of course, that the
state of knowledge and of theoretical development were
sufficiently advanced to permit this.

75

in this chapter with some form of organizational theory
might be fruitfully consummated.

Finally, the lack of an adequate theory whereby
responses on the ALI can be linked to the social structure
and to behavior seems to be partly a result of a lack of
information concerning the meanings which the symbols possess.
That is, no adequate theory is possible until we know what
the self-statements mean and we won't know that until we let
the respondents tell us. There are, however, some likely
avenues of approach to the development of sudh a theory
which have been suggested already in this chapter. The
one that has some empirical confirmation behind it is the
rationale employed by McPartland and Cumming and McPartland,
Cumming and Garretson, in which it is suggested (in different
terminology, of course) that there is a certain isomorphism
between the structure of responses on the TST and the
structure of behavior associated with it, e.g., bizarre
self—responses and bizarre behavior, nonsocial self—

14 Theoretical

responses and non-social behavior, etc.
developments could be forthcoming too from the observation
that self-symbols may vary in the extent to which the relation-

ships implied by them were to formally or to informally

 

l4In criticism "5" of Chapter II it was noted that the
obviousness and narrowness of the categories on the ALI
might be associated with strong but theoretically trivial
relationships. This example of McPartland et al. is an
instance of manifestly hghrtrivial and hgh-obvious relation-
ships yielded by a rather specific and mph-obvious theoretical
proposition.

76

constituted groups. Still other theoretical propositions
might follow from a knowledge of the specificity or
generality¢1fclasses of objects and relationships which
might be associated with the degree to which any self-

symbol implies a narrow or broad range of behavior.

Similarly, the degree of specificity or generality of the
class of actors to which a self-symbol referred could be
related to the degree of similarity or uniqueness of behaviors

and social-structural involvements of a number of individuals.

Summary

The central focus of this thesis has been an assess—
ment of the Asset—Liability Instrument. This instrument
was developed as a technique for eliciting self-symbols
that would enable respondents to be linked in important
Ways to their location in the social structure and which
would enable predictions to be made regarding their behavior.
This was to be accomplished in the ALI by combining self-
descriptions and self-evaluations into one instrument.
This combination would, it was reasoned, accomplish two things.
First, it would yield categories according to which indivi-
duals thought of themselves and which had already been
proven useful in research with the TST. Second, the most
salient values held by the respondents would also be forth-
coming. It was argued that these values would reflect the
groups in which the respondent was significantly involved,

and would also be of consequence for his behavior.

77

A number of problems were encountered in using the
ALI and in interpreting results from it. This in turn led
to a reconsideration of the theoretical justification behind
the basic assumption that self-symbols were an adequate
way of linking an actor to the social structure and to his
behavior by means of self-symbols. This reconsideratiOn
had two major consequences for the thesis: (1) it
reconfirmed the rationale behind the assumption of the
theoretical and empirical value of self-representations,
and (2) it also confirmed the fact that the ALI is incapable
of eliciting such useful self—descriptions.

From this new perspective the links between self-
symbols, social—structure, and behavior are seen to be a
consequence of two characteristics of the symbolic self.
First, they are derived from a unique context which is
significant to the actor and in which symbols and their
meanings are socially shared and developed. Second, the
meanings of self—symbols establish behavioral rslationships
between certain socially defined categories of actors and
objects.

For a number of reasons the ALI fails as a method
of empirically linking an actor to his social-structural
involvement and to his behavior. Hewever, one reason in
particular is sufficientto condemn it: the combination
of self-descriptive and self-evaluative statements into a

single instrument which was to give the ALI its unique value

78

as a method of revealing these links, is just that feature
which obscures them. Self—description and self-evaluations
are theoretically and empirically distinct; as a consequence
of combining them the meaning of the responses on the ALI,
and hence the relationships which they were to reveal,
are hopelessly confounded.

Finally, some very general suggestions for further
development of the theory and for developing instruments
which avoid the major flaws exhibited by the ALI are

presented and briefly discussed.

BIBLIOGRAPHY

Brookover, Wilbur B., Ann Patterson and Shailer Thomas.
Self-Concept of Ability and School Achievement.
Final Report of Cooperative Research Project No.
845. Office of Research and Publications, Michigan
State University, 1962.

Clark, Burton R. Educating the Expert Society. San Francisco:
Chandler Publishing Company, 1962.

Cohen, Arthur R. "Some Implications of Self-Esteem for
Social Influence," in H0vland and Janis (eds.).
Personality and Persuasibility. New Haven:
Yale University Press, 1959.

 

.Coombs, Clyde H. A Theory of Data. New York: John Wiley
and Sons, Inc., 1964.

 

Davis, James A. with David Gottlib, Jan Hajda, Carolyn Huson,
Joe L. Spaeth, Stipends and SpousesZThe Finances
of American Arts and Science Graduate Students.
Chicago: University of Chicago Press, 1962.

Gaier, Eugene L. and Helen S. Wambach. "Self-Evaluations
of Personality Assets and Liabilities of Southern
White and Negro Students," Journal of_§ocial
Psychology, Vol. 51 (1960), 135-143.

Gottlieb, David and Benjamin Hodgkins. "College Student
Subcultures: Their Structure and Characteristics
in Relation to Student Attitude Change,“ The School
Review, Vol. 71 (1963), 266-289.

Hickman, C. Addison and Manford Kuhn. Individuals, Groups,
and Economic Behavior. New York: The Dryden
Press, 1956.

Hoijer, Harry. "The Sapir-Whorf Hypothesis," in Readings
in Anthropology, Vol. I. New York: Thomas Y.
Crowell, 1959.

Inkeles, Alex. "Industrial Man: The Relation of Status

to Experience, Perception and Value," American
Journal of Sociology, V01. 66 (1960), 1-31.

79

80

Littunen, Yrjo and Eugene L. Gaier. “Occupational Values
and Modes of Conformity," Journal of Social
Psychology, V01. 51 (1960), 123-133.

McPartland, Thomas. Manual for the Twenty Statements Problem,
(Revised, Department of Reserach, the Greater
Kansas City Mental Health Foundation, 1959).

Mead, George H. Mind, Self and Society. Chicago: The
University of Chicago Press, 1934.

. The Philosophy of the Act. Chicago: The
University of Chicago Press, 1938.

 

Merton, Robert K., and Alice S. Rossi. ”Contributions to
the Theory of Reference Group Behavior,” in
Robert K. Merton, Social Theory and Social Structure.
Revised and enlarged; Glencoe, Ill.: The Free

Press, 1957.

Miyamoto, S. Frank, and Sanford Dornbusch. "A Test of
Interactionist Hypotheses of Self-Conception,”
American Journal of Sociology, Vol. 61 (1956).

Riesman, David. Faces in the Crowd. New Haven, Conn.:
Yale University Press, 1950.

Sherif, Muzafer and Carolyn W. Sherif. An Outline of Social
Psychology. New York: Harper and Brothers, 1956.

Stotland, Ezra, Stanley Thorely, Edwin Thomas, Arthur R.
Cohen, and Alvin Zander. “The Effects of Group
Expectations and Self-Esteem Upon Self-Evaluations,"
J. of Abnormal and Social Psychology, V01. 54
(1957), 55-63.

Waisanen, F. B. "Self-Attitudes and Performance Expectations,"
Sociological Quarterly, V01. 3, No. 3 (1962),
208-219.

‘Aiz‘iHIGAH ATE UNIVERSITY LIBPARIPS

I !g‘ ”i“ V 1‘ I!) III .
:1 l 1“ {'1’ \I

'2‘ [:(W ”I 1‘ #1) i

3

1 1!
93 03146 1969