Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
This dissortation has boen
microfilmod exactly as received 69-16,656
KEIM, Willard Dennis, 1932-ROLE-TAKING ACCURACY AND DISOWNINGPROJECTION: TIIEIR RELATION TOADVOCACY Of<' CIIANGE AND POLITICALSUCCESS.
University of lIawaii, Ph.D., 1969Political Science, general
University Microfilms. Inc.. Ann Arbor, Michigan
ROLE-TAKING ACCURACY AND DISOWNING PROJECTION:
THEIR RELATION TO ADVOCACY OF CHANGE
AND POLITICAL SUCCESS
A DISSERTATION SUBMITTED TO THE GRADUATE DIVISION OF THEUNIVERSITY OF HAWAII IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
IN POLITICAL SCIENCE
JANUARY 1969
ByI
Willard D~'Keim
Dissertation Committee:
Marshall N. Goldstein, ChairmanMichael J. ShapiroYasumasa KurodaRobert B. StaufferMilton Bloombaum
PLEASE NOTE: Appendix paHcsare not original copy. Printis indistinct on many pages.Filmed in the best possibleway.
UNIVERSITY MICROFILMS.
ABSTRACT
The purpose of this research was to determine the
relationship of role-taking accuracy, assumed similarity,
~ctual similarity, and disowning projection to other
political variables, of which the most important for the
study were advocacy of change and political success.
Success was measured by scores on self-assessed influence
and having held office in an organization.
A questionnaire was sent to all legislators, house of
representatives and senate, of the State of Hawaii. From
the results of this questionnaire, 35 "agree-disagree"
items were derived on which there was substantial con
sensus among the legislators. In addition, six items
on which legislators indicated a preferred response to a
hypothetical situation showed consensus.
Legislators' consensus responses were built into
another questionnaire in which the "agree-disagree" items
were to be answered by respondents from two voluntary po
litical organizations (1) for thems-elves and (2) again as
if they were legislators. Matching these scores against
the legislators' own responses gave a behavior score, the
sum of correct guesses by respondents on situation
questions; an actual similarity score, the respondents
answers for himself on the "agree-disagree" items matched
iii
against the legislators responses; an assumed similarity
score, on which the respondents own answers were matched
with his guesses for legislators; and a role-taking ac
curacy score, on which the respondents guesses for a
legislators' responses were matched against the legis
lators' answers.
These five variables were used together with 31
other variables and attributes of political content in a
number of analyses. First, the groups from which re
spondents were drawn were compared in several one-way
analyses of variance. No differences between groups on
the role-taking variables were discovered.
Second, a series of cross-tabulations were analyzed
between the variables most pertinent to this analysis and
the remaining variables and attributes. This analysis
merged with a factor analysis of the entire 36X36 matrix.
An initial orthogonal varimax solution was rotated to a
biquartimin oblique solution in order to attain the best
simple structure. Factors emerging from this analysis
were named (in order of variance explained): Left-radical,
Real efficacy, Role-taking, Political empathy, Change, Age,
and Group success.
Third, regression equations predicting to advocacy
of change and political success indicated the variables
from among eleven that explained the most variance in
advocacy of change and political success. On the level of
iv
analysis with the whole group of respondents, the role
taking variables were not strikingly associated with
political success or advocacy of change.
Finally, an empirical method of grouping the re
spondents isolated eight political-role types. The
profiles of each type over 24 variables were compared.
On this level of analysis, the role-taking variables were
related in several subgroups in accordance with the
hypotheses relating role-taking accuracy to political
success and disowning projection to advocacy of change.
TABLE OF CONTENTS
ABSTRACT ...
LIST OF TABLES
LIST OF -FIGURES.
ii
. vii
xii
CHAPTER I. THE PROBLEM -_ ~
1.1 Why This Problem?1.2 A Brief Survey of the
Science LiteraturePolitical
1
3
41
3234
CHAPTER II.
CHAPTER III.
CHAPTER IV.
DEFINITIONS, DESIGN, AND HYPOTHESES2.1 Defin~tio~s and Operation-
al~zat~ons . . . . . . . . .. 102.2 Some Measuring Techniques
Rejected . . . . . .. ... 202.3 The Research Technique. . . . 232.4 Hypotheses. . . . . . . 27
LEGISLATIVE CONSENSUS3.1 The Choice of the Political-
Other. . . . . . . . . .3.2 The Criterion of Consensus.3.3 Representativeness of the
Legislative Respondents.
THE NATURE OF THE RESPONSE GROUPS4.1 The Choice of Respondents. 454.2 Description of Group A. . . . 484.3 Description of Group B. . . . .. 634.4 Between Groups Differences. . 67
. 122
CHAPTER V. PRELIMINARY COMMENTS ON THE ROLE-TAKINGAND DEPENDENT VARIABLES AND CROSSTABULATIONS
5.1 Scales. . . . . . . .. . ..5.2 Evidence on Role-Taking from
Cross-tabulation . .5.3 Hint~ toward the Factor
b."~alysi~ '. . . . . . . .
73
77
CHAPTER VI. POLITICAL DIMENSIONS OF ROLE-TAKING6.1 The Orthogonal Factor Analysis .. 1256.2 Oblique factors of Political
Role-Taking. . . . . . . . . . 139
vi
CHAPTER VII. THE BEST PREDICTORS OF SUCCESS ANDADVOCACY OF CHANGE AND A POLITICAL-ROLETYPOLOGY
7.1 Regression Equations.. . .. 1477.2 Political Profiles. . . 156
CHAPTER VIII.
APPENDICES .
BIBLIOGRAPHY
CONCLUSIONS8.1 Reconsiderations of the
Hypotheses .8.2 Suggestions for Further
Research
175
. . . . 183
. . 188
. . 224
_...
TABLE
LIST OF TABLES
PAGE
3.1 QUESTIONS FROM THE LEGISLATIVE QUESTIONNAIRESHOWING CONSENSUS. . . . . . . . . . . . . .. 36
4.1
4.2
DESCRIPTION OF RESPONDENTS, GROUPS A AND
ROLE QUESTIONS ANSWERED FOR SELF BYGROUPS A AND B . . . . . . . . . . .
B • • 51
59
4.3 ROLE QUESTIONS ANSWERED AS "LEGISLATORS" BYGROUPS A AND B . . . . . . . . . . . . . . .. 61
4.4 ANALYSIS OF VARIANCE BETWEEN GROUPS A AND B,PARTICIPATION INDEX. . . . . . . . . . . . .. 69
4.5 ANALYSIS OF VARIANCE BETWEEN GROUPS A AND B,ADVOCACY OF CHANGE . . . .. 70
4.6 ANALYSIS OF VARIANCE BETWEEN GROUPS A AND B,POLITICAL KNOWLEDGE INDEX. . . . . . . . . .. 71
5.1
5.2
5.3
5.4
CORRELATES OF HELD OFFICE HELOFF/ACQLEG.
CORRELATES OF HELD OFFICE HELOFF/POSOFF.
CORRELATES OF HELD OFFICE HELOFF/ISSACT.
CORRELATES OF HELD OFFICE HELOFF/SOCIAL.
80
80
81
81
5.5 CORRELATES OF HELD OFFICE HELOFF/PARTIC. . .. 82
5.6
5.7
CORRELATES OF HELD OFFICE HELOFF/INFSEL.
CORRELATES OF HELD OFFICE HELOFF/POLINT.
. . .
. . .82
83
5.8 CORRELATES OF HELD OFFICE HELOFF/POLKNO. . .. 83
5.10 CORRELATES OF HELD OFFICE HELOFF/PROJEC.
5.11 CORRELATES OF HELD OFFICE HELOFF/ASPGAP.
5.9 CORRELATES OF HELD OFFICE HELOFF/BSCORE. 84
84
85
5.12 CORRELATES OF INFLUENCE OF SELF INFSEL/ACQLEG. 90
5.13 CORRELATES OF INFLUENCE OF SELF INFSEL/POSOFF. 90
viii
TABLE PAGE
5.14 CORRELATES OF INFLUENCE OF SELF INFSEL/ISSACT. 91
5.15 CORRELATES OF INFLUENCE OF SELF INFSEL/SEX . . 91
5.16 CORRELATES OF INFLUENCE OF SELF INFSEL/ELDONL. 92
5.17 CORRELATES OF INFLUENCE OF SELF INFSEL/EGOSTG. 92
5.18 CORRELATES OF INFLUENCE OF SELF INFSEL/POLCYN. 93
5.19 CORRELATES OF INFLUENCE OF SELF INFSEL/POLEFF. 93
5.20 CORRELATES OF INFLUENCE OF SELF INFSEL/PARTIC. 94
5.21 CORRELATES OF INFLUENCE OF SELF INFSEL/POLINT. 94
5.22 CORRELATES OF ADVOCACY OF CHANGE CHANGE/POSOFF 95
5.23 CORRELATES OF ADVOCACY OF CHANGE CHANGE/ISSACT 96
5.24 CORRELATES OF ADVOCACY OF CHANGE CHANGE/EMPJOB 96
5.25 CORRELATES OF ADVOCACY OF CHANGE CHANGE/RADICAL 97
5.26 CORRELATES OF ADVOCACY OF CHANGE CHANGE/EQUAL. 97
5.27 CORRELATES OF ADVOCACY OF CHANGE CHANGE/INNOUT 98
5.28 CORRELATES OF ADVOCACY OF CHANGE CHANGE/HUMANE 98
5.29 CORRELATES OF ADVOCACY OF CHANGE CHANGE/POLEFF 99
5.30 CORRELATES OF ADVOCACY OF CHANGE CHANGE/PARTIC 99
5.31 CORRELATES OF ADVOCACY OF CHANGE CHANGE/INFGRP 100
5.32 CORRELATES OF ADVOCACY OF CHANGE CHANGE/FINDEX 100
5.33 CORRELATES OF ADVOCACY OF CHANGE CHANGE/APOLIT 101
5.34 CORRELATES OF ASSUMED SIMILARITY ASSUME/ACQLEG 105
5.35 CORRELATES OF ASSUMED SIMILARITY ASSUME/ISSACT 105
5.36 CORRELATES OF ASSUMED SIMILARITY ASSUME/RADICL 106
5.37 CORRELATES OF ASSUMED SIMILARITY ASSUME/FAITHP 106
ix
PAGE
CORRELATES OF ASSUMED SIMILARITY ASSUME/EQUAL 107
CORRELATES OF ASSUMED SIMILARITY ASSUME/SOCIAL 107
CORRELATES OF ASSUMED SIMILARITY ASSUME/POLEFF 108
CORRELATES OF ASSUMED SIMILARITY ASSUME/ACTUAL 108
CORRELATES OF ASSUMED SIMILARITY ASSUME/PROJEC 109
TABLE
5.38
5.39
5.40
5.41
5.42
5.43 CORRELATES OF ROLE-TAKING ACCURACYROLEAC/POSOFF .
5.44 CORRELATES OF ROLE-TAKING ACCURACYROLEAC/ISSACT .
5.45 CORRELATES OF ROLE-TAKING ACCURACYROLEAC/POLEFF .
5.46 CORRELATES OF ROLE-TAKING ACCURACYROLEAC/ASSUME .
5.47 CORRELATES OF ROLE-TAKING ACCURACYROLEAC/ACTUAL. . . . . . . . . . . .
5.48 CORRELATES OF ROLE-TAKING ACCURACYROLEAC/PROJEC .
5.49 CORRELATES OF ROLE-TAKING ACCURACYROLEAC/FINDEX. . . . . . . . . . .
5.50 CORRELATES OF THE BEHAVIOR SCOREBSCORE/ISSACT. . . . . . . . . .
5.51 CORRELATES OF THE BEHAVIOR SCOREBSCORE/FLEXBL. . . . . . . . . .
5.52 CORRELATES OF THE BEHAVIOR SCOREBSCORE/SOCIAL. . . . . . . . . . .
5.53 CORRELATES OF THE BEHAVIOR SCOREBSCORE/ROLEAC .
5.54 CORRELATES OF THE BEHAVIOR SCOREBSCORE/ASSUME. . . . . . . . . .
5.55 CORRELATES OF THE BEHAVIOR SCOREBSCORE/PROJEC .
110
110
111
111
112
112
113
114
114
115
115
116
116
TABLE
5.56 CORRELATES OF DISOWNING-ASSIMILATIVEPROJECTION PROJEC/ISSACT . . . . . .
5.57 CORRELATES OF DISOWNING-ASSIMILATIVEPROJECTION PROJEC/RADICL . . . . . .
5.58 CORRELATES OF DISOWNING-ASSIMILATIVEPROJECTION PROJEC/SOCIAL . . . . . .
5.59 CORRELATES OF DISOWNING-ASSIMILATIVEPROJECTION PROJEC/POLINT . . . . . .
5.60 CORRELATES OF DISOWNING-ASSIMILATIVEPROJECTION PROJEC/FINDEX . . . . . .
x
PAGE
117
118
118
119
119
6.1 VARIABLES ENTERING BUT SLIGHTLY INTO THECOMMON FACTOR SPACE, AS INDICATED BY LOWCOMMUNALITIES. . . . . . . . . . . 126
6.2 FACTOR I: LEFT-RADICAL (Reflected) 127
6.3 FACTOR II: REAL EFFICACY.
6.5 FACTOR IV: POLITICAL EMPATHY
6.6 FACTOR V: CHANGE.
6.4
6.7
FACTOR III: ROLE-TAKING ...
FACTOR VI: AGE .
131
132
133
135
136
6.8 FACTOR VII: GROUP SUCCESS (Reflected). . 137
6.9 PATTERN MATRIX OF OBLIQUE BIQUARTIMINFACTORS FOR POLITICAL-ROLE VARIABLES . 141
6.10 CORRELATIONS AMONG OBLIQUE FACTORS FORPOLITICAL-ROLE VARIABLES . . . . . . . . 143
7.1 LOADINGS ON ELEVEN ORTHOGONAL FACTORS BYINDEPENDENT VARIABLES FOR THE REGRESSIONEQUATIONS. . . . . . . . . . . . . . . . 148
7.2 REGRESSION FROM SELECTED FACTOR SCORES TOHELOFF . . . . . . . . . . . . . . . . . .. 150
TABLE
7.3 REGRESSION FROM SELECTED FACTOR SCORESINFSEL . . . . . . . . . . . . . . . .
7.4 REGRESSION FROM SELECTED FACTOR SCORESCI-IANGE . . . • . . . . . . . . . . . .
xi
PAGE
TO. . . 152
TO. . . 153
7.5 STANDARD DEVIATION OF FACTOR SCORES FORSEVEN FACTORS FROM A COMMON FACTOR ANALYSISOF A PRODUCT-MOMENT CORRELATION MATRIX . .. 158
FIGURE
LIST OF FIGURES
PAGE
2.1 A SCORING PARADIGM FOR THE STUDY OF POLITICALROLE-TAKING ACCURACY . . . . . . . . . . . .. 17
5.2 RELATIONSHIPS EXCEEDING A CHI-SQUARE OF 3.84AND A PHI OF .2325 AMONG THE VARIABLES ANDATTRIBUTES USED IN THE STUDY OF THEPOLITICAL-ROLE SPACE . . . . . . . . . . . 124
5.1
7.1
7.2
CURVE REPRESENTING FREQUENCIES OF PHI-COEFFICIENT SIZES .
EIGHT POLITICAL-ROLE PROFILES .....
EIGHT POLITICAL-ROLE TYPES FOUND IN TWOVOLUNTARY POLITICAL ORGANIZATIONS .'..... . .
122
163
172
_....
CHAPTER I
THE PROBLEM
1.1 WHY THIS PROBLEM?
Along with other political resources, there are certain
psychological traits and skills that may be essential for
success in a given political system. For example, one
possible base for political success might be ego strength.
An important skill would seem to be empathy, or, as it will
be called in this study, role-taking accuracy.
To operate effectively in a given political system,
it intuitively would appear useful to be able to predict the
roles of relevant "others," as these roles are defined by
attitudes and behavior. This ability is evidence of a
mental process, for the accurate role-taker does not actually
imitate the other, although he can predict his reactions.
Among the innumerable problems of nations in transition,
one is the inability of certain large groups of people to
imagine themselves in the role of a political actor. It is
possible that political apathy may be in part the result
of the difficulty people have in adjusting quickly to new
roles.
It may also be the case that community development
projects falter because they are staffed with people who
2
are inaccurate role-takers. Persons who cannot predict the
attitudes of villagers may be less successful than they
might be in attempting to modernize village life.
In the United States, there is reason to be concerned
with the results of the poverty programs and attempts to
induct slum dwellers into the polity. Among white, middle
class Americans, including many of the decision-makers, it
is unusual to find one who can understand or predict the
attitude of slum dwellers to the legal apparatus. Inability
to understand how problems affect others results in the
introduction of well-intentioned but irrelevant measures
designed to alleviate these problems.
Are some persons, therefore, better (more accurate)
role-takers than others? Is this role-taking accuracy
pertinent in a study of politics? For example, are there
some people who are more successful in their political
goal attainment because they can perceive the attitudes of
those who are already successful political actors?
On the other hand, there is another type of person
who might be called a projector. The projector does not
construct the others' role on the basis of reality but on
the basis of his own needs, goals, and motives. If the
projectors' own attitudes differ greatly from present
occupants of a political role, then it might be assumed
that they would be a force for change whenever they
entered the political arena. Are these persons successful
3
in a given political system? Do they tend to advocate
greate~~litical change than others do?
One surmise might be that projectors who tend to dis
own certain qualities of self and attribute these to
political role-holders might be persons who advocate
relatively large changes in the political system, while
those who tend to liken political role-holders to themselves
would be less likely to be advocates of large political
changes.
As a step toward analyzing these two major types of
individuals, the accurate role-takers and the projectors
(disowning and assimilative), this dissertation will first
provide a method of measuring the differences. If this
method is fruitful in contributing to an analysis of the
styles of politically active persons in Honolulu, then its
improvement and application in other countries and situations
may be recommended, and some vital facets of political life
may be made more clear.
1.2 A BRIEF SURVEY OF THE POLITICAL SCIENCE LITERATURE.
A complete survey of all the literature pertinent to
the subject of role-taking would comprise a book, without
advancing directly the research to be conducted. The
bibliography will list background literature for the
dissertation, and particular citation will be made as these
works illuminate specific problems. However, some justifi
cation is necessary in order to establish the relevance of
4
role-taking to political science concerns. This literature
is not very extensive.
The concept of role-taking was implicit in Gabriel
Almond's The Appeals of Communism. 1 In his concentric
model of Communist organization, he suggested that those
who reached the center councils of the party lost their
ability to interact with non-Communists, leaving the
proselytization work to members of the party who were still
on the organizational periphery.
The nature of the respondents as a sample of Communist
party members made it difficult for Almond to support this
model except qualitatively. His research and results are
nevertheless suggestive for further research into the role
taking variables.
In the field of game theory and bargaining, Thomas C.
Schelling encounters a number of typical role-taking
situations. He notes, "People can often concert their
intentions or expectations with others if each knows that
the o~~er is trying to do the same." 2 This suggests that
in politics the person capable of guessing what the other
person is going to do is likely to be more successful than
an inaccurate guesser.
1princeton, N.J., 1954.
2The Strategy of Conflict, New York, 1963, p. 57.Chapters 3 and 4 are especially pertinent to a ro1etaking concept.
5
While Schelling does not have people explicitly guess
the attitudes of others, some of the games he presents have
people guessing numbers in common or points of rendezvous
on a map, which might entail some knowledge of the attitudes
of other people.
Role-taking is a concept used in public administration,
which has generally relied on results of studies by social
psychologists. A recent publication that treats extensively
of empathic responses is Empathy and Ideology: Aspects of
Administrative Innovation. 3 The essays in this collection
treat empathy as one of the important behavioral dimensions
of administrative success. None of the research reported,
however, empirically substantiates the importance of role
taking accuracy for persons operating in the political
arena. For example, predicting a political-other is not
necessarily the same skill as predicting such others as
marital partners, and persons skillful in predicting some
others may be unsuccessful when presented with political
actors. It is therefore necessary that role-taking and its
associated concepts be studied in connection with distinctly
political roles.
In the field of political development, one of the most
interesting attempts to apply the concept of empathy to the
3Charles Press and Alan Arian, editors, Chicago, 1966.
6
study of the developing nations is Daniel Lerner's The
Passing of Traditional Society.4 Unfortunately, the con
cept is there used in a fashion that greatly diminishes the
value of the findings. By using the term as a shorthand
for several very different mechanisms, the results Lerner
obtains may be greatly contaminated. Projection, "assigning
to the object certain preferred attributes of the self,"
distantiation or negative identification, which "results
when one projects onto others certain disliked attributes
of the self," and introjection, which enlarges identity by
attributing to self certain desirable attributes of objects,
all are included in empathy as measured by Lerner. He
defends his operationalization as "a pragmatic, not theoretic,
intent," and states that his interview data "does not permit
systematic discrimination between introjective and projective
mechanisms. ,,5
As a possible result of using the concept of empathy
in this way, some of Lerner's findings are subject to
question. For example, his Table 6, "Empathy to Modern
Press Standards,,,6 shows that modern types for five middle
4Glencoe, Ill., 1958.
5Ibid ., p. 49. But, since Ralph D. Norman and PatriciaAinswo'rt'El;" "The Relationships Among Projection, Empathy,Reality, and Adjustment, Operationally Defined," Journal ofConsulting pstchology , 18 (1954), 54-55, find empathy andreaI~ty posit vely related, projection and reality negativelyrelated, there is little doubt that the two concepts shouldbe distinguished.
6Ibid ., p. 97.
7
eastern nations appear to be less empathetic to press
standards~-than transitionals. This result is disconcerting
if it is assumed that empathy characterizes modern society
as contrasted with traditional society. Moderns should be
expected to rank at least as high as transitionals on an
empathy measure.
Table 3, Appendix C, "Complete Latent St~cture of
Item Patterns," also indicates a spread of the empathy
variable among moderns, traditionals, and transitionals. In
fact, two of the largest groupings of traditional item
patterns are characterized by positive empathy.7 This sub
stantiates the comments of R. M. Marsh, from whose theses
much of the theory on empathy was taken, "The Index of
Empathy is more sensitive to the demographic and media
behavior characteristic of the low educated and the low
(and sometimes middle) SES groups than to the high educated
and high SES groups.,,8 Among the moderns there appeared to
be a fairly even distribution of Empathy Index types.
It is clear that more work is necessary before empathic
concepts can be shown to be fruitful for political scien
tists. In this research, role-taking accuracy will replace
Lerner's empathy concept. Lerner enumerated responses to
nine projective questions, considering a high total as
7Lerner, p. 442.
8Ibid ., p. 435.
indicative of empathy, whereas our measure will give a
score for each individual according to his accuracy in
predicting others' attitudes. Role-taking accuracy is
operationally different from Lerner's empathy concept,
and it may well tap a different trait.
Furthermore, an attempt will be made to break up
Lerner's broad pragmatic usage of empathy into a number of
possibly distinct components. A more discriminating
measurement device will be developed to reveal empathic
differences in already modern societies. Presumably, if
role-taking and associated concepts are found to make a
8
difference in political success and other political
variables, it ought to be possible to distinguish political
actors within an already modern polity.
This research is not a replication of Lerner's study.
The concepts used are not necessarily attached to the same
personality mechanisms as the broad concept of empathy used
by Lerner.
Lerner has made the most use of the empathy concept,
but it is alluded to elsewhere in the political science
literature. Robert Chin holds that ". the single most
important improvement the change-agent can help a c1ient
system to achieve is to increase its diagnostic sensitivity
to the effects of its own actions upon others.,,9 He
9"The Utility of System Models and DevelopmentalModels," in Jason L. Finkle and Richard W. Gable, editors,Political Development and Social Change, New York, 1966,p. 12.
suggests that such concepts as role-taking and projection
can pertain to individuals, groups, governments, and
nations. Other studies have suggested the use of ro1e
taking accuracy to tap perceptions persons have of other
nations. Hadley Cantril-and William Buchanan and Bruce
Russett have gone furthest in connecting such a concept
to empirical data. lO
This bibliographic section shows that empathy con-
9
cepts are not foreign to interests of political scientists.
However, the concept has not usually received a0~quate
empirical treatment. It is too easy to suggest the ob-
vious notion that to perceive others accurately is an
important political attribute leading to possible con-
sequences such as success in achieving goals. This
common-sense proposition, however, is open to question
until it is verified empirically. Finally, it was shown
that the most important application of the empathy concept
in political research had some serious shortcomings that
this dissertation will attempt to overcome.
lOCantril and Buchanan, How Nations See Each Other,Urbana, Ill., 1953; Russett, Community and Contention,Cambridge, Mass., 1963.
CHAPTER II
DEFINITIONS, DESIGN, AND HYPOTHESES
2.1 DEFINITIONS AND OPERATIONALIZATIONS
Levinson provides three defini~ion~~hat have oriented
role research in socio10gy.1 A role may be defined as the
structurally given demands (norms, expectations, taboos, and
responsibilities) associated with a given social position.
In the case of a political role, such as that of legislator,
a role might be determined by the state constitution and
the laws that prescribe certain activities and prohibit
others. Other structural pressures, such as those from
voters and the governor, would also channel a legislator's
behavior. Political knowledge from books on government or
personal occupancy of similar positions as that of legis
lators would inform a person of a role in this sense.
A second definition of role is the orientation or
conception of it by the members playing a part in an
organization. This is the inner definition of what a
person in a particular social position is supposed to think
1Danie1 J. Levinson, "Role, Personality, and SocialStructure in the Organizational Setting," The Journalof Abnormal and Social Psychology, 58 (1959), 172.
11
and do. Finally, a role may be taken as the actions of
individual members of a position, even if these actions are
violations of the organizational norms.
The meaning of role in this dissertation is closest
to the second definition. Although questions regarding
behavior in particular situations were asked the legis-
lators, it must be recognized that responses to question
naire items are probably closest to a role-holder's
conception of how he ought to behave. It was not determined
empirically, although it would be interesting to have done
so, whether questionnaire responses were accurate indicators
of role in the other two definitions.
Respondents in this research were asked to judge the
qualities occupants of a particular political position might
be expected to say they have. Care was taken to have the
respondent predict questionnaire items as though he were a
legislator responding to items on a questionnaire. These
distinctions must be borne in mind so that the conclusions
are actually in keeping with the task presented to role
takers.
The concept of empathy has been defined in a number of
ways. For example, Katz holds that "when we empathize, we
lose ourselves in the new identity we have temporarily
assumed. ,,2 This is similar to the original meaning of the
2Robert L. Katz, Empathy: Its Nature and Uses, Glencoe,
Ill., 1963, p. 9.
12
term as it was used by Theodor Lipps, for whom Einfuehlung
(empathy) implied the loss of self-awareness on the part
of an observer, as when he confronted a painting or a piece
of sculpture and "fused" with the object that absorbed his
attention. This deep~r emotional experience is not to be
expected during performance of the task of answering a
questionnaire, and since the term empathy ought to be
reserved for its original meaning, the term role-taking
accuracy will hereafter be used to refer to the skill
analyzed in the succeeding research. Empathy certainly
does not refer to accuracy. It may well refer to assumed
similarity,3 another concept used in this analysis, but it
is more appropriate to regard empathy as one of the
emotions, best left, for the time being, to the field of
psychology. No assumption about the internal emotional
state of role-takers will be made, making unnecessary the
term empathy.
Role-taking accuracy refers to the accuracy with which
a person (or group) predicts how another person (or group)
responds to a set of items, as on a questionnaire. 4 This
3As in Henry Clay Smith, Sensitivity to People, NewYork, 1966, p. 22.
4This is called stereotype accuracy by Victor B. Clineand James M. Richards, "Components of Accuracy of Interpersonal Perception Scores and the Clinical and StatisticalPrediction Controversy," The Psychological Record, 12, 1962,pp. 373-381. The same operat~onal technique is usea-forempathf by H. H. Remmers, "A Quantitative Index of SocialPsycho ogical Empathy," American Journal of Orthopsychiatry,
13
role-taking aptitude involves "the skill in shifting
perspective from one's own position to that of the other,
in vicariously oscillating between self and role. More
specifically, it is the ability of the person to behave,
with or without observable enactment, as if he were in a
social position other than the one he is actually occupying. 1I5
To an extent, the position will be taken that role-taking
. . d k' 6~s att~tu e-ta ~ng.
Role-taking is not role-playing, for the former is a
mental or cognitive process, while the latter is a concept
that refers to a social function which persons in a
particular position or status are expected to perform in
overt conduct. 7 Only in a very loose sense might
20 (1950), 161-165.
5Theodore R. Sarbin and Donal S. Jones, "An Experimental Analysis of Role Behavior," in Eleanor E. Maccoby,Theodore M. Newcomb, and Eugene L. Hartley, Readings inSocial psycholog~, New York, 1958, pp. 465-472. It is unfortunate that t e terms empathy, insight, social sensitivity, and role-taking have been used for the same operationalprocedures. The most careful distinction is maintained bySheldon Stryker, "Conditions of Accurate Role-taking: ATest of Mead's Theory," in Arnold M. Rose, Human Behavior andSocial Processes, Boston, 1962, pp. 41-62.
6See Stanley Stark, "Role-Taking, Empathic Imagination,and Rorschach Human Movement Responses: A Review of TwoLiteratures," Perceptual and Motor Skills, 23 (1966), 247.
7See the distinctions in Walter Coutu, "Role-playingvs. Role-taking: An Appeal for Clarification," AmericanSociological Review, 16 (1951), 180-187.
14
respondents be said to engage in role-playing when
responding to a questionnaire as though they were legis
lators. To role-play, they might be asked to deliver a
speech before a mock legislature.
Projection will also be analyzed. The projector might
construct a role as he would if he himself with all his
present attitudes were in the situation. Or, he might
project (disown) negative values and attitudes onto the
role. Thus, the identity of the other to be predicted is
irrelevant to the role content since the subject simply
imputes his own role conceptions to the other. 8
Projection in the psychoanalytic literature does not
mean that we project our own feelings on to the other in
the sense of mentally transforming ourselves into another
person. Projection is viewed as a defense against anxiety,
and, as such, the projector is actually deficient in the
capacity to project into himself the identity of another
person. 9
If projectors are actually deficient in perceiving
what others are like, then, taking the psychological
evidence as given in the case of this concept, it follows
that they will be less accurate in role-taking. Low
8Ra1ph H. Turner, "Role-taking: Process VersusConformity," in Rose, Human Behavior, p. 318.
9Katz , Empathy, p. 42.
15
role-taking scores will therefore be evidence of the
propensity to project. Thus, the measure of role-taking
accuracy also provides a measure of projection, for
projectors will fall into the range of low scores. 10
To provide further information on projection, beyond
its being the low end of the scale for role-taking accuracy,
it may be useful to distinguish two forms of projection,
assimilative projection and disowning projection. The first
is the tendency to assume without valid supporting evidence
that others are like oneself. Disowning projection implies
that the person attributes unjustifiably to others that
which is actually his own while at the same time disclaiming
it for himself. ll
Two additional important variables are actual
similarity and assumed similarity. If a respondent answers
items on a questionnaire as his own responses, actual
similarity between him and an other may be scored as those
responses which are the same as the responses of an other,
who has answered the same items. The respondent's assumed
lOThe psychological evidence supports the logic of thiscontention. Ideally, low-scoring role-takers ought to havebeen located and tested by means of another measure, suchas the Rorschach. Such evidence was obtained by RosalindDYmond, "A Preliminary Investigation of the Relation ofInsight and Empathy," Journal of Consultin Ps cholo ,12(1948), 228-233. Respon ents n t 1S stu y, owever, coUIdnot be retraced (see Chapter 4).
llNorman Cameron, The Psychology of Behavior Disorders,New York, 1947, pp. 166-168.
16
similarity score is the number of items he answers when he
guesses an other's responses in the same way as he has
responded himself. 12
In addition to these role-taking scores, another set
of questions calls on a respondent to choose from among
several behavior resp~nses to hypothetical situations. This
behavior score furnishes an internal validity check for the
attitude type responses a respondent makes as if he were a
legislator. It is also possible that behavior scores may
vary independently with other variables than does the
attitude-type role-taking score.
These, then, are operationalizations of the concepts
role-taking accuracy, actual similarity, assumed similarity,
assimilative projection, and disowning projection. The
scoring procedure is set forth in Figure 2.1. This figure
shows the scoring for the two types of response a respondent
makes against the single response of the other.
Figure 2.1 indicates how ~hese concepts will be
operationalized by means of a questionnaire technique. It
will be noted that these concepts overlap. Some indicator
problems are avoided, since it is role-taking accuracy, a
skill, and not empathy, an emotion, which is being tapped.
A person who assumes similarity in cases in which this
l2Henry Clay Smith, Sensitivity to People, New York,1966, pp. 95-96.
17
FIGURE 2.1 A SCORING PARADIGM FOR THE STUDY OF POLITICALROLE-TAKING ACCURACya
Judge's Judge's Others'sResponseb Guess for Actualfor Self Others's Response Response Variables
XC X X Role-takingaccuracy, actualsimilarity, assumedsimilarity
y X X Role-takingaccuracy
X X y Assimilativeprojection, assumedsimilarity
X y X Disowningprojection, actualsimilarity
aThis fiRure is adaptl·d from Bernard Spilka and MarvinLewis, 'Empathy, Ass _milative Projection, and DisowningProjection," The PSfichological Record, 9 (1959~, 99-102.In their paradigm, owever, the series "agree, I "disagree," "agree" was termed error. It is entirely inkeeping with Cameron's formulat10n of disowning projection,nevertheless, to score this series as such. For example,when a person regards himself as, say, open-minded, healso disowns the quality of dogmatism, which he mayattribute to others. Once the scores are altered, thefigure can be abbreviated to four distinctions, as presented above.
bThe responses referred to are responses to role questionson the questionnaire in Appendix II.
cAccording to the questionnaire items, X may be either an"agree" or a "disagree" response.
18
similarity exists is going to be in such instances an
accurate role-taker. But the problem of overlap cannot be
ignored.
For example, taken separately, real similarity and
assumed similarity are not by definition correlated with
role-taking accuracy, nor is assumed similarity of necessity
correlated with our measure of the assimilative-disowning
dimension of projection. In the former case, a person may
be similar to members of a group and obtain a distinctly
low accuracy score unless he also assumes similarity. In
the latter case, he may assume similarity correctly and
any errors may be the result either of disowning or as
similative projection.
However, on anyone item, with each series of two
variables, the third can be predicted. That is, given
actual similarity and accuracy on a single item, assumed
similarity follows necessarily. If a group has very high
actual similarity, it follows that high assumed similarity
will be correlated in their case with high accuracy.
Another group, in which there is marked lack of real
similarity, will, if their assumed similarity is high, be
markedly inaccurate.
The attempts to control for one or another of these
variables--to reduce the effect of actual or assumed
similarity on accuracy--have been unsuccessful. Redefining
the terms has not eliminated the necessary artefactual
19- .- ~
relationships between variables. 13 On each item, they are
simply linearly dependent.
The Figur~, however, shows that none of the concepts
is completely defined by the others. To an extent, they
must vary together, but their remaining variance may be
independent of the others. Thus, a factor analysis in-
eluding these variables would show a necessary dimension
in which they are all related, a dimension descriptive of
the group (or persons) being analyzed. But other factors
might be pulled out of the common factor space, on which
the variables were related to other variables independently.
One means of resolving the problem of linearity among the
role-taking concepts, therefore, is factor analysis.
It can be left to empirical research to reveal whether
accuracy, assumed similarity or actual similarity are con-
sistently associated with the same non-role-taking variables.
But a conclusion such as "accuracy is highly positively
associated with assumed similarity" will be specific for
each group (or persons) being analyzed and will depend on
the actual similarity of the judges to the judged.
13The effort to produce refined empathy scores was afailure. See A. H. Hastorf, I.E. Bender, and D.J. Weintraub,"The Influence of Response Patterns on the 'Refined EmpathyScore, '" The Journal of Abnormal and Social Ps cholo ,51(1955), 3 . e 1na 1ty 0 t e so-ca e re 1ned-empathy score to solve the problem is shown by Bernard I.Murstein, "Some Comments on the Measurement of Projectionand Empathy," Journal of Consulting Psychology, 21 (1957),81-82. --
20
That these variables are impure, because of overlap
among them, is a not uncommon event in analysis of this
nature. It is by no means a difficulty that ought to
paralyze research:
It is seldom necessary . . . to measure anyof these variable quantities absolutely. It usuallysuffices to achieve merely relative comparisons onsuch variables, for our propositions most often specify only that a difference in one variable is relatedto a difference in another variable. Given suchrelative measurement, a good many powerful researchdesigns and statistical techniques are at ourdisposal.14
In a sense, the search for "refined" and uncontaminated
measures is unnecessary. The "refinements" sought may be
construed as stages along a continuum. The measures used
in this research are more refined than those employed by
Lerner, but they are nonetheless "impure."lS
2.2 SOME MEASURING TECHNIQUES REJECTED.
A decision to adopt a research design is also a decision
to reject alternative ways of studying role-taking accuracy.
l4George J. McCall and J. L. Simmons, Identities andInteractions, New York, 1966, p. 269.
lSMoreover some correlations are not spurious.L. L. Thurstone, Multiple Factor Analysis, Chicago andLondon, 1947, p. 458, makes the point that correlationbetween stature and intelligence shown by factoranalysis represents reality quite well when thecorrelations occur among children of different ages. Theresearcher, however, needs to be careful about interpretations based on selections of cases.
21
The Kerr-Speroff Empathy Test is a commercial test designed
to measure empathy.16 However, this test in several in-
vestigations has not correlated highly if at all with
other tests that supposedly measure the same skill. 17 For
this reason, it is given a negative review in the Mental
Measurement Yearbook. 18 A further disadvantage is that
it does not tap attitudes in the political realm. The
assumption that role-taking accuracy is general over
several realms of experience is not made in this analysis,
which is concerned solely with the political realm. l9
Therefore, the Kerr-Speroff Empathy Test was rejected.
16W. A. Kerr and B. J. Speroff, The Empathy Test,Chicago, 1954.
17"The failure of the Kerr-Speroff Empathy Test tocorrelate with variables with which it might be expected tocorrelate raises a serious question as to whether it is atest of empathy," according to C. H. Patterson, "A Note onthe Construct Validity of the Concept of Empathy," ThePersonnel and Guidance Journal, 40 (1962), 803-806.---
l80scar Krisen Buros, Editor, The Fifth Mental Measurement Yearbook, Highland Park, N.J., 1959, pp. 120-121.The reviewer Robert L. Thorndike concludes, " ... thistest [the Kerr-Speroff] cannot be recommended as either auseful practical device or a contribution to the descriptionand understanding of an individual.
19The evidence on this point is contradictory, WaymanJ. Grow and Kenneth R. Hammond, "The Generality' of Accuracyand Response Sets in Interpersonal Perception, ' The Journalof Abnormal and Social Psychology, 54 (1957), 384-390,conclude that ~nterpersona1 role-taKing accuracy is notgeneralizable. See also Henry Clay Smith, Sensitivityto People, New York, 1966, pp. 10-11.
22
A movie would provide a stimulus with which persons
could mentally interact. There are a number of studies in
which judges were shown brief films of the people whom they
were called upon to predict. This technique appears to be
especially applicable when people are called upon to judge
other individuals, that is, in testing for sensitivity to
individual differences. The technique has not been applied
to judgements, about generalized others, although there is
no reason why this cannot be done. The filming of a session
of the Hawaiian legislature was beyond the means of this
researcher, however. Moreover, such a technique places
restrictions on the means of obtaining respondents, although,
since voluntary political groups provided respondents, this
limitation is not insurmountable providing the movie is
entertaining enough.
There are a number of projective techniques that
suggest themselves for an exploration of empathic responses.
Two of the most often used are the Rorschach and the
Thematic Apperception Test (TAT). The latter, which would
seem to be most interesting for role-taking purposes, has
an unfortunately low reliability.20 To tap the political
realm, a new set of TAT pictures should probably be de
veloped. Indeed, this technique, in spite of problems of
20See Raymond B. Cattell, The Scientific Analysis ofPersonality, Baltimore, Md., 1965, p. 128.
23
reliability, offers a fine opportunity for validation of
the measures of role-taking accuracy. Using voluntary
political organizations to obtain respondents, however,
made necessary a mailed questionnaire, and the names of
respondents were not known to the researcher.
Finally, there is the psychodrama technique, in which
persons would be called upon to stage a brief performance of
a situation in which they played, in this case, the role of
a legislator, a campaign orator, or a voter. Such a
technique would place great limitations on the selection of
a suitable and willing sample. Again, the rating technique
used in scoring such experiments as these suffers from the
same unreliability as the projective techniques. Cattell
regards the questionnaire as better than a rating technique
for this reason. 21 However, the psychodrama offers an
opportunity to test the socialization of children into
political role-taking accuracy. Such a research aim would
make it highly appropriate.
These are among the chief alternatives to the question-
naire technique used in this study, and in one way or
another they appear less appropriate in the research
questions posed in this project. For that reason, the
procedure in the next section was adopted.
21catte11, Scientific Analysis of Personality, p. 137.Cattell complains about the use of ratings by panels as ameans to validate questionnaire results. Such a validationprocedure is indeed made use of in several of the empathy
24
2.3 THE RESEARCH TECHNIQUE.
The first step of the research was to test the political-
other whose role persons were to predict. A salient
political-other had to be chosen, a person (or group) who
was visible to the respondents and from whom completed
questionnaires could be obtained. It seemed best to tap a
group of persons for a consensus of attitudes rather than
to rely on a single respondent who could not be grantee
anonymity. The state legislators met these criteria of
accessibility and saliency. Their attitudes were obtained
by a questionnaire, a step that will be discussed in detail
in Chapter III. From this questionnaire, items on which
there was consensus among the legislators were cOMpiled.
These items were used in the second portion of the
research to construct a questionnaire combining them with
other political attitude and behavior measurements. The
scores on the role-taking items and on the variables of
interest to this study were derived from these question-
naires, which were mailed to persons in two voluntary
political organizations in Honolulu.
The ensuing analysis proceeded on three levels.
Through a one-way analysis of variance, an effort was made
experiments. It is highly questionable, especially inview of the low scores often attained by "skilled" observerson predictive questionnaires.
25
to ascertain relevant differences between the two groups
as a whole. On what variables did the two groups vary
significantly from each other, ignoring for the moment
the within-groups variance?
The analysis then continued to search for differences
among the individuals without respect to their group des
ignation. In this analysis, the concern was with the
relationships of the role-taking variables to political
variables among a group of politically-aware respondents.
No statistical inference can be made, of course, to the
population of the politically-aware of Hawaii, but the
implications of role-taking relationships extend to this
population.
In the analysis at this second level, the first stage
of data reduction was a Guttman-scaling of the scale
questions in the questionnaires returned by the role-taking22respondents. The scale scores and the other items on the
questionnaire, including the role-taking variables, were
dichotomized and a correlation matrix was obtained.
22Factor analysis was an alternative at this stage, andit was llsed later. It was decided that the Guttman scalesincorporated traditional and readily-identifiable scales thathave been used in many other studies of political attitudes.For a comparison of Guttman scales and a factor analysis ofthe same items, see Jeanne E. Gu11ahorn, MultivariateAtProaches in Survey Data processin¥: Comearisons of Factor,Custer and Guttman Anal ses and 0 Mult~ Ie Re ress~onan anon ca orre at~on et 0 s, Fort Wort, exas,Society of MUltivariate Experimental Psychology, 1967.Also see Jeanne E. Gu11ahorn and John T. Gu11ahorn, "TheUtility of APtP1ying Both Guttman and Factor Analysis toSurvey Data,' Sociometry, 31 (1968), 213-218.
26
Significant interrelationships were noted among the
variables included in the study. This process was a simple
cross-tabulation procedure in which all individuals from
both groups entered the computation. 23
The pattern of relationships in the common factor
space was obtained for all the dichotomized variables. This
was done with the highest row correlations in the principal
diagonal of the matrix to be factored. Rotation was
carried out first to an orthogonal solution. This solution
was then rotated to a biquartimin oblique factor solution
to determine whether a better simple structure was possible.
Multiple regression and correlation was used to
predict to the dependent variables of the study from in
dependent dimensions of the attitude, behavior, and role
taking space to ascertain which of these variables gave a
better prediction. (The raw scores of the role-taking
variables were used in this step.) This was the final
step in the search for relationships among the variables
on the level of the entire group of politically-aware in
dividuals.
Finally, a process combining a distance program with
a subsequent factoring of the distance matrix produced
23These 2X2 tables can be quite revealing. Theyvide information not obtained from any of the othertechniques in this research. See Fred N. Kerlinger,Foundations of Behavioral Research, New York, 1964,espec~a11y Chapter 12.
pro-
27
subgroups from among the politically-aware respondents of
the study. This empirical classification technique yielded
information on relationships among the role-taking and
political variables within subgroups representing different
styles or patterns of political activity.
Thus, the analysis proceeded through three different
levels: comparisons between the two groups, comparisons
among the respondents as a whole, and comparisons among a
number of subgroups empirically derived from among the
respondents. Caution must therefore be exercised through
out the discussion, so that conclusions reached during the
analysis at one level are not assumed to hold for another
level. This is the ecological fallacy, and its reverse,
the individualistic fallacy, that must be constantly
avoided during the course of the investigation. 24
2.4 HYPOTHESES.
There are two major hypotheses which will be con
sidered: 25
24The ecological fallacy is discussed in an initialarticle by William S. Robinson, "Ecolo~ical Correlationsand the Behavior of Individuals," Amer~can SociologicalReview, 15 (1957), 351-57. The problem is well-discussedand explained in Erwin K. Scheuch, "Cross-National Comparisons Using Aggregate Data: Some Substantive andMethodological Problems," Richard L. Merritt and SteinRokkan, editors, Comparing Nations, New Haven and London,1966, pp. 131-167.
2SThe nature of the respondents is such that there isno random sample of any delimitable population. Inferenceto any population, such as "members of voluntary political
28
1. Accurate role-takers will accept the politicalsystem enough to want to work within it as itis presently constituted. Disowning projectorswill tend to be Qersons advocating relativelydrastic changes. Zb Thus, we expect to findthat disowning projection is highly correlatedwith advocacy of change. Accurate role-takerswill not be expected to advocate major changein the political system.
2. Role-taking accuracy is positively related topolitical success as measured subjectively(according to the subject's own assessment)and objectively (as measured according to anoutside criterion). Disowning projectors willhave little feeling of political success andlittle objective political success in comparison. 27
These major hypotheses are formulated according to
results obtained from many non-political studies. "Those
who accept the system work within it and gain experience
organizations in Hawaii," may be made only at the reader'srisk. Since the questionnaires studied are only thosereturned by certain of the members of two political groups,inference cannot be made even to these groups. Statisticalsignificance indicates only relationships that hold amongthe respondents in this particular study.
26"people who anchor themselves at extremes tend to seeothers as more unlike them," according to Leonard Berkowitz,"The Judgmental Process in Personality Functioning,"Psychological Review, ~ (1960), 134.
27See David A. Rogers "Personality Correlates ofSuccessful Role Behavior,~ The Journal of Social Ps~cho1ogy,46 (1957~, 115; Leonard S. Cottrell, Jr., and Rosa11nd F.Dymond, 'The Empathic Responses, A Neglected Field ofResearch," Psychiatry, 12 (1949), 358. Note however thatespecially in po11tics,~ere may be an alternative competing hypothesis: " ... it will be contended that muchof the most efficient collective action occurs . . . undercircumstances which make accurate social perceptionrelatively unimportant," according to Ivan D. Steiner,"Interpersonal Behavior as Influenced by Accuracy of SocialPerception," Psychological Review, 62 (1955), 270.
29
--it pays off to role-take.,,28 This might be a surrnnation
of the results hypothesized above.
In addition to the major hypotheses, there are a
number of secondary results that are predicted. The
citations that follow them provide support for the
propositions:
a.
b.
c.
d.
e.
f.
Persons who have interacted with politicians willhave higher role-taking accuracy.29
Members of the "socializin& professions" will bemore accurate role-takers (salesmen, for example)than members of more "introspective professions"(engineers, for instance).
Role-taking accuracy increases with age. Theyoung tend to project, perhaps in the directionof disowning projection.
Long-time residents of Hawaii among the politically active will be better able to take the roleof Hawaii's legislators.
Younger children of families with more than onechild will have greater role-taking accuracy(when adults), while the only and eldest childrenwill tend to be projectors. 3D
Motivated persons will be better role-takers thanthose with less motivation. Presumably, motivation
28Thomas J. Scheff, "Toward a Sociological Model ofConsensus," American Sociological Review, 32 (1967), 45.
29Ronald Taft, '~ccuracy of Empathic Judgments ofAcquaintances and Strangers," Journal of Personalityand Social Psychology, 1 (1966), 603.
30suggested by Ezra Stotland and Robert E. Dunn,"Empathy, Self-Esteem, and Birth-Order," The Journal ofAbnormal and Social Psychology, 66 (1963), 532, 539.
30
compels us to learn more about others. Thismay hold, however, only for the moderatelymotivated. 31
g.
h.
i.
j .
Men will be more accurate than women as politicalrole-takers.
The politically cynical will be high on disowningprojection, low on role-taking accuracy.
Role-taking accuracy will be positively relatedto equalitarianism and faith in people. 32
Role-taking accuracy will be positively relatedto ego strength.33
k. Accuracy on the two types of prediction questions(attitudes and behavior) will be positively related.
These hypotheses are the result of a study of the
literature, none of it dealing directly, however, with
taking the role of a political-other. Since role-taking
accuracy may be specific to particular groups, it is
necessary that these hypotheses be tested for political
variables and with political others to be predicted. 34
Because multivariate techniques are used in the analysis,
31David E. Berlew, "Interpersonal Sensitivity andMotive Strength," The Journal of Abnormal and SocialPsychology, ~ (1961), 393.
32Rosalind F. Dymond, "Personality and Empathy,"Journal of Consulting pShchology~ 14 (1950), ~49,.foundthe highs on empathy to e trust~ng and equal~tar~an.
33Smith, Sensitivity to People, p. 178.
34I bid., p. 139.
there will be other relationships of interest discovered
during the course of the investigation. The power of
multivariate analysis as opposed to bivariate analysis
techniques ought to be amply demonstrated.
31
.- CHAPTER III
LEGISLATIVE CONSENSUS
3.1 THE CHOICE OF THE POLITICAL-OTHER.
In order to determine whether or not political groups
in Hawaii are able to take the role of the political-other,
it was necessary to define the role of some political
other. This might have been done by defining the po1itica1
other in a number of ways. The other might have been the
governor, the President of the United States, the Congress,
or the Republican Party of the State of Hawaii.
The criteria of choice were salience to the respondents
and access to a questionnaire survey. For Hawaiian volun
tary political organizations, the legislature meets these
criteria. While potential respondents might not be able
to distinguish individuals among the legislators, they
might be expected to have formed an image of the group as
a whole, a role-image. While the governor might be un
willing to submit to a questionnaire on attitudes, the
legislators would be more likely to return such a question
naire, provided anonymity was granted and the results would
characterize the entire legislature rather than individuals
in it. It must be kept in mind, however, that respondents
were tested for sensitivity to generalized others rather
33
than to individua1s,1 and role-taking accuracy with regard
to groups is not necessarily generalizable to accuracy with
regard to individuals.
No single legislator determines the role of legislator,
but together they agree on certain definitions of a common
role. It was necessary to determine the consensus of the
legislators on what comprised certain attitudinal and
behavioral aspects of their ro1e. 2 Moreover, it seemed
undesirable to include items that called for a special
knowledge of state laws or parliamentary procedure.
Presumably, institutional role-taking of this nature would
quite simply be correlated with political experience,
knowledge of politics, and education. To tap role-taking
accuracy in the second sense of role (see Chapter II,
page 11), attitudinal and behavioral items were necessary.
The literature proved rather sparse for the purpose
of framing items for a legislative-consensus questionnaire.
A number of sources provided attitudinal statements, how
ever, and clues from these allowed a choice from among
1For elaboration on these distinctions, see UrieBronfenbrenner, John Harding, and Mary Ga11wey, "TheMeasurement of Skill in Social Perception," in DavidE. McClelland, et a1., Talent and Society, Princeton,N.J., 1958, p. 17.--
2Thus , Erving Goffman, Encounters, Indianapolis,Ind., 1961, p. 93, defines role as Ii ••• the typicalresponse of individuals in a particular position." Hegoes on to distinguish the typical role from the actualperformance of a concrete individual. It might be said thatthe consensus sought is itself the typical response, or role.
34
other commonly-used political attitude items. 3
A three-part questionnaire with a total of 66 questions
and six semantic differential concepts with ten scales each
was mailed to all the current members of the Hawaiian State
legislature, Senate and House (as of November, 1967).4 Of
75 questionnaires mailed,S 45, or 60 per cent, were re-
turned.
3.2 THE CRITERION OF CONSENSUS.
The criterion of consensus used to select the items
to be included in the final questionnaire was 30 or more
responses on either side of the Likert scale on the
consensus questionnaire and 30 or more responses on any
3Useful sources included John B. McConaughy, "CertainPersonality Factors of State Legislators in South carolina,"The American Political Science Review, 44 (1950), 897-903;Donald R. Matthews, "The FolkWays of theUnited StatesSenate~ Conformity to Group Norms and Legislative Effectiveness," The American political Science Review, 53 (1959),1064-1089, and u.s. Senators and The~r World, university ofNorth Carolina, 1960; Robert K. Merton, "BureaucraticStructure and Personality," Social Forces, 18 (1940),560-568; Clem Miller, Member of the House, New York, 1962;Thomas Vernon Smith, The Legislative Way of Life, Chicago,1940; Duane Lockard, "The Tribulations of a State Senator,"in John C. Wahlke and Heinz Eulau, editors, LegislativeBehavior, Glencoe, Ill., 1959, pp. 294-298; Leonard b.Wh~te and Thomas Vernon Smith, Politics and Public Service,New York, 1939.
4The questionnaire, except for the semanticdifferential portion, is reproduced in Appendix I.
Sane House member had just resigned to run for acounty office and was not included.
35
single item of the first nine situation questions. This
figure was derived from the confidence interval of the
percentage using the most conservative ca1cu1ation,6 which
gave + 14.6% at p < .05. Thus, a 50 per cent split would
be at 23 cases. Fifteen per cent of the cases would be
seven. Any split at 30-15 might safely be said to occur
by chance only about five per cent of the time. The
division seemed appropriate for a definition of legislative
consensus.
By this criterion, there were six of the situation
questions that qualified for the final questionnaire, while
39 questions qualified for Likert-type items. This pro-
vided sufficient items for the second role-taking question
naire. The complete questionnaire for the legislators is
reproduced in Appendix I. To avoid problems of level on
the role-taking responses, the Likert items were dichoto
mized, and consensus was defined by "agree-disagree"
disregarding neutral choices. These dichotomized items
and the responses of the legislators are shown in
Table 3.1. As noted in footnote (b), items 13, 23, and
43 were dropped on the final role-taking questionnaire
when a pilot study showed them inadequate in distinguishing
accurate from inaccurate role-takers.
6Hubert M. Blalock, Social Statistics, New York,1960, p. 164.
36
TABLE 3.1 QUESTIONS FROM THE LEGISLATIVEQUESTIONNAIRE SHOWING CONSENSUS
(Only consensual results are shown)
PART I: BEHAVIOR-TYPE QUESTIONSNumber of
Choices
1. I have made a campaign promise to support acertain bill. This won quite a few votes forme. When the bill comes before the legislature, my political party is very much opposedto its passage. I will
d. tell the members of my party about mycampaign commitment and then vote infavor of the bill. 42
2. I have won my first electionlegislature, and the sessionA bill is up for discussion.an expert on the bill, but Iknowledge and support it. I
to the Hawaiihas just opened.
I am by no meanshave a littlewill . . .
d. keep quiet and vote for the bill.
3. At a private dinner party, I find that I havebeen seated next to a long-time politicalenemy with whom I have been feuding recently.I will . . .
d. sit next to my political enemy and usethe occasion to get on good terms withhim.
4. A letter comes into my office written inpencil and in very bad grammar. It requestsinformation that will take some time tocollect. I will . . .
34
34
b. have my secretary obtain the informationand write a full reply. 35
5. Mr. Jones is a lobbyist with whom I am friendlyand who has helped me a few times in the past.He argues in favor of a bill, but he doesn'tconvince me of its worth. I will . . .
a. tell Mr. Jones how I intend to vote and why. 42
37
Table 3.1 (continued)
PART I: BEHAVIOR-TYPE QUESTIONSNumber ofChoices
6. A bill of mine is before a committee of whichI am a member. I can foresee that it willprobably lack several votes of being approved.Before the day of the vote, a member of theopposition on the committee suggests that Iaccept an amendment to the bill. This amendment will make the bill less effective but willprobably allow it to pass. I will ...
b. discuss his amendment with other interestedgroups and accept it if most of the othersgo along. 34
PART II: CONSENSUAL LIKERT ITEMS (Dichotomized)
Agree Neutral Disagree
2. A civil service exam would bea better way than an electioncampaign for choosing ourlegislators. 2 1 32
4. Elections are too heated formy taste. 2 43
5. It is quite natural to getopposing opinions from thesame set of facts. 40 5
6. You cannot really be surewhether an opinion is trueor not unless people are freeto argue against it. 39 2 4
9. I feel uncomfortable beinglaughed at. 34 11
11. A man who won't compromiseisn't a good citizen 6 6 33
38
Table 3.1 (continued)
PART II: CONSENSUAL LIKERT ITEMS (Dichotomized)
Agree Neutral Disagree
12. It's better to settle issuessomehow than to get themsettled absolutely correctly.
13. I like praise from others. b
31
39
3
2
11
3
15. It is self-defeating toembarrass a political opponentneedlessly. 43
17. People ought to make clear tolegislators how they want themto vote. 30
20. Most of those who disagree withmy political views are simplyuninformed. 7
21. It is good to be popular evenwith opponents, because they maybe on my side later. 38
22. Private and informal discussionsometimes clears up difficultproblems that public debatewould not solve. 45
23. The essence of good politics isteamwork. b 40
24. I'm almost never impolite topeople. 30
26. Special interests are usuallyagainst the public interest. 11
27. In a better age, we will beable to do without politicians. 6
28. A person who hides behind thelaws when he is questionedabout his activities doesn'tdeserve much consideration. a 14
29. I enjoy political arguments. 34
1
3
2
2
1
4
3
1
1
5
1
12
36
5
4
11
31
38
30
3
39
Table 3.1 (continued)
PART II: CONSENSUAL LIKERT ITEMS (Dichotomized)
Agree Neutral Disagree
31. If all my friends were university graduates, I would hidp thefact that I was a commonlaborer. 4 41
32. Sometimes I supportprinciples I don't entirelyagree with.
33. I don't mind a politician'smethods if he gets the rightthings done.
34. It's good politics to praiseyour opponents when you can.
35. Most politicians are inpolitics to be of "publicservice."
36. I like to be able to plan aset routine for my dailywork.
37. All elections ought to benonpartisan.
38. I get along with personsfrom all levels of society.
40. When I get home from work, Ilike to forget my job andrelax
43. I'm glad to live in theseexciting times. b
45. If I really wanted a lawpassed, it would be cowardlyto accept a watered-downversion of it.
47. It's wrong for people to putpressure on legislators to votetheir way.
33
12
32
32
32
5
41
39
45
8
4
1
1
6
3
1
2
1
1
11
31
7
10
13
38
2
6
36
40
40
Table 3.1 (continued)
PART II: CONSENSUAL LIKERT ITEMS (Dichotomized)
Agree Neutral Disagree
48. Those people who hate our wayof life should have a chanceto speak and be heard.
49. In politics, it's every manfor himself.
50. Politicians are usually"bitterly attacking" or"reacting violently."
51. All major interests in ournation are equally worthy ofconsideration.
53. I'd rather stand off and lookat events from a distance
54. The major aim of governmentis efficiency.
40
12
3
38
5
13
1
1
4
2
4
1
3
32
38
5
36
31
55. The records of legislators aretoo little known to the public. 43 2
56. Politics are dull in comparisonwith sports. 6 1 38
aThis question was dropped as a result of the checkfor the mode of a respondent runs-group. See Chapter III,p. 43.
bThese questions were dropped on the final questionnaire when a pilot study showed that they did little todistinguish accurate from inaccurate role-takers. Item23 is an interesting case. Respondents in a pilot studyanswered this "agree" both for themselves and forlegislators even though many of them had agreed that "Inpolitics, it's every man for himself"! Item 23 is justthe type of question that elicits acquiescence.
41
No argument is advanced that legislators following
a consensus most closely in their own attitudes are more
successful than maverick legislators. It may be more
important that a legislator knows when his own attitudes
are consensual or not. Nor can an argument be supported
that the responses given on the questionnaire are what the
typical legislator really feels or would do in a situation.
For this reason respondents were later instructed to
respond to a series of questions as if they were typical
legislators answering a questionnaire.
3.3 REPRESENTATIVENESS OF THE LEGISLATIVE RESPONDENTS.
Since 30 legislators did not return questionnaires, it
was advisable to check whether there were significant dif
ferences between the respondents and the non-respondents.
Information on the non-respondents was obtained from
Who's Who in Government, State of Hawaii, published for the
current (1967) legislative session by the Hawaii Chamber of
Commerce.
The non-respondents were distinguishable because the
questionnaires were numbered and the names marked off as
they were returned. Two identifying numbers, however, were
marked out by the respondents, although the questions had
been answered. Since these two legislators were uniden
tifiable, they were counted necessarily as non-respondents.
The two groups, respondents and non-respondents, were
divided for differences on the following: political party,
42
members from Oahu versus members from the Outer Islands,
Oriental versus non-Oriental, length of service (service
with territorial legislature versus newer members), and
age. On none of these divisions was there a statistically
significant chi-square (p < .10) between the respondents
and non-respondents. Age, however, approached significance
(p < .20).
A Wa1d-Wolfowitz runs test, which is addressed to any
sort of difference between groups, including central
tendency, skewness, and variability, showed a z-score of
-1.71 for the age attribute, which is significant for a
one-tail test at p < .05. The apparent difficulty was a
run in non-respondent birth dates from 1894 to 1906 (six
non-respondents) and two runs in the respondents from 1911
to 1915 (six respondents) and from 1916 to 1920 (11
respondents).
Since two of the larger runs were respondents, they
could be checked against the questions chosen on the
consensus criterion. On no item with the exception of
Question 28 was the mode of the split of this runs-group
on the side opposite to the consensual answer. Question 28
(as numbered on the complete questionnaire, Appendix I)
showed a dissensus of nine (out of 17) from the runs-group,
and it was therefore dropped from the consensus questions. 7
7Question 11 showed a dissensus of seven, Questions 24and 35 showed a dissensus of six each, and Question 36 had
43
One of the schisms in a legislature might well be that
of political partisanship. There were 13 Republicans among
those returning their questionnaire (a proportion reflect
ing quite accurately the proportion in the legislature as a
whole). Since there were so few Republicans, however, dif
ferences resulting from partisan splits would appear as a
consensus with our criterion. It would be fallacious to
argue, however, that partisan attitudes necessarily re-
fleeted a consensus on the role of the legislator as such.
Therefore, the distribution of Republicans on the questions
was examined.
On none of the consensus questions did a Republican
mode appear to differ from that of the Democrats. None of
the questions, therefore, were dropped because they tapped
a partisan attitude or behavior. 8 The foregoing arguments
appear sufficient to defend the criterion of consensus which
defined the questions used on the final questionnaire for
role-taking respondents.
It might still be argued that the really crucial
difference between respondents and non-respondents was that
five dissenters from this runs-group. These questionswere retained.
8Since the decision was made to drop the semantic differential questions in the final questionnaire, no more willbe said about them. However, the concept Governor was oneon which there was"an obvious party split. In spite ofclear instructions on the abstract nature of the concepts,the legislators were not able to avoid regarding the concept Governor as synonymous with the present incumbent.
44
the latter had not returned their questionnaire and that
had they been tested on the questionnaire their answers
might have differed significantly from those of the
respondents. However, since the responding group was
larger than the non-responding one, it may be argued that
no change would be likely on those questions chosen on the
basis of a criterion of consensus. In fact, the consensus
of the legislature seemed to be that one returns a question
naire of the type mailed them.
CHAPTER IV
THE NATURE OF THE RESPONSE GROUPS
4.1 THE CHOICE OF RESPONDENTS.
The major hypotheses with which this research is con
cerned made it desirable to obtain respondents for whom
the legislature would be part of their perceptual domain. 1
It seemed proper to seek respondents, therefore, from
political organizations which they had joined voluntarily,
presumably to accomplish some political aim. A union, as
an example of one organization that partakes in politics,
would not have been appropriate for the reason that many
of its members join because they have to and not because
they happen to be politically active. Although it would
be interesting to obtain union respondents on these ro1e
taking variables, they might not be expected to be as
politically aware as members of other groups. Legislators
mayor may not be significant others for them.
Therefore, organizations were sought whose membership
was more likely to be active in the political arena. An
example of such a group might be the NAACP, which persons
1She1don Stryker, "Role-Taking Accuracy and Adjustment,"Sociometry, 20 (1957), 287, states categorically that ro1etaking requires that the predictor and other be mutuallyimplicated in an ongoing social situation.
join in order to accomplish the aim of extending civil
rights to all races. People in such a group are likely
to regard the legislature as a part of their political
domain, because it is the legislature, state or federal,
that passes or refuses to pass civil rights 1egis1ation. 2
Among groups of this nature, it was necessary in
order to test the hypothesis regarding advocacy of change
that respondents be obtained ranging from those who ad
vocated considerable change in the political system to
those who were relatively satisfied with the system as it
was. Ideally, these respondents should range across the
political spectrum from the extreme left, through the
center, and to the extreme right. The two groups from
which respondents were drawn did cover a wide portion of
the spectrum from left to right. Not a few of them ex-
pressed a desire for a maximum of political change on3Questions 2.13 and 2.14.
46
2The NAACP was not tapped for respondents in thisstudy. No group names will be used, and they will bediscussed only in general terms. Some respondents expressed the fear that their responses would be taken forthe position of the group.
3A11 question numbers will refer to the numbering asit appeared on the questionnaire, which is reproduced inAppendix II. Appendix III provides a definition of allthe variables that were used in the analysis. Hereafter,all questions will be referred to by pafte and number, as2.13, '\V'hich reads "page 2, ques tion 13. I
47
It proved impossible, however, to cover the political
spectrum completely. Many voluntary organizations prohibit
in their by-laws the distribution of questionnaires. Al-
though respondents were guaranteed individual anonYmity,
experience showed the desirability of granting anonYmity
to the organization as well; respondents did not want their
answers to represent the official stand of the organization
to which they belonged. The latter guarantee ought to have
been made explicitly in the introduction to the question
naire, and it was conveyed to Group B in the form of a
cover letter. 4 Naturally, the same anonYmity was extended
to the first group. In spite of these precautions, however,
the ideal purpose of selection was not achieved. S
4Reproduced in Appendix II.
SAlthough enough respondents among the politicallyaware were tapped for the purpose of this research, themethod of reaching respondents cannot be recommended toany researcher whose access to such organizations must befrom the outside. Some lesson might be salvaged from thisexperience, however. A possible approach to a voluntarypolitical organization's membership might be through aninitial letter to all the members explaining the nature ofthe research, assuring anonYmity for the individual and theorganization, and requesting those who are interested inreceiving such a questionnaire to return an addressed cardto the researcher. A questionnaire would then be mailedto those requesting one.
This approach ought to eliminate criticism frompersons who do not wish to receive a questionnaire in themails under any circumstances. Unfortunately, the techniqueeliminates anonYmity for some, and the objection oforganizations to the distribution, in effect, of even aportion of their mailing list through the addressed requestsremains a problem.
It is also unfortunate that no statistical inferencesmay be made either to the group or to the population of the
48
4.2 DESCRIPTION OF GROUP A.
Eighty-five questionnaires were mailed to members of
Group A. Ultimately, 47 of these, or 54 per cent, were
completed and returned. Since not all the members were sent
a questionnaire, approximately every third or fourth name
was chosen from the mailing list. In cases where both the
husband and wife held a joint membership, the selection of
respondent was left to chance. Mailing lists of voluntary
political organizations are not given to non-members, and
reliance had to be placed on secretarial help within the
organization to mail the questionnaires, cover letters, and
return envelope. This lack of control over the question
naires is a distinct disadvantage to the researcher. It is
not possible, for example, for a follow-up study in order
to make a validity check on certain of the questions.
However, the questionnaire had been prepared for self
administration, and a pilot study showed that students in
a beginning political science course had no difficulty in
following the directions. Presumably, the respondents from
voluntary organizations had no difficulty either, and an
indication of their sophistication can be seen from the
responses to Question 13.2 (Education), among others in
Table 4.1.
politically-aware. This may still offer a group ofrespondents preferable for many reasons to the usual sampleof college students. For, even if a random sample isobtained of a beginning political science class inferencemay go no further than that class.
49
Group A was tapped in order to reach respondents who
would generally range from moderately to further left
politically. An indication of the validity of this surmise
is given by the answers to Question 3.19. Some of the
issues mentioned do not place the individuals politically;
the l8-year-old vote, constitutional convention, laws on
alcoholism, taxes, education, air pollution, and model
cities are issues that enlist all sections of the political
spectrum. Civil rights and the civic ethics commission do
not offer a criterion for placement. However, the issue
most often mentioned was opposition to American partici-
pation in the war in Vietnam, followed by draft resistance,
and support for Senator Eugene McCarthy's campaign for the
Democratic nomination to the presidency. In the political
climate of 1968, these issues mark this group as left of
center. Other suggestive issues supporting this conclusion
were the open housing bill, disarmament, poverty, and the
Dr. Oliver Lee case. 6
The results presented in Table 4.1 provide a charac
terization of the respondents of Group A. (By Group A
hereafter will be meant respondents from Group A.) Of
course, no inference about the entire organization may
6The last-named case concerned questions of freespeech, academic freedom, and due process with regard togranting and taking away tenure. Those involved in thecase were not necessarily from the political left, but amajority of them probably were.
50
be reliably drawn from these figures, since a mailed
questionnaire is necessarily subject to bias. These re
sults show that the respondents were indeed a highly
sophisticated group politically. This is shown not only
by the sociological variables but by the participation and
political knowledge indices as well. Yet, except possibly
for the knowledge index, there are enough respondents from
the lower ranges for contrast. If important distinctions
among them are forthcoming from the role-taking variables,
then it may be surmised that these distinctions might be
even greater among both these respondents and those of a
broader portion of the population.
Although the mean of role-taking accuracy is high,
there are also individuals among the lower range of
accuracy, as indicated by the range and the standard de
viation. The same is true of the other role-taking variables.
The mean for advocacy of change is somewhat higher than
expected for Group A, but there are those who fall into the
range of more moderate change and a few in the lower levels
of desire for change. It may be concluded that Group A
provided the type of respondents that were desired for a
test of the hypotheses.
It would not be surprising to discover in a relatively
stable and mature political society that political groups
with pretensions to successful operations in the extant
political system are similar in their attitudes to the
Frequency of responseGroup A Group B
(N=47) (N=24)
TABLE 4. 1 DESCRIPTION OF RES PONDENTS, GROUPS A AND B
PART I. FREQUENCIES a
Questionnaire Itemb Variable namec and responses
1.7 HELOFF10 1437 10
33 1414 10
112 623 11
2 19 6
15 420 1311 7- -1
INFSEL
ACQLEG
YesNo
YesNo
A lotMore than most peopleAbout averageLess than most peopleAlmost none
POLINT
1.8
1.9
2.10A lotMore than most peopleAbout averageLess than most peopleAlmost none
aCne Group A respondent omitted the questions on page 13 of the questionnaire.
bRead 1.7 as page 1, question 7.
cVariable names are abbreviated in all tables and figures.Appendix III ~vill give the full description and names of all variables and attributes.
U1t--'
TABLE 4.1 DESCRIPTION OF RESPONDENTS, GROUPS A AND B (continued)
PART I. FREQUENCIES
Questionnaire Item Variable name and responsesFrequency of response
Group A Group B(N=47) (N=24)
2.11
2.12
2.14
ASPIREA lot moreMore than I have nowAbout as much as I have nm.;Don't really care
INFGRPA lotMore than most other political groupsAbout the same as most other political
groupsLess than most other political groupsAlmost none
CNGVRBA great dealA lot, but not too muchA moderate amountMore than a littleVery littleI like things as they are now
102410
3
214
1811
221110
22
514
41
22
1541
64
1022
U1N
TABLE 4. 1 DESCRIPTION OF RES PONDENTS, GROUPS A AND B (con t inued)
PART I. FREQUENCIES
Questionnaire Item Variable name and responsesFrequency of response
Group A Group B(N=47) (N=24)
2.15
2.16
3.17
3.18
SUCSELA good deal of successSome successPossibly a little successAlmost no success
SUCGRPA good deal of successSome successPossibly a little successAlmost no success
POSOFFYesNo
ISSACTYesNo
~
21224
8
72612
1928
2522
14
109
28
104
915
717
V1VJ
TABLE 4.1 DESCRIPTION OF RESPONDENTS, GROUPS A AND B (continued)
PART I. FREQUENCIES
Frequency of responseQuestionnaire Item Variable name and responses Group A Group B
(N=47) (N=24)
3.20-26 POLKNO7 18 66 10 45 9 3
Number correct: 4 6 53 3 42 1 210
11-12.1-6 BSCORE6 2 55 15 64 6 4
Number correct: 3 11 32 6 31 5 30 2
13.1 SEXMale 34 13Female 12 11
13.2 EDUCATHigh school and below 1 3Above high school 45 21
VI+:'-
TABLE 4.1 DESCRIPTION OF RESPONDENTS, GROUPS A AND B (continued)
PART I. FREQUENCIES
Questionnaire Item Variable name and responsesFrequency of response
Group A Group B(N=47) (N=24)
13.3
13.4
13.5-6
13.7
RESIDETen years or lessOver ten years
EMPJOBEmpathicNonempathic
ELDONLEldest and only childYounger child
AGE20-3031-4041-5051-6061+
2224
1333
3016
81411
76
1113
915
168
1572
V1V1
TABLE 4.1 DESCRIPTION OF RESPONDENTS, GROUPS A AND B,
PART II. DESCRIPTIVE STATISTICS a
Variab1eb Group A (N=45) Group B (N=23)Range Range
name Mean S.D. Max Min Mean S.D. Max Min
PARTIC 8.49 3.46 15 1 6.48 3.85 13 0INFSEL 2.82 1. 07 5 1 2.65 1.11 4 1POLINT 4.02 0.89 5 1 3.87 0.69 5 3ASPIRE 2.84 0.82 4 1 3.00 0.74 4 1INFGRP 3.02 1.08 5 1 3.04 0.88 5 1CHANGE 7.36 2.42 10 1 5.74 1.91 9 1SUCSEL 2.13 0.84 4 1 1.83 0.83 4 1SUCGRP 2.76 0.88 4 1 2.39 0.84 4 1POLKNO 5.69 1.35 7 2 4.87 1.74 7 2BSCORE 3.40 1.63 6 0 3.83 1.70 6 1ROLEAC 26.82 3.77 34 17 27.00 3.79 33 19ASSUME 25.24 5.25 34 19 25.39 4.62 33 18ACTUAL 27.91 3.77 33 20 28.13 2.24 33 25PROJEC 13.60 5.01 25 1 14.87 4.51 26 7
aTwo cases for Group A and one case for Group B are omitted.
bSee Appendix III for the full names and definitions of variables.
U1(j\
57
legislators. A separate analysis of the role-questions of
the two groups of respondents shows this to be true.
As can be seen from Table 4.2, Group A is similar to
the legislators as shown by the responses on the role
questions for "self." Only on Questions 11, 14, 22, 25,
and 31 do the respondents in Group A not fall significantly
on the same side of the agree-disagree dichotomy as the
legislators. 7 In no case, is the mode of the respondents
significantly (statistically) on a side opposite to the
legislators.
The questions on which the respondents deviate the
most from the legislators' consensus are:
_.11. It is good to be popular even with
opponents, because they may be on myside later.
14. Special interests are usually againstthe public interest.
22. I like to be able to plan a set routinefor my daily work.
25. When I get home from work, I like toforget my job and relax.
31. All major interests in our nation areequally worthy of consideration.
Group A is a relatively issue-oriented group. It is
possible that the popularity indicated in Question 11 is not
as consonant with issue-activity as it is with garnering
7Question 5 was a borderline case.
58
votes. A measure of combativeness against "interest"
groups is also apparent. As close as these respondents are
to the legislators, their situation differs. It is less
important to placate all interests when the group is en
gaged in issue tactics. The remainiug questions are more
indicative of possible personality differences between
legislators, who may be more jealous of what leisure they
have, and persons who engage in issue activity for its own
sake.
It may be seen from the "self" answers for Group A
that the assimilative projector will be more successful
in accurate role-taking than will the disowning projector.
In a factor analysis, it is predictable that accurate role
taking will load positively on the same factor as assumed
similarity, while the measure of disowning projection will
load on the same factor, but negatively. It remains for
that analysis to indicate with what other variables these
may vary independently of one another.
It does not follow, however, that a group close to
the legislators in their attitudes should of necessity be
accurate role-takers. In case they are not, they will
then be necessarily high as disowning projectors.
In fact, Group A respondents were able as a group to
clearly designate the correct responses of the legislators
on 29 attitude questions. On six other questions, the mode
of the respondents was not significantly the same as that
59
TABLE 4.2 ROLE QUESTIONS ANSWERED FOR SELF BY GROUPS A AND B
Questiona Group A Group B Legis1atorsb
number Agree Disagree Agree Disagree Agree Disagree
1 5 42 1 23 0 X2 4 43 1 23 0 X3 42 5 23 1 X 04 40 7 19 5 X 05 29 18 19 5 X 06 7 40 2 22 0 X7+ 34 13 8 16 X 08 39 8 23 1 X 09 46 1 24 X 0
10 9 38 24 0 X11·k 23 24 17 7 X 012 45 2 22 2 X 013 33 14 20 4 X 014·k 27 20 7 17 0 X15 6 41 3 21 0 X16 39 8 19 5 X 017 3 44 1 23 0 X18 32 15 15 9 X 019 6 41 4 20 0 X20 40 7 17 7 X 021 32 15 13 11 X 022·k 19 28 16 8 X 023 5 42 3 21 0 X24 46 1 21 3 X 025 24 23 19 5 X 026 12 35 8 16 0 X27 4 43 1 23 0 X28 46 1 22 2 X 029 7 40 11 13 0 X30 2 45 3 21 0 X31 26 21 14 10 X 032 13 34 4 20 0 X33 4 43 3 21 0 X34 34 8 18 6 X 035 4 43 3 21 0 X
aThese are the 35 questions, pages 10-11 of the question-naire, Appendix II, as they were answered for the self inthe scale portion of the questionnaire, Part II, pp. 4-9.
bThe mode (consensus) is indicated by an "X."*On these ~uestions Group A differed in mode from the
answers 0 the legislators.+On these ~uestions Group B differed in mode from theanswers 0 the legislators.
60
of the legislators, and on one of these questions, Question
6, the mode of response was significantly inaccurate.
Table 4.3 gives these totals.
The following questions were most difficult for role
takers in Group A:
6. A man who won't compromise isn't a goodcitizen.
10. Most of those who disagree with my politicalviews are simply uninformed.
19. I don't mind a politician's methods if hegets the right things done.
25. When I get home from work, I like to forgetmy job and relax.
29. In politics, it's every man for himself.
31. All major interests in our nation areequally worthy of consideration.
The errors appeared to be evident in questions in which
some measure of political toughness was attributed to
legislators, Questions 19 and 29, which they did not say
they had. Question 6 may indicate an idea of legislators
as being more prone to compromise than they are. On
Questions 25 and 31, there may have been a larger measure
of projection onto the legislators of attitudes on which
the respondents differed most from them (see Table 4.2).
Five questions were chosen from the responses of the
legislators as examples of a lack of consensus within the
legislature. On all five of the questions the respondents
of Group A were as a group significantly convinced that the
TABLE 4.3 ROLE QUESTIONS ANSWERED AS "LEGISLATORS" BY GROUPS A AND B
Questiona Group A Group B Legislatorsnumber Agree Disagree Agree Disagree Agree Disagree
1 2 45 - 24 0 X2 6 41 2 22 0 X3 41 6 22 2 X 04 35 12 21 3 X 05+ 31 16 11 13 X 06* 31 15 9 15 0 X7 45 2 18 6 X 08 37 10 16 8 X 09 41 6 23 1 X 0
10 23 24 9 15 0 X11 41 6 21 3 X 012 45 2 24 - X 013 41 5 20 4 X 014 14 33 3 20 0 X15 2 45 - 24 0 X16 44 3 21 3 X 017 14 33 7 17 0 X18 41 5 21 3 X 019* 28 19 12 12 0 X20 40 7 14 9 X 021 42 5 19 5 X 0
aThese are the answers respondents gave as if they were legislators, Part III, pages10-11 of the questionnaire. The last five of the questions were non-census questionsaccording to the legislators' own responses. See Appendix for the questions in full.
*This indicates a wrong mode for the guesses of Group A+This indicates a wrong mode for the guesses of Group B
'"r-'
Table 4.3 Role Questions Answered as "Legislators" by Groups A and B (continued)
Question Group A Group B Legislatorsnumber Agree Disagree Agree Disagree Agree Disagree
22 31 15 22 2 X 023 1 46 4 20 0 X24 46 1 24 - X 025+* 19 28 9 15 X 026 6 41 9 15 0 X27 13 34 8 16 0 X28 33 14 18 6 X 029+* 24 23 13 11 0 X30 7 40 1 23 0 X31 25 22 19 5 X 032 10 37 2 22 0 X
33 13 34 8 16 0 X34+ 32 15 11 13 X 035 3 44 1 23 0 X
(36) 41 6 23 1 NC NC(37) 42 5 21 3 NC NC(38) 40 7 20 4 NC NC(39) 40 7 16 7 NC NC(40) 44 1 20 2 NC NC
*This indicates a wrong mode for the guesses of Group A.+This indicates a ~vrong mode for the guesses of Group B.
0'\N
63
correct guess for consensus in all cases was "agree."
This is a clue as to the nature of role-taking in predicting
the generalized-other. It is akin to stereotype accuracy,
in which the individual's assessments are made in broad
terms of experience rather than on an emotional feeling
himself-into-the-others'-shoes that would be evidence of
empathy.
This result indicates that a more searching question
naire would have offered three choices for the respondents;
besides the "agree-disagree" choices offered on the role
taking questions, there ought to have been a "no consensus"
choice. The questionnaire might then have been a somewhat
better test of accuracy. Such a response choice ought to
be part of any replication of this research project. Care
must be taken, of course, to select only those questions
dividing the role-others equally.
4.3 DESCRIPTION OF GROUP B.
One-hundred-and-five questionnaires were provided for
distribution among members of Group B. Ultimately, 24 of
these, or somewhat less than 23 per cent, were completed and
returned. Group B was chosen in order to tap the moderate
right-wing, and as this section of the political spectrum
is less demonstrative and active than the left wing, given
the present political climate, it is possible that this is
the explanation for the reduced percentage of returns.
64
The issues mentioned by the respondents of Group B
included 18-year-01d voting, Diamond Head and a Ka1akaua
mall (both urban conservation issues), removal of the
four-per-cent tax on food and drugs, and foreign travel
tax. These issues indicate, if anything, that the group
includes the moderate right. Omitted are the issues of
greatest interest to the generally left-wing respondents
of Group A. It is interesting that respondents from
Group B did not mention as an "issue" working for the
nomination of any political candidates. The moderate
left, on the other hand, regarded support and work for
Senator McCarthy's candidature as an "issue."
Table 4.1 presents a characterization of the
respondents from Group B. The general conclusion is that
this group, like Group A, is politically sophisticated.
Group B does not emerge as issue active as the first group,
nor does it appear so high in political knowledge or
political participation. Respondents do not differ in
sociological variables such as profession and education.
Table 4.2 shows that Group B respondents are quite
similar in their responses on the attitude-type questions
to the legislators. Possible differences might be discerned
from Questions 7, 21, 29, and 31. In Question 7 the
respondents in Group B fall in a mode on the opposite side
from both the legislators and Group A.
65
The questions on which Group B diverged most from the
legislators included:
7. It's better to settle issues somehowthan to get them settled absolutelycorrectly.
21. Most politicians are in politics to be ofpublic service.
29. In politics, it's every man for himself.
31. All major interests in our nation areequally worthy of consideration.
Questions 7 and 31 indicate a greater degree of attachment
to principle than to political expediency than is apparent
among law-makers, who must make deals as a normal part of
their task if they are to pass bills. On the other hand,
Question 14 concerning "special interests" did not cause
a perceptible divergence from the legislators. It is
possible that "special interests" were construed as economic
lobbies, the normal operation of which ought not to perturb
the moderate right, which is generally friendly to business.
Questions 21 and 29 indicate an amount of cynicism with
regard to the political motives of politicians. The dif-
ference indicated on the last question was also remarked
for Group A.
Like respondents of Group A, those of Group B were
fairly accurate as a group in their answers as simulators
of the legislative role. Questions 5, 25, 29, and 34 show
most of the respondents falling into a mode on the inaccurate
side of the answers, while the group divides evenly on
Question 19.
66
These questions were:
5. I feel uncomfortable being laughed at.
25. When I get home from work, I like toforget my job and relax.
29. In politics, it's every man for himself.
34. The records of legislators are too littleknown to the public.
Questions 5 and 34 show Group B differing from both the
legislators and the guesses of Group A. The first question
may indicate an idea that legislators are more thick-
skinned than they really are, while Questions 29 and 34 may
indicate some amount of political cynicism, which was
possibly present in some self-attitudes, as was shown. Both
Groups A and B differed from the legislators on Questions
25 and 29. On the first, there may be an image of the
politician as hyperactive, while the second question may
suggest a belief that politics is even more competitive
than it appears to legislators. The respondents may be
conceiving of politics as a struggle for issues and
elections, while the legislators may have had in mind the
teamwork necessary to win political offices.
Both Groups A and B emerge as very accurate role
takers, and this concurs with the expectations that led to
the choice of respondents from this portion of the political
population. The groups are not entirely similar, however,
as the next section will show. The picture of role-taking
shows a large consensus in the United States on general
67
principles and political style, at least on the broad
portion of the political spectrum tapped. This sub
stantiates the results reported by Gabriel A. Almond and
Sidney Verba, of course. 8 It remains to be seen whether
within this particularly active group of respondents there
exist relationships such as those hypothesized between the
role and the other variables.
4.4 BE~~EN GROUPS DIFFERENCES.
The respondents from the two groups were compared for
differences by a one-way analysis of variance. Role
accuracy, assumed similarity, actual similarity, and dis
owning projection were not significantly different between
the two groups. The variance of the role variables is con
tained not in the between-groups variance but in the within
groups variance.
The groups did have differences, however, and the major
one was advocacy of change, as shown in Table 4.5. The
commonsense notion that it is the left-wing which advocates
the greatest amount of change in the present system is
vindicated, not unexpectedly.
In addition, Group A was significantly higher on the
participation index and on the knowledge index as well.
8The Civic Culture, Princeton, N.J., 1963.
68
Group A was more than the other group oriented toward issue
activity, as can be seen from Table 4.1. Issue activity
indicates motives that find outlet in greater participation.
In addition, the very nature of Group A as more change
oriented than Group B makes it probable that individuals
would be plunged into more types of participation and with
more intensity than individuals in the less change-oriented
group. When one wants to change the system in a major way,
he expects to work harder at it than those that want to
preserve most aspects of the system.
The knowledge index indicated that on the whole the
individuals from both groups were politically aware to a
far greater degree than is reported from surveys of a
cross section of the population. The questions asked were
not intended to be difficult, and it was necessary to be
stringent in talleying the score; for example, all of
Hawaii's representatives had to be named for a point to be
given on Question 3.23.
The difference on the knowledge index may be part of
the issue-activity and change-orientation of members of
Group A, for whom it is important to know the answers to
the knowledge questions. The desire for change offers a
strong motivation to keep up to date on the news, and most
members of Group A were aware of the recent change that
occurred in the Department of Defense, to which a new
secretary had just been appointed. It is also possible
TABLE 4.4 ANALYSIS OF VARIANCE BETWEEN GROUPS A AND B,
PARTICIPATION INDEX (NA = 47, NB = 23)a
Source of Sum of Degrees of Mean squarevariation squares freedom variance
Total 936.28 69
Between groups 52.33 1 52.33
Within groups 883.95 68 12.999
F ratio = 4.03 p £....05
aCne respondent in Group B was not included in this analysis.
0'\0
TABLE 4.5 ANALYSIS OF VARIANCE BETWEEN GROUPS A AND B,
ADVOCACY OF CHANGE (NA = 47, NB = 23)
Source of Sum of Degrees of Mean squarevariation squares freedom variance
Total 380.58 69
Between groups 39.59 1 39.59
Within groups 240.99 68 5.01
F ratio = 7.90 p L... .01
-...Jo
TABLE 4.6 ANALYSIS OF VARIANCE BETWEEN GROUPS A AND B,
POLITICAL KNOWLEDGE INDEX ( NA = 47, NB = 23 )
Source of Sum of Degrees of Mean squarevariation squares freedom variance
Total 164.80 69
Between groups 9.64 1 9.64
Within groups 155.16 68 2.28
F ratio = 4.22 p <:... .05
-...J.....
72
that some of the difference between groups on political
knowledge was the result of a number of newcomers to
Hawaii in Group B, who had not yet learned the details
of local politics for Questions 3.21-23.
The role variables, however, on the group level did
not distinguish Group A from Group B. If role variables
are related to other variables and attributes of political
interest, it is evidently not on the level tested for in
this section. The search for the correlates to role
accuracy and disowning projection must proceed on another
level in order to tap the within-groups variance. This
search will be pursued in the next chapter.
CHAPTER V
PRELIMINARY COMMENTS ON THE ROLE-TAKING AND
DEPENDENT VARIABLES AND CROSS-TABULATIONS
5.1 SCALES.
The limited number of respondents in this study made
appropriate some technique other than factor analysis to
produce attitudinal dimensions. Besides the 35 questions
for which the respondent answered the role questions for
himself, there were 82 other attitudinal items on pages
four through nine of the questionnaire. These questions
were selected as politically pertinent from a variety of
sources in which they had been shown to scale according to
Guttman's technique for testing for unidimensional scales.
Other items, as for the radicalism scale, were original.
Just as with factor analysis, however, scaling depends not
only on the items to be scaled, but on the group selected
as respondents as well. As observed in the Appendix III,
certain of the scales do not have ideal marginals, for the
respondents ranked too high or too low in general on cer
tain attitudes. On at least two groups of questions,
scales were not possible, and indexes were formed for
authoritarianism and apolitical tendencies.
The Guttman technique also made a positive contri
bution. Items on the questionnaire have been used on many
74
occasions in studies of political behavior and attitudes,
and researchers are familiar with the scales derived from
them, such as political efficacy, flexibility, and se1f
society. On the other hand, a factor analysis is likely
to produce factors loaded with items that might not be
comparable in other studies. This state of affairs need
not be irremediable, of course, for just as Raymond Cattell
has identified 16 or more dimensions of personality, so
might political scientists isolate the dimensions of
political behavior. l
Thus, rep1icabi1ity is another argument for using the
more common Guttman scaling technique. The process of data
reduction was accomplished using the BMDOSS Guttman Scale
Number 1. 2 The items on the questionnaire were entered as
dichotomous agree-disagree items, and those with a marked
unanimity of response were dropped. The minimum co-
efficient of reproducibility accepted was 0.89, and the
minimum number of items for each scale was four.
An attempt to experiment with a program using Lingoes'
empirical technique for scaling the entire set of 82 (and
1Raymond B. Cattell, The Scientific Analysis ofPersonality. Baltimore, Md., 1965.
2This program assigns proper weights to the data,ranks respondents from most favorable to least favorable,and assigns a Guttman scale score to each respondent.
75
more) items met with failure. Further experimcuts using
subsets of items known to contain scales and reducing the
stringency of the Multiple Scalogram Analysis criteria
gave evidence of the nature of the difficulty. The MSA
program used as a criterion for retaining items the phi
coefficient (of minimum .80) from 2X2 tables of scale items.
However, the nature of a proper scale is to have one cell of
the four virtually empty, and the phi coefficient for such
a table is much reduced below, say, a tetrachoric r under
such a condition. If the items had met the usual criteria
for scalability, they would have been discarded by the
program.
For example, the cross tabulation between INFSEL and
ACQLEG produces a table that shows perfect scalibility
between the two items: All 19 persons high in their assess
ment of political influence also are acquainted with
legislators. Yet, this table produced a phi coefficient of
only .43. It was no wonder that the MSA program generated
so many one-item scales. 3
Scales were formed for the following sets of items,
and these scales are fully presented in Appendix III:
Ego strengthInnovativenessPolitical efficacy
3For a discussion of this quality of the phi coefficient,see J. P. Guilford, Fundamental Statistics in Psychology andEducation, New York, 1965, pp. 335-336.
76
EqualitySociabilityHumanitarianismRadicalismFlexibilityFaith-in-peoplePolitical cynicism
Eight items from among the authoritarianism set scaled
with reproducibility of 0.89, but this scale was dropped
because it was not sufficiently discriminatory. Most
respondents were nonauthoritarian, and none of the items
were split near the 50-50 division and below. It was re
tained, with apolitical tendency, as an index. If these
indexes fail to discriminate adequately, the error will be
in the lack of correlation between them and the other
variables.
The self-society items produced a scale with a high
reproducibility, but on two of the five items in this
scale, the response was nearly unanimous. There was near
unanimous acquiescence for the following questions:
I~ person does not need to worry about otherpeople if only he looks after himself."(Reflected)
"Even if some minority groups are treatedbadly, it is no business of mine."(Reflected)
Respondents were drawn from too narrow a spectrum of
society to provide contrast on the above items. Minorities
are too popular a cause among the politically aware.
People who are as politically active as the respondents in
this study are also quite aware of the existence of
77
minorities who have not achieved the full practical content
of their constitutional rights, and they are also aware of
what their attitude toward these minorities ought to be.
The division between the moderate left and the moderate
right on such attitudes as self-society can only be found
in such behavior as issue activity, where the left puts its
attitudes into active practice more readily than the right.
Judging from studies on political attitudes, these ten
scales tap a large portion of politically relevant attitudes.
The Cornell scores of persons on each scale provided a
criterion for a cutting point dividing the respondents into
roughly equal halves of high and low for each scale. The
undichomized Cornell scores were also used in portions of
the following analysis, particularly in the regression
predictions, and this will be noted where appropriate.
5.2 EVIDENCE ON ROLE-TAKING FROM CROSS-TABULATION.
Cross-tabulation of the dichotomized variables pro-
vides some evidence regarding the associations between them.
There is much to be learned from the simple 2X2 table that
is not always evident from other more sophisticated data
analysis techniques. 4 In this chapter single relationships
4For example, a relationship in which both the issueactive and those not issue-active believe the group is veryinfluencial will not be statistically significant, but itmay be suggestive for other reasons. In a case in which atrend is shown among the high in assumed similarity thatdoes not exist among low scorers, the table may not besignificant, but it may be important in a search forpossible controls.
78
among the variables will be discussed, while the patterns
that emerge through factor analysis will be covered in the
next chapter. The two chapters nevertheless overlap, and
an attempt will be made to avoid redundance in the two
discussions. Cross-tabulation was performed between the
three "dependent" and four role variables and all the other
variables and attributes of the study. In addition, the
interested reader may refer to the information in Figure
5.2.
Controls have been experimented with but the results
with respect to the role-taking variables were not en
couraging. S Moreover, some of the background variables
were constant in any case, because the respondents were
taken from among the more politically active. The choice
of voluntary political organizations from which to tap
respondents assured that income, education, and professional
status were almost uniformly high.
The dependent variables of the study will be dealt
with first, but briefly, since many results of the cross
tabulations may be reviewed in the next chapter on factor
SRole-taking was cross-tabulated with self-assessedinfluence and success in alterring a law, with controlsfor holding office, participation, and issue-activity.Advocacy of change was cross-tabulated with self-assessedinfluence, projection, and assumed similarity, controllingfor issue activity, projection, and ego strength. None ofthe results were significant or indicative of trends.
79
analysis. These included a measure of objective success,
for which having held an office was chosen; a measure of
subjective success assessment, for which self-assessed in
fluence was chosen; and an indicator of advocacy of change,
for which the ladder item provided a value. Ideally, out
siders' appraisals of a person's political success would
probably provide the soundest criterion of actual political
success, but such a measure was not possible in this study,
for the respondents were assured anonYmity. However, those
who have held a political office of some kind have pre
sumably been assessed by their fellows. It was possible,
of course, that a person's personal appraisal of his own
political success would also be an appropriate indicator of
real success, when correlated with the objective criterion.
Indeed, having held office (HELOFF) appears to be part
of a political success pattern. Those who have held some
office are acquainted with legislators (ACQLEG), consider
political office-seeking a distinct possibility for them
selves (POSOFF) , are issue active (ISSACT), are high in
political knowledge (POLKNO), regard themselves as more
influential (INFSEL), and are more interested in politics
(POLINT) than those who have held no office. These asso
ciations are shown in the accompanying tables.
Having held office demarcates politically active
individuals who are most willing to engage the system in
seeking its offices and working to influence issues.
ACQLEG
80
TABLE 5.1
CORRELATES OF HELD OFFICE
HELOFF/ACQLEGa
HELOFF
Yes
No
No
25
22
Yes
22
2
Total
47
24
Total
h. b
C l.-square =
47 24
10.51, p « .01, Phi =
71
.38
aAll definitions and full names of these attributes andbvariables may be found in Appendix III.
These chi-squares may not be used to infer to thepopulations from which the respondents came. Only hypotheses which relate specifically to the 71 respondentsare being tested.
TABLE 5.2
CORRELATES OF HELD OFFICE
HELOFF/POSOFF
POSOFF HELOFF
No Yes Total
Yes 12 16 28
No 35 8 43
Total 47 24 71
Chi-square = 11. 26, p « .001, Phi = .40
TABLE 5.3
CORRELATES OF HELD OFFICE
HELOFF/ISSACT
ISSACT HELOFF
No Yes Total
Yes 15 17 32
No 32 7 39
Total 47 24 71
Chi-square = 9.72, p <. .01, Phi = .37
TABLE 5.4
CORRELATES OF HELD OFFICE
HELOFF / SOC IAL
SOCIAL HELOFF
No Yes Total
High 19 14 33
Low 28 10 38
Total 47 24 71
Chi-square = 2.05, NS, Phi = .17
81
TABLE 5.5
CORRELATES OF HELD OFFICE
HELOFF/PARTIC
PARTIC HELOFF
No Yes Total
High 20 15 35
Low 27 9 36-Total 47 24 71
Chi-square = 2.53, NS, Phi = .19
TABLE 5.6
CORRELATES OF HELD OFFICE
HELOFF/INFSEL
INFSEL HELOFF
No Yes Total
High 6 13 19
Low 41 11 52
Total 47 24 71
Chi-square = 13.89, p < .001, Phi = .44
82
TABLE 5.7
CORRELATES OF HELD OFFICE
HELOFF/POLINT
POLINT HELOFF
No Yes Total
High 7 12 19
Low 40 12 52
Total 47 24 71
Chi-square = 9.99, p < .01, Phi = .38
TABLE 5.8
CORRELATES OF HELD OFFICE
HELOFF/POLKNO
POLKNO HELOFF
No Yes Total
High 21 17 38
Low 26 7 33
Total 47 24 71
Chi-square = 4.37, p < .05, Phi = ,25
83
TABLE 5.9
CORRELATES OF HELD OFFICE
HELOFF/BSCORE
BSCORE HELOFF
No Yes Total
High 22 16 38
Low 25 8 33
Total 47 24 71
Chi-square = 2.52, NS, Phi = .19
TABLE 5.10
CORRELATES OF HELD OFFICE
HELOFF/PROJEC
PROJEC HELOFF
No Yes Total
High 27 8 35
Low 20 16 36
Total 47 24 71
Chi-square = 3.70, p < .10, Phi = -.23
84
85
TABLE 5.11
CORRELATES OF HELD OFFICE
HELOFF / AS PGAP
ASPGAP HELOFF
No Yes Total
High 19 3 22
Low 28 21 49
Total 47 24 71
Chi-square = 5.79, p <: .02, Phi = -.29
Explanations for a few possible anomalies do not dispel
this interpretation. For example, the results indicate that
the persons who have held office also tend to feel polit
icalYy' efficacious (POLEFF), which is in keeping with the
greater tendency for them to be among those high in pol it-
ical influence. However, this table was not statistically
significant because the persons who have held no office also
tend to feel politically efficacious. For the same reason,
the aspirations (ASPIRE) of office holders are not distin
guished from the aspirations of other politically-aware
individuals .....
It is interesting that the cross-tabulation with self-
success in alterring a law (SUCSEL) shows that the office
holders fall in the same direction as the non-office
.~.
86
holders; that is, they do not assume that they will be
successful in engaging the political system as individuals.
Experience has taught these persons that the individual is
not often effective by himself, and the question seems to
have been taken in this sense. Given teamwork and organi
zation, however, the persons with experience in office are
able to exercise what they consider to be their political
influence (Table 5.6). And, they, like the non-office
holders, assume that the group to which they belong will
have more success than not in influencing a political
decision.
The office holders are not more accurate role-takers
than others, although their behavior score (BSCORE)
approaches significance (p <. .10), but they have less
tendency toward disowning projection (PROJEC--Tab1e 5.10).
Having held office themselves, they error less than others
by assuming a difference between themselves and fellow
politicos when predicting the attitudes of legislators. Of
some interest is the tendency of office holders to rate
higher in sociability (SOCIAL). These are persons who put
themselves forward in social situations, as can be seen
from the nature of the questions in this scale.
A second dependent variable attempted to tap political
success; the first choice for this was Question 2.15, the
87
perception of success of self in alterring a law. 6 In one
sense, it might be anticipated that this variable would
correlate highly with the objective measure, but this did
not prove to be the case. Those who have held office showed
no more tendency than others to attribute success to self.
There has already been a hint that the reason for this may
lie in the interpretation of the question as indicating
what the individual is capable of doing in politics without
organization and the aid of others. It therefore tapped an
entirely different context than did the attribute of having
held office.
For example, the results indicate that self-success
distinguishes among individuals who regard seeking political
office as a possibility. Such individuals are nevertheless
no more active in issues than others. These persons are
also high in political efficacy, which is no surprise, since
this attitude is a similar self-assessment of success.
These individuals also fall into the low category of polit
ical cynicism--they are sure of their powers. They tend to
attribute influence (INFGRP) and success (SUCGRP) to the
groups to which they belong as well. Finally, they are
high in actual similarity to the attitudes of the legis-
lators on consensual items. That is, their answers on
6What follows is based on cross-tabulations on 57cases. A later calculation on the entire 71 cases did notalter the conclusions.
88
these items for self tended to be the same as the legis
lators' own answers, but this did not make them better
role-takers. 7
An anomaly somewhat difficult to explain is why among
persons who regard themselves as potentially successful in
alterring a law is there no significant difference between
them and others on influence, self-assessed? The explana
tion may lie in the nature of the success question, which
does not ask a person to assess what success he actually
has enjoyed, but the success he might have if he were to
try to change a law. While success of this kind is poten
tial, influence is real, and, although these persons feel
they would be successful if they tried, they are honest in
assessing their real influence. That this interpretation is
probable is borne out by the results of political participa
tion (PARTIe) and issue activity, which show that these
persons do not engage in political activity any more than
do others who feel less potentially successful. Indeed, it
is among these individuals that we distinguish those who
feel that the group to which they belong is themost in
fluencial--this is probably perceived as real. And they are
almost unanimous in attributing success to the group.8
7No tables have been included for this variable, butFigure 5.2 supports these statements.
8This was. a result of the 57-case analysis.
89
The cross-tabulations with self-success in alterring
a law have indicated a more accurate subjective assessment
of success to be self-assessed influence. An exploration
of cross-tabulations with self-assessed influence substan
tiates this.
Individuals who assess themselves as high in political
influence are those who have held office, are acquainted
with legislators, consider seeking a political office a
possibility, are high in ego strength (EGOSTG), political
efficacy, political interest, and participation. The en
tire context of "influence" appears to be more tangible
than the "success" tapped by SUCSEL.
The high influentials also tend to be male, one of the
few significant correlations with the sex attribute. More
over, they also tend to be the eldest or only children of
families. Although it is not the purpose of this research
to follow this subsidiary line of inquiry, it is possible
that eldest or only children are placed early in situations
calling for responsibility.
Self-assessed influence does not correlate significant
ly with the feeling of possible success nor with political
knowledge. It is not highly correlated with the ladder
value of advocacy of change, and none of the role variables
appear to be associated with it directly. Because this
variable is more tangibly related with other accoutrements
of success, it will be regarded as the best subjective
indicator of that feeling among the measures used here.
TABLE 5.12
CORRELATES OF INFLUENCE OF SELF
INFSEL/ACQLEG
ACQLEG INFSEL
Low High Total
Yes 28 19 47
No 24 24
Total 52 19 71
Chi-square = 13.25, p<::' .001, Phi = .43
TABLE 5.13
CORRELATES OF INFLUENCE OF SELF
INFSEL/POSOFF
POSOFF INFSEL
Low High Total
Yes 15 13 28
No 37 6 43
Total 52 19 71
Chi-square = 9.12, p < . 01, Phi = .36
90
TABLE 5.14
CORRELATES OF INFLUENCE OF SELF
INFSEL/ISSACT
ISSACT INFSEL
Low High Total
Yes 17 15 32
No 35 4 39
Total 52 19 71
Chi-square = 12.03, p < .001, Phi = .41
TABLE 5.15
CORRELATES OF INFLUENCE OF SELF
INFSEL/SEX
SEX INFSEL
Low High Total
Men 30 17 47
Women 21 2 23
Total 51 19 70
Chi-square = 5.89, p < .02, Phi = .21
91
TABLE 5.16
CORRELATES OF INFLUENCE OF SELF
INFSEL/ELDONL
ELDONL INFSEL
Low High Total
Yes 29 17 46
No 22 2 24
Total 51 19 70
Chi-square = 6.53, p < .05, Phi = .23
TABLE 5.17
CORRELATES OF INFLUENCE OF SELF
INFSEL/EGOSTG
EGOSTG INFSEL
Low High Total
High 18 15 33
Low 34 4 38
Total 52 19 71
Chi-square = 10.99, p < .001, Phi = .39
92 ..
TABLE 5.18
CORRELATES OF INFLUENCE OF SELF
INFSEL/POLCYN
POLCYN INFSEL
Low High Total
High 24 9 33
Low 28 10 38
Total 52 19 71
Chi-square = .008, NS, Phi = .01
TABLE 5.19
CORRELATES OF INFLUENCE OF SELF
INFSEL/POLEFF
POLEFF INFSEL
Low High Total
High 20 15 35
Low 32 4 36
Total 52 19 71
Chi-square = 9.13, p < .02, Phi = .36
93
TABLE 5.20
CORRELATES OF INFLUENCE OF SELF
INFSEL/PARTIC
PARTIC INFSEL
Low High Total
High 21 14 35
Low 31 5 36
Total 52 19 71
Chi-square = 6.17, p < .02, Phi = .29
TABLE 5.21
CORRELATES OF INFLUENCE OF SELF
INFSEL/POLINT
POLINT INFSEL
Low High Total
High 11 8 19
Low 41 11 52
Total 52 19 71
Chi-square = 3.12, p < .10, Phi = .21
94
95
The third dependent variable is advocacy of change
(CHANGE).9 The advocates of greatest change are, quite
naturally, those who appear most issue active. The de
sire for change appears as a major motivation for political
activity of this kind among these respondents; apparently,
there are other motives for other kinds of political
act~'ity, such as seeking office, with which attribute the
change-oriented correlate negatively. The change-oriented
are also among the high in political participation and tend
to be higher (but not significantly) in political interest,
although they do not rank higher in political knowledge.
TABLE 5.22
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/POSOFF
POSOFF
Yes
No
Total
Low
18
19
37
CHANGE
High
10
24
34
Total
28
43
71
Chi-square = 2.75, p < .10, Phi = -.20
9CHANGE means the ladder-value. The verbal criterion,Question 2.14 had a product-moment correlation with thisvalue .74 and was dropped from the analysis.
TABLE 5.23
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/ISSACT
ISSACT CHANGE
Low High Total
Yes 13 19 32
No 24 IS 39
Total 37 34 71
Chi-square = 3.08, p < .10, Phi = .21
TABLE 5.24
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/EMPJOB
EMPJOB CHANGE
Low High Total
Yes 16 6 22
No 20 28 48
Total 36 34 70
Chi-square = 5.83, p < .05, Phi = -.30
96
TABLE 5.25
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/RADICAL
RADICAL CHANGE
Low High Total
Left 13 23 36
Right 24 11 35
Total 37 34 71
Chi-square = 7.49, p <: .01, Phi = .32
TABLE 5.26
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/EQUAL
EQUAL CHANGE
Low High Total
High 20 26 46
Low 17 8 25
Total 37 34 71
Chi-square = 3.90, p< .05, Phi = .23
97
TABLE 5.27
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/INNOUT
INNOUT CHANGE
Low High Total
High 14 25 39
Low 23 9 32
Total 37 34 71
Chi-square = 9.12, p < .01, Phi = .36
TABLE 5.28
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/HUMANE
HUMANE CHANGE
LoW High Total
High 21 27 48
Low 16 7 23
Total 37 34 71
Chi-square = 4.15, p < .05, Phi = .24
98
TABLE 5.29
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/POLEFF
POLEFF CHANGE
Low High Total
High 14 21 35
Low 23 13 36
Total 37 34 71
Chi-square = 4.06, p <. .05, Phi = .24
TABLE 5.30
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/PARTIC
PARTIC CHANGE
Low High Total
High 14 21 35
Low 23 13 36
Total 37 34 71
Chi-square = 4.06, p < .05, Phi = .24
99
TABLE 5.31
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/INFGRP
INFGRP CHANGE
Low High Total
High 7 15 22
Low 30 19 49
Total 37 34 71
Chi-square = 5.26, p « .05, Phi = .27
TABLE 5.32
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/FINDEX
FINDEX CHANGE
Low High Total
High 22 11 33
Low 15 23 38
Total 37 34 71
Chi-square = 5.23, p .( .05, Phi = -.27
100
101
TABLE 5.33
CORRELATES OF ADVOCACY OF CHANGE
CHANGE/APOLIT
APOLIT
Low
CHANGE
High Total
High 19
Low 18
Total 37
Chi-square = 1.95, NS, Phi =
23
11
34
.17
42
29
71
The change-oriented are engaging the system rather
than ignoring the channels open for changing it. This
does not prevent the advocates of change, however, from
being high on political cynicism (POLCYN). It is in
teresting to see an association that allies political
efficacy and political cynicism in the same context.
The advocates of change tend to be high in attitudes
of equality (EQUAL), humanitarianism (HUMANE), radicalism
(RADICL), and innovativeness (INNOVT). The last two at
titudes are obvious relationships and help to explain the
tendencies on the first two. Along the political spectrum
from which these respondents corne, it is the left that
seeks the greatest change, a finding that is shown by the
analysis of variance between the groups. The list of
issues engaged in by these change-oriented individuals is
102
strong in civil rights and peace orientations, areas tapped
by the scales EQUAL and HUMANE. Implicated in this group
of relationships is the negative correlation with the
change-oriented and the authoritarian index (FINDEX).
This may be regarded as a sort of moderate left-wing
pattern, given the present political environment in the
United States. This pattern was verified in the factor
analysis presented in the next chapter.
In passing, it is interesting to note that flexibility
(FLEXBL) does not distinguish the change advocates from
others. It appears to be tapping variance distinct from
innovativeness and radicalism, which might be assumed to
be more closely related to it. Also, the change-oriented,
although they have high political aspirations are not dis
tinguished on this variable from persons not so change
oriented. Once more, political aspiration is associated
with other aims than changing one's environment.
The role-taking variables do not indicate distinctions
between advocates of great change and others in these cross
tabulations. Moreover, the change-oriented are not more
prone to disowning projection than others. It cannot be
assumed that a strong motivation for change in the respon
dents tapped in this study has a direct effect in causing
them either to project their own self-images into the role
of political-other or to project negative images into them.
It is quite likely that these persons, formidable socially
103
and politically, are not frustrated enough in their hopes
for accomplishing their aims to be high on PROJEC, in
spite of the tendency toward political cynicism. In
general, the perception of reality is not impaired by the
fairly high intensity of desire for change among the change
oriented as a whole.
In dealing with the role variables, it is necessary to
treat first of their interrelationships (see Tables 5.41,
5.42, 5.46, 5.47, 5.48, and 5.53-55). Not surprisingly,
the behavior score is found to be strongly associated with
role accuracy (ROLEAC). The latter is strongly associated
with assumed similarity (ASSUME) and less strongly with
actual similarity (ACTUAL). The linearity between these
variables, although reduced as a result of the procedure
of dichotomization, still shows up strongly.
Projection, of course, as an artefact of scoring and
of the nature of the respondent group (as shown in Chapter
IV), shows a strong negative relationship to role accuracy
and assumed similarity. However, the relationships of the
behavior score with projection indicate that these relation
ships are not entirely artefactua1 (see Table 5.40). Not
only were projectors inaccurate role-takers, which might be
explained on the basis of the scoring procedure, but they
also were strongly related negatively to the behavior score,
which is positively related to assumed similarity (p < .02).
Thus, the interrelationships among the role variables
104
received an internal validation through the behavior score.
Assumed similarity, the variable probably most closely
associated with empathy, had the richest associations be
yond the role variables. Acquaintance with a legislator
approached a significance of .10. Faith-in-people (FAITHP)
was strongly associated with assumed similarity, while, as
the tables show, both radicalism and equality related strong
ly in the negative direction. Here is an indication that
those associated with a desire for change, as radicals, do
not assume their own attitudes match the legislators. Per
sons high on assumed similarity are also high in sociability
and tend to be somewhat higher in political efficacy. It
is also possible that if one assumes similarity with a
holder of moderate power such as a legislator, he will also
tend to regard himself as so much less a cornmon man, hence
the tendency for the negative relationship with equality.
This tendency was also evident in the pilot study, although
among those student respondents it did not reach statistical
significance (taken as p < .10).
Oddly, role accuracy is not enhanced by having held
political office; in fact, assumed similarity is apparently
not associated, as might be thought, with having held
office. The possibility of running for public office, how
ever, not only is associated with assumed similarity but
with role accuracy as well. This, even more than acquaint
ance with a legislator, may elicit from an aspirant to
TABLE 5.34
CORRELATES OF ASSUMED SIMILARITY
ASSUME/ACQLEG
ACQLEG ASSUME
Low High Total
Yes 18 29 47
No 14 10 24
Total 32 39 71
Chi-square = 2.58, N. S . , Phi = .19
TABLE 5.35
CORRELATES OF ASSUMED SIMILARITY
ASSUME/ISSACT
ISSACT ASSUME
Low High Total
Yes 11 21 32
No 21 18 39
Total 32 39 71
Chi-square = 2.69, Almost p .(..., .10, Phi = .19
105
TABLE 5.36
CORRELATES OF ASSUMED SIMILARITY
ASSUME/RADICL
RADICL ASSUME
Low High Total
High 20 16 36
Low 12 23 35
Total 32 39 71
Chi-square = 3.24, p <- .10, Phi = -.21
TABLE 5.37
CORRELATES OF ASSUMED SIMILARITY
ASSUME/FAITHP
FAITHP ASSUME
Low High Total
High 9 28 37
Low 23 11 34
Total 32 39 71
Chi-square = 13.43, p< .001, Phi = .44
106
TABLE 5.38
CORRELATES OF ASSUMED SIMILARITY
ASSUME/EQUAL
EQUAL ASSUME
Low High Total
High 28 18 46
Low 4 21 25
Total 32 39 71
Chi-square = 13.17 , p < .001, Phi = -.43
TABLE 5.39
CORRELATES OF ASSUMED SIMILARITY
ASSUME/SOCIAL
SOCIAL ASSUME
Low High Total
High 10 23 33
Low 22 16 38
Total 32 39 71
Chi-square = 5.43, p < .02, Phi = .28
107
TABLE 5.40
CORRELATES OF ASSUMED SIMILARITY
ASSUME/POLEFF
POLEFF ASSUME
Low High Total
High 12 23 35
Low 20 16 36
Total 32 39 71
Chi-square = 3.24, p< .10, Phi = .21
TABLE 5.41
CORRELATES OF ASSUMED SIMILARITY
ASSUME/ACTUAL
ACTUAL ASSUME
Low High Total
High 12 30 42
Low 20 9 29
Total 32 39 71
Chi-square = 11.31, p < .001, Phi = .40
108
109
TABLE 5.42
CORRELATES OF ASSUMED SIMILARITY
ASSUME/PROJEC
PROJEC ASSUME
Low High Total
High 26 9 35
Low 6 30 36
Total 32 39 71
Chi-square = 23.80, p <: .001, Phi = -.58
public office more consideration as to just how well he
might fit into the role. A socializing influence may also
be at work here to draw into active office-seeking those
among the politically aware whose attitudes are already
most like encumbents. Issue activity also may stimulate
study of legislators attitudes, for the legislators must
often be influenced in the course of pursuing goals. The
implication is that rather than preceding political ex
perience as a general skill, role accuracy appears here to
be the outgrowth of certain motivations that cause persons
to become more accurate role-takers in the political realm.
TABLE 5.43
CORRElATES OF ROLE-TAKING ACCURACY
ROLEAC/POSOFF
POSOFF ROLEAC
Low High Total
Yes 9 19 28
No 23 20 43
Total 32 39 71
Chi-square = 3.12, p ~ .10, Phi = .21
TABLE 5.44
CORRELATES OF ROLE-TAKING ACCURACY
ROLEAC/ISSACT
ISSACT ROLEAC
Low High Total
Yes 10 22 32
No 22 17 39
Total 32 39 71
Chi-square = 4.49, p < .05, Phi = .25
110
TABLE 5.45
CORRELAT£S OF ROLE-TAKING ACCURACY
ROLEAC/POLEFF
POLEFF ROLEAC
Low High Total
High 12 23 35
Low 20 16 36
Total 32 39 71
Chi-square = 3.24, p < .10, Phi = .21
TABLE 5.46
CORRELATES OF ROLE-TAKING ACCURACY
ROLEAC/ASSUME
ASSUME ROLEAC
Low High Total
High 8 31 39
Low 24 8 32
Total 32 39 71
Chi-square = 21.08, p <: .001, Phi = .54
111
TABLE 5.47
CORRElATES OF ROLE-TAKING ACCURACY
ROLEAC/ACTUAL
ACTUAL ROLEAC
Low High Total
High 15 27 42
Low 17 12 29
Total 32 39 71
Chi-square = 3.64, p < .10, Phi = .23
TABLE 5.48
CORRELATES OF ROLE-TAKING ACCURACY
ROLEAC/PROJEC
PROJEC ROLEAC
Low High Total
High 23 12 35
Low 9 27 36
Total 32 39 71
Chi-square = 11.88, p< .001, Phi = -.41
112
113
TABLE 5.49
CORRELATES OF ROLE-TAKING ACCURACY
ROLEAC/FINDEX
FINDEX ROLEAC
Low High Total
High 18 15 33
Low 14 24 38
Total 32 39 71
Chi-square = 2.24, NS, Phi = -.18
These are the only non-role associations with role
accuracy. There is apparently no direct relationship be
tween it and many of the other attitude factors presumed
to be so related. Thus, equality, faith-in-peop1e,
sociability, and humanitarianism show no relationship,
while the authoritarianism index indicates a slight negative
but not a significant trend.
The behavior score, like role accuracy is related to
issue activity. It is negatively related to flexibility,
perhaps because these respondents were too flexible in de
parting from what the less flexible found to be more
obvious answers to the situation questions. Considering the
nature of the items in the sociability scale, it is hearten
ing that its relationship to the behavior score at least
114
approaches significance (p < .10). But, with none of the
other variables, oddly, is the behavior score related.
TABLE 5.50
CORRELATES OF THE BEHAVIOR SCORE
BSCORE/ISSACT
ISSACT BSCORE
Low High Total
Yes 11 21 32
No 22 17 39
Total 33 38 71
Chi-square = 3.43, p < .10, Phi = .22
TABLE 5.51
CORRELATES OF THE BEHAVIOR SCORE
BSCORE/FLEXBL
FLEXBL BSCORE
Low High Total
High 20 15 35
Low 13 23 36
Total 33 38 71
Chi-square = 3.16, p < .10, Phi = -.21
TABLE 5.52
CORRELATES OF THE BEHAVIOR SCORE
BSCORE/SOCIAL
SOCIAL BSCORE
Low High Total
High 12 21 33
Low 21 17 38
Total 33 38 71
Chi-square = 2.54, NS, Phi = .19
TABLE 5.53
CORRELATES OF THE BEHAVIOR SCORE
BSCORE/ROLEAC
ROLEAC BSCORE
Low High Total
High 13 26 39
Low 20 12 32
Total 33 38 71
Chi-square = 6.01, p <: .02, Phi = .29
115
TABLE 5.54
CORRELATES OF THE BEHAVIOR SCORE
BSCORE/ASSUME
ASSUME BSCORE
Low High Total
High 13 26 39
Low 20 12 32
Total 33 38 71
Chi-square = 6.01, p < .02, Phi = .29
TABLE 5.55
CORRELATES OF THE BEHAVIOR SCORE
BSCORE/PROJEC
PROJEC BSCORE
Low High Total
High 24 11 35
Low 9 27 36
Total 33 38 71
Chi-square = 13.54, p< .001, Phi =: -.44
116
117
The projection score enjoys but one non-role relation-
ship, a negative correlation with political interest. No
relationship is apparent here with advocacy of change.
Interestingly, the authoritarianism index, although an un-
certain indicator, is not related to projection as might
be anticipated. If authoritarians do engage in a measure
of disowning projection, as some sources have shown, this
projection and role inaccuracy is not apparent with the
mild authoritarians among these respondents, who, if they
tend to disowning projection at all, do not seem to choose
legislators as targets. lO
TABLE 5.56
CORRELATES OF DISOWNING-ASSIMILATIVE PROJECTION
PROJEC/ISSACT
ISSACT PROJEC
Low High Total
Yes 21 11 32
No 15 24 39
Total 36 35 71
Chi-square = 5.19, p <: .05, Phi = -.27
10A1vin Scode1 and Paul Mussen, "Social Perceptions ofAuthoritarians and Non-Authoritarians," The Journal ofAbnormal and Social Psychology, ~ (1953), 181-184, haveshown authoritarians to project onto other individuals incertain situations. The situation in the research_..on.political role-taking may not be comparable.
TABLE 5.57
CORRELATES OF DISOWNING-ASSIMILATIVE PROJECTION
PROJEC/RADICL
RADICL PROJEC
Low High Total
High 14 22 36
Low 22 13 35
Total 36 35 71
Chi-square = 4.08, p < .05, Phi = .24
TABLE 5.58
CORRELATES OF DISOWNING-ASSIMILATIVE PROJECTION
PROJEC / SDC IAL
SOCIAL PROJEC
Low High Total
High 22 11 33
Low 14 24 38
Total 36 35 71
Chi-square = 6.29, p < .02, Phi = -.30
118
TABLE 5.59
CORRELATES OF DISOWNING-ASSIMILATIVE PROJECTION
PROJEC/POLINT
POLINT PROJEC
Low High Total
High 13 6 19
Low 23 29 52
Total 36 35 71
Chi-square = 3.26, p <: .10, Phi = -.21
TABLE 5.60
CORRELATES OF DIWOWNING-ASSIMILATIVE PROJECTION
PROJEC/FINDEX
FINDEX PROJEC
Low High Total
High 17 16 33
Low 19 19 38
Total 36 35 71
Chi-square = 0.02, NS, Phi = -.02
119
120
It remains to say a few words on four variables that
might have been possible controls for the role-taking and
related dependent variables, sex (SEX), age (AGE), empathic
professions (EMPJOB), and eldest or only child (ELDONL).
Sex appears in the political environment of the United
St t t b . d' .. . 11 d' h d .a es 0 e qU1te un 1scr1m1nat1ng, an 1n t e oma1n
of this study it turned out to be related to almost nothing
else, although there was a tendency, not statistically
significant, for males to be high in disowning projection
compared to females; this tendency was present (p .10)
among the student respondents of the pilot study. Although
females appeared to be as active, interested, and aware in
politics among these respondents, they did not feel as in
fluencial, as we have seen. There is still a psychological
gap apparent in their complete political emancipation. This
is supported by the factor analysis in the next chapter.
Age did not become a variable of interest in this
study. There was some reason to believe that with age,
one's ability to assess the attitudes of others might
improve, but it is possible that these changes in role
taking accuracy occur mainly among the very young.
11Although women are generally found to be lessknowledgeable on all class levels than men on foreignaffairs, as in Alfred O. Hero, Americans in World Affairs,Boston, The World Peace Foundations, 1955, pp. 83-85.
121
The division of professions into empathic and non-
empathic turned up no relationships to the role-taking
variables. This was not altogether anticipated, but it is
quite in keeping with innumerable results from social
psychological studies that consistently fail to reveal dif
ferences between those whose profession calls upon them to
make attitude and personality assessments and those who be
long to less empathic professions. 12 There was a relation
ship between change and empathic profession because of the
tendency for the change-oriented to be in non-empathic
professions as they were defined here. It is also possible
that the operationalization of this attribute was faulty.
The attribute ELDONL showed only one relationship of
interest. The tendency for the eldest or only children is
to fall into the high self-assessed influence category.
Again, while cognizant of the need for replication when a
single significant relationship among many appears,13 it
would be interesting to speculate on the possible reasons
l2Henry Clay Smith, sensitivitt to People, New York,1966, p. 102, and Robert t. Katz,mpathy: Its Nature andUses, Glencoe, Ill., 1963, p. 172.
l3In cases in which there are a great number ofcorrelated variables, as in these cross-tabulations, andfew are significant, as was the case with a number ofvariables, it is unsafe to assume that a single significantresult merits attention. Its significance is also a functionof the number of correlations. In this case, however, thereis some evidence of the result from the pilot study as well.For a discussion on this sometimes neglected aspect ofcross-tabulation and correlation studies, see Hanan C. Selvin,
122
for this. It is possible that eldest and only children
enjoy in a middle-class family a real position of influence
and responsibility not accorded to the younger members.
They may be prone to seek out such positions in maturity
as a result.
5.3 HINTS TOWARD THE FACTOR ANALYSIS.
The cross-tabulation study does not allow an opti
mistic appraisal of the importance of the role-taking
variables in a political study. It is possible, however,
that they may appear on dimensions with other political
variables, even though the cross-tabulation relationships
with other variables may not be very high.
150
Frequencies
o80
Phi correlation
FIGURE 5.1 CURVE REPRESENTING FREQUENCIES OFPHI-COEFFICIENT SIZES
"A Critique of Tests of Significance in Survey Research,"American Sociological Review, 22 (1957), p. 526.
123
A prediction of the results in the next chapter is
given by Figure 5.1, which shows the frequencies of size
among the correlations of all pairs of variables in the
study.14 Although the phi-coefficients, which were used
for the cross-tabulations and for this figure, are gen
erally low, it is the shape of the curve that indicates
the possible result of a factor analysis. The high
frequency of low correlations indicates a group of vari
ables that belong to many different dimensions. It is
unlikely that the factor analysis will turn up many
interesting relationships not found in the cross-tabula-
tions for the role-taking variables.
Figure 5.2 is an illustration of the interrelatedness
of the variables. The role variables have been depicted
by heavier lines. Here too, it can be seen that the
strongest relationships do not occur between the role
variables and the political and attitude variables.
l4See Johan Galtung, Theor~ and Methods of SocialResearch, New York, 1967, pp. 03-30S.
5
1/
/2. 13 9
8
I
3'1
33
124
Key to numbers (for complete description of variables seeAppendix III):1 HELOFF2 ACQLEG3 POSOFF4 ISSACT5 SEX6 RESIDE7 ELDONL8 IMPJOB9 EGOSTG
FIGURE 5.2
10 RADICL 19 PARTIC 28 BSCORE11 FLEXBL 20 INFSEL 29 ROLEAC12 FAITHP 21 POLINT 30 ASSUME13 EQUAL 22 ASPIRE 31 ACTUAL14 SOCIAL 23 INFGRP 32 PROJEC15 INNOVT 24 CHANGE 33 AGE16 HUMANE 25 SUCSEL 34 ASPGAP17 POLCYN 26 SUCGRP 35 FINDEX18 POLEFF 27 POLKNO 36 APOLIT
RELATIONSHIPS EXCEEDING A CHI-SQUARE OF 3.84AND A PHI OF .2325 AMONG THE VARIABLES ANDATTRIBUTES USED IN THE STUDY OF THE POLITICALROLE SPACE (Role-taking variables areindicated by heavier lines)
CHAPTER VI
POLITICAL DIMENSIONS OF ROLE-TAKING
6.1 THE ORTHOGONAL FACTOR ANALYSIS.
A phi-correlation matrix was entered into a factor
analysis program to determine the orthogonal dimensions of
common variance among the 36 variables discussed in the
last chapter. 1 The highest row correlations furnished the
communalities for the matrix. Factoring was accomplished
with the Mesa I program on a 7040 computer. 2 Factors above
a minimum eigenvalue of 0.95 only were extracted and ro
tated to an orthogonal varimax solution. This gave seven
factors descriptive of the common variance among the 36
attributes and variables.
~uch criticism has been directed against the phi coefficient to provide correlations as input into a factoranalysis. Ultimately, no correlation coefficient is ideal.The case against the use of phi and phi-over-phi-max isably presented by John B. Carroll, "The Nature of the Data,or How to Choose a Correlation Coefficient," Psychometrika,26 (1961), 347-372. Unfortunately, computer programs avai1ao1e to this researcher used a cosine-pi approximation to thetetrachoric correlation coefficient, which can only be usedwhen variables are dichotomized near their medians. Therefore, the argument presented in Andrew L. Comrey and EdwardLevonian, '~ Comparison of Three Point Coefficients inFactor Analyses of MMPI Items," Educational and PsychologicalMeasurement, XVIII (1958), 739-755, was accepted. ~s thediscussion will show, the tetrachoric matrix was comparedwith the factor results from the phi, and the latter showedthe same structure on the orthogonal results without theproblems of communalities exceeding 1.00, which is commonwith the tetrachoric.
2This program was prepared at the University ofChicago by Florence Bradford.
126
A number of variables participated but little in the
common variance, and a hint as to this result could be
obtained from the cross-tabulation results. Table 6.1
gives a list of variables that generally partook in a
single dimension only in the political role-taking space
together with their communalities. These communalities
show the amount of variance in common with the other
variables that was explained over the seven dimensions of
the common factor space.
TABLE 6.1 VARIABLES ENTERING BUT SLIGHTLYINTO THE COMMON FACTOR SPACE,
AS INDICATED BY LOW COMMUNALITIES
Variable name
SEXELDONLEMPJOBFLEXBLSOCIALPOLCYNPOLINTINFGRPCHANGESUCSELSUCGRPPOLKNOBSCOREACTUALASPGAPAPOLIT
Communality
.14
.13
.24
.32
.25
.32
.39
.31
.38
.27
.39
.28
.38
.36
.38
.35
Factor one (see Table 6.2) has been suggested in the
cross-tabulation discussion; it appears to be a pattern of
variables descriptive of a left-wing moderate political
outlook. The variable loading most highly on this factor
127
is equality, followed by the authoritarian index with
opposite sign and radicalism with the same sign. The
correlation between radicalism and equality on the phi
matrix being positive, it is apparent that leftism is
here associated with its correlated attitudes, such as
flexibility, innovativeness, and humanitarianism; and
..
sociability and authoritarianism correlate negatively.
In addition, there are small loadings by political interest
and assumed similarity. The dimension seems to describe
the outlook of left-wing moderates in the present political
situation. The causes they espouse are motivated by at
titudes of humanitarianism and equality, and the outlook
they adopt tends to create some feeling of distance from
encumbents of the present political system as indicated by
the negative loading of assumed similarity.
TABLE 6.2 FACTOR I: LEFT-RADICAL (Reflected)(29.5 per cent of common variance)
Variable name
EQUALFINDEXRADICLINNOVTFLEXBLHUMANEASSUMEPOLINT
Loading
.71-.68
.63
.60
.46
.39-.38
.32
128
A previous factor analysis of 57 cases,3 more biased
by the selection of left-wing respondents, associated change
and issue activity with the variables on this factor. In
the present analysis 14 cases have been added from right
wing respondents, and the change orientations have broken
off to form another orthogonal dimension, which will be
discussed later. Interestingly, one dimension of the seven
obtained in the analysis of the pilot study respondents
using a comparable selection of 29 variables also turned up
a pattern associating the left-wing with attitudes such as
humanitarianism, issue activity, and change-orientation.
In the pilot study, there were also moderate loadings
with role-taking accuracy and disowning projection, both in
the same direction! This may be interpreted with regard to
the student respondents in the following manner: Among
students the chief form of political activity is associated
with issues. Although the radical students tend to dis-
associate themselves to a degree from the political system
as it stands, which is evidenced by the projection variable
loading on the factor, they also are, relative to other
students, more knowledgeable about legislators. The same
conditions do not hold when the entire group of respondents
is politically aware.
3The pilot study, it will be recalled, was an exploratory analysis using 69 student respondents. The 57case analysis was done prior to the receipt of 14additional questionnaires from conservative respondents.
129
In the present selection of respondents, there are
other motives than issues that enlist the interest of
political participants. For example, these respondents
are possible candidates for public office. They are more
influencial than students in the election of other can-
didates for office, which mayor may not be associated with
issues. Thus, with motivations toward other ends than
issue activity, there does not appear to be a positive
association between role-taking accuracy and radicalism
a~ong the respondents of the final study.4 Moreover, the
lack of assumed similarity is notable among the relation-
ships produced by this group of politically-aware individ
uals on the left-radical dimension.
It might be argued, considering the results of the
57-case factor analysis, the pilot study, and the final 71-
case analysis that these factors are simply an artefact of
the groups included. It is true that selection plays an
important part in any factor study, not only selection with
regard to the variables studied but selection with regard
to the respondents as well. Nevertheless, the respondents
of the final study came from a fairly broad portion of the
political spectrum, and the result fits with qualitative
assessments of the present political climate of the United
4Indeed, the oblique factor analysis revealed just theopposite.
130
States. Left-wing moderates are associated with the array
of attitudes and attributes that appear on this factor.
Moreover, the factors emerging from the three results do
not appear to be widely divergent, in spite of differences.
All the analyses yielded seven dimensions, for example, and
some of the factors were recognizably ana1ogous. 5
Factor two (see Table 6.3) might be called an efficacy
dimension, characterized by the variable that loaded most
highly on it, self-assessed influence. The cross
tabulations indicated that assessment of influence was a
more active quality than the assessment of success in al
tering a law. Success appeared to convey a sense of
potentiality, as opposed to influence, which was inter
preted as an indication of a tangible and more active
relationship with the political world. This interpretation
seems to be borne out on this dimension, on which the
possibility of seeking office, having held office, political
efficacy, acquaintance with a legislator, issue activity,
and ego strength also load. In addition, the factor shows
a negative relation with the gap between actual se1f-
assessed influence and the desire for influence. There is
5Future replications using the same variables mightprofit from a factor comparison technique.
131
also a small loading by the sex attribute, slightly asso-
ciating males with efficacy actual and perceived. This is
the only loading above .30 by the sex attribute on the
seven dimensions.
TABLE 6.3 FACTOR II: REAL EFFICACY(21.2 per cent of common variance)
Variable name
INFSELACQLEGASPGAPHELOFFISSACTPOSOFFEGOSTGPOLEFFSEX
Loading
.77
.61-.55
.51
.43
.43
.39
.34
.34
Factor three (see Table 6.4) might be called a role-
taking factor, since all the role-taking variables, with
the exception of the indicator of passive actual similarity,
load highly on it. Their interrelationships are much
stronger than any relationships with non-role variables.
A number of the political attributes and indicators do
load together with them, however, including having held
office, issue activity, political interest, and sociability.
These loadings, although small, tend to substantiate a
number of predicted associations with the role-taking
variables. Although simply being acquainted with a legis
lator does no~ appear to make a person a more accurate role
taker, at least among these respondents, having held office
132
TABLE 6.4 FACTOR III: ROLE-TAKING(12.7 per cent of common variance)
Variable name
PROJECROLEACBSCOREASSUMEPOLINTHELOFFISSACTSOCIAL
Loading
-.69.59.59.49.37.36.36.30
and engaging in issue activity seems to motivate persons to
be more accurate in role-taking. Perhaps having held of-
fice allows a person to assume that he is more similar to
a fellow office holder such as a legislator. Some of these
associations have already been discussed in the cross
tabulation results, although there it was noted that the
possibility of seeking a public office was associated
significantly with role-taking accuracy, whereas in the
factor analysis this association was too slight for in
clusion.
The name chosen for the factor explaining the next
largest portion of the common variance is presented
tentatively: political empathy (see Table 6.5). No argu
ment is made that these names, all of them judgmental
interpretations of the associations indicated on the
1336factors, are "correct" or "best." Nevertheless, this
fourth dimension taps a certain euphoric sense of personal
potential for success. The discussion of the cross
tabulations with success in altering a law and other
variables may be recalled here. This dimension allies it
with faith-in-peop1e, political efficacy, and, negatively,
political cynicism and the index of apolitical tendencies.
Associated with this pattern of related variables is actual
and assumed similarity. Subjectively, this factor de-
scribes persons who feel personally effective in the
political realm, but they have not subjected themselves to
tangible tests of their potential powers by engaging in
issue activity or office seeking. Perhaps it is this lack
of real experience that explains the lack of role-taking
accuracy or an associated loading with the behavior score.
TABLE 6.5 FACTOR IV: POLITICAL EMPATHY(10.8 per cent of common variance)
Variable name
FAITHPACTUALAPOLITASSUMESUCSELPOLCYNPOLEFF
Loading
.61
.55-.54
.51
.43-.36
.35
6en the naming of factors, see Chapter 18, mimeographof draft, April, 1966, Rudolph J. Rummel, Applied FactorAnalysis, forthcominfi. The subject is also treated byRaYmond B. Cattell, 'The Dimensions of Culture Patterns ofFactorization of National Characters," Journal of Abnormaland Social Psychology, 44 (1949), 443-469.
134
This dimension has possibly tapped an aspect of
voluntary political organizations which draw in not only
the active, but those persons whose political activity is
largely subjective and who adopt an otherwise passive re
lationship to politics. It is somewhat interesting to find
that this political dimension is not associated with sex.
This dispels another possible common-sense assumption about
the feminine psyche.
In the 7l-case factor analysis, the change factor has
separated from the more simple relationship with other
radical variables that appeared in the 57-case study.7 In
factor five (see Table 6.6) radicalism is related not only
to variables of outlook or attitude, but to activity vari
ables as well. Thus, participation loads most highly on
this factor, and associated with it are issue activity,
radicalism, innovativeness, humanitarianism, and advocacy
of change. Unlike factor one, which was a group of
associated passive variables, this factor, which also in-
dicates a left-wing pattern, is an active factor, combining
not only attitude variables, but acti.vity indicators as
well. Is it accidental, therefore, that it is on this factor
rather than on the more passive dimension that political
cynicism loads?
7From Table 6.10, however, it can be seen that theoblique factors, left-radical and change, are correlated.22, which is a moderate relationship.
135
TABLE 6.6 FACTOR V: CHANGE(9.9 per cent of common variance)
Variable name
PARTICHUMANERADICLCHANGEPOLCYNISSACTEMPJOB
Loading
.62
.50
.49
.47
.40
.36-.31
Evidently, political cynicism is no barrier to a very
active stance with regard to attitudes such as humanitarian-
ism. In spite of it, persons can combine issue activity,
innovativeness, and concrete participation. Yet, when the
individual's desire for change runs high, he is bound to
find the political system a block to his plans for a better
society, and the apparent opportunism of the establishment,
which makes compromises that slow the aims of the change
oriented, may make him somewhat cynical about politics.
But this cynicism does not stultify involvement. It seems
to be an outgrowth of confrontation.
It is on this factor, that the hypotheses presented
in Chapter II lead us to expect a loading by disowning
projection. But disowning projection is absent from the
change factor, nor is the loading of role-taking accuracy
(-.26) high enough for extended comment. The desire for
extensive change and activities to that end have not
perceptibly diminished the ability of the change-oriented
136
as a group to correctly assess the attitudes and behavior
of legislators. At least, this is the conclusion based on
the orthogonal factors.
Factor six (see Table 6.7) is an age factor, on which
age loads most highly together with residence in Hawaii.
Considering that the division for residence was made at
ten years, this is no surprise, for many of the members of
active voluntary organizations are former mainlanders, and
those who have lived in Hawaii for at least ten years are
naturally older. With age, aspirations in politics also
appear to diminish, and there is a substantial negative
loading for political aspirations.
TABLE 6.7 FACTOR VI: AGE(8.9 per cent of common variance)
Variable name
AGERESIDEASPIRE
Loading
.72
.66-.48
The seventh factor (see Table 6.8) has been named the
group success factor. The variables associated on this
factor include the assessment of the success of the group,
the influence of the group, and, negatively, political
knowledge. There are low associations with ego strength
and political efficacy. It is possible that this factor
has tapped yet another motive for association with a
voluntary political organization. Whereas factor four
137
indicated a feeling of potential power for oneself, this
factor seems to be composed of those variables that show a
vicarious psychological participation in the powers of the
organization. The negative association with political
knowledge and the lack of any activity variables or at-
tributes indicates that for some persons the major end
involved in joining a political organization is to partake
in the influence enjoyed by the organization itself. No
role-taking variables are associated with this dimension.
TABLE 6.8 FACTOR VII: GROUP SUCCESS (Reflected)(7.0 per cent of common variance)
Variable name
SUCGRPINFGRPPOLKNOEGOSTGPOLEFF
Loading
.55
.48-.42
.35
.32
It is evident from the results of the orthogonal
factor analysis that a number of hypothesized associations
between the role-taking and other attributes and variables
have not appeared in the orthogonal factor analysis of the
entire group of politically-aware respondents. Although
the left-radical factor shows assumed similarity loading
negatively, which is in keeping with the results expected,
this has not affected role-taking accuracy. Moreover,
disowning projection has failed to appear on any factor
with the exception of the role factor itself.
138
Among the student respondents of the pilot study, dis-
owning projection loaded, as hypothesized, with advocacy of
change and radicalism. Among these older respondents, how
ever, the place of projection may have been taken by
political cynicism, and the ability to assess the attitudes
and behavior of legislators has not been impaired by a
strong desire for change. The easiest explanation for this
is that continued political activity and consequent maturity
has made these older respondents more accurate role-takers
and less apt to mistake the nature of role-encumbents tested
for here.
It was disappointing to find that the role-accuracy or
behavior score were not loaded with the efficacy factor,
which the hypotheses of Chapter II indicated. In fact, the
associations of these two correlated variables were
singularly arid.
Because of the nature of the results, the 57-case
study was refactored using another correlation coefficient.
The problems of the phi coefficient are well-known. In
the 2X2 table where one cell is empty, the phi coefficient
tends to give low correlations in the neighborhood of .30
to .53, whereas a correlation coefficient such as the, .
tetrachor~c will provide a higher correlation. If the
differences tapped by the role-taking variables are only
differences distinguished by either high or low scores on
these measures, then the phi coefficient would give
139
consistently low and possibly misleading correlations.
Thus, it might be possible that whereas low role-taking
accuracy may not distinguish respondents on an attribute,
there may be a drastic effect among respondents with high
role-taking accuracy. Under these circumstances, another
correlation coefficient might bring out such associations
more clearly.
Thus, although the cross-tabulations did not indicate
the foregoing as a major result, the tetrachoric correla
tion coefficient was chosen to compute another 33X33 matrix
(for the 57-case study) to input into the factor analysis
program. Although the loadings of the variables were con
sistently higher in the tetrachoric results, the pattern of
the factors was unchanged. No differences in interpreta
tion would have been necessary if the tetrachoric-matrix
had furnished the factors obtained with the phi-matrix in
the 57-case study. This attested to the stability of the
factors. Since this result was presumed to hold for the
same variables over the 7l-case matrix, the experiment was
not repeated for the final analysis.
6.2 OBLIQUE FACTORS OF POLITICAL ROLE-TAKING.
It is often the case, especially when work is done
with attitude scales and psychological variables, that the
factors extracted in an orthogonal analysis do not give the
best representation of the factor structure and pattern.
140
Since many of these factors are likely to be oblique, it
is best to seek the preferred simple structure in an oblique
rotation. The results of the seven-factor orthogonal study
were therefore entered into an oblique factor routine. 8
Table 6.9 presents the results of this oblique factor
rotation. It can be seen that while the loadings on some
variables increased or decreased slightly, the factor
pattern and structure were not altered very much. It is
proper to interpret the orthogonal results, which are
similar to the oblique rotation; in fact, the descriptive
names of the variables do not have to be changed. For the
analysis of the role-taking variables, however, the oblique
analysis is somewhat preferable, as the discussion will
show.
The intercorrelations among the seven oblique factors
are generally low. However, two of the highest inter
correlations between factors include the role factor. The
role factor correlates .40 with the factor that has been
called political empathy. Thus, there would appear to be
a moderate association between the two oblique factors that
somewhat vindicates the use of the term "empathy" in
describing factor four.
8The Oblimin program by John Carroll. A biquartiminsolution was derived, using 30 cycles and 50 iterations.The total of 10,500 iterations were used in deriving theresults, which are probably accurate to the two decimalplaces required.
141
TABLE 6.9 PATTERN MATRIX OF OBLIQUE BIQUARTIMIN FACTORSFOR POLITICAL-ROLE VARIABLES
(Only loadings greater than .30 are shown)a
Factors
Variables
HELOFFACQLEGPOSOFFISSACTSEXRESIDEELDONLEMPJOBEGOSTGRADICLFLEXBLFAITHPEQUALSOCIALINNOVTHUMANEPOLCYNPOLEFFPARTICINFSELPOLINTASPIREINFGRPCHANGESUCSELSUCGRPPOLKNOBSCOREROLEACASSUMEACTUALPROJECAGEASPGAPFINDEXAPOLIT
IRealEfficacy
4758383834
31
43
78
-56
IILeftRadical
6048
75-36
5836
-52
32
-66
IIIPoliticalEmpathy
67
-36
42
5558
-60
aFactor loadings are present without decimals. The numberof the factors does not correspond with the numbers givenon the orthogonal factors, but the factor names are thesame.
Table 6.9 Pattern Matrix of Oblique Biquartimin Factorsfor political-Role Variables (continued)
Factors
IV V VI VIIRo1e- Group
Variables taking Success Change Age
HELOFF 42ACQLEGPOSOFF 30ISSACT 35SEXRESIDE 67ELDONLEMPJOB -30EGOSTG -34 31RADICL 43FLEXBLFAITHPEQUALSOCIALINNOVTHUMANE 48POLCYN 42POLEFF 33PARTIC 62INFSELPOLINT 41ASPIRE -47INFGRP 48CHANGE 43SUCSELSUCGRP 56POLKNO -42BSCORE 62ROLEAC 63 -39ASSUME 32ACTUALPROJEC -66AGE 73ASPGAPFINDEXAPOLIT
142
TABLE 6.10 CORRELATIONS AMONG OBLIQUE FACTORS FOR POLITICAL-ROLE VARIABLES a
1 2 3 4 5 6 7
1 Real efficacy2 Left-radical 063 Political empathy 23 194 Role-taking 12 09 405 Group success 00 -04 03 056 Change 12 22 07 31 037 Age -08 09 -09 -01 -01 05
aThe factor numbers here correspondent to the numbering of the oblique factorson Table 6.9. These are the correlations among the primary factors of the patternmatrix given by TIT.
J-'+='VJ
144
The next largest correlation between factors is .31,
which relates the role factor to the change factor. This is
an interesting result with relation to the hypotheses which
introduced the analysis, for role-taking accuracy has ap
peared on the oblique change factor negatively associated
with the variables and attributes descriptive of change.
Disowning projection still plays no part in these associa
tions of active change variables, but it does playa small
part in the passive change variables associated on the
left-radical factor.
The conclusion that may be drawn from the factor
analysis is not entirely negative with regard to the hypoth
eses presented. The role-taking dimension does appear re
lated to some other political dimensions, including
variables found on the efficacy dimension, although the
loadings of these variables and attributes are rather low
when they appear on the oblique role-taking dimension. And
some role-taking variables appear on the change factor and
on the political empathy and left-radical factors. The
oblique factor matrix would seem to be more revealing than
the orthogonal analysis, and the role-taking associations
are more easily distinguished.
While the conclusions as to the importance of the
role-taking variables and their influence on the politically
aware ought not to be too sanguine as a result of the moder
ate associations shown on the oblique factor analysis, the
145
relationships are nevertheless present to a greater degree
than often characterizes comparable variables brought into
the political realm from other disciplines.
The results shown here for a matrix of 36 variables are
at least encouraging. To an extent, the results depend on
how well the political domain has been covered by the choice
of variables entered with the role-taking variables into
the factor analysis. If the domain of variables and at
tributes is fairly complete, then the associations shown in
the oblique factor analysis will tend to reappear with com
parable studies of the politically aware elsewhere in the
political culture of the United States. In any case, the
associations will not be diminished by the addition of new
variables to future analyses, for this will turn up new
dimensions and new associations. The common variance of
the variables will increase.
The conclusions derived from the factoring will be
returned to in the last chapter. In the next chapter, the
analysis of the entire group will be continued in a search
for the best predictors of advocacy of change and the two
success indicators. This will be done on the same level of
analysis as the factoring and cross-tabulations, using the
entire group of respondents. This section will be supple
mented with the results of a typology of political styles,
in which the search for associations of the role-taking
variables will be conduced on another level of analysis,
the level of subgroups or types among the respondents.
146
CHAPTER VII
THE BEST PREDICTORS OF SUCCESS AND ADVOCACY
OF CHANGE AND A POLITICAL-ROLE TYPOLOGY
7.1 REGRESSION EQUATIONS.
In order to determine those independent variables,
which would give the best prediction of the dependent vari-
ables, success and advocacy of change, a series of re-
gressions were carried out. This procedure provided an
exact test of how well the dependent variables of the study
were predicted to by disowning projection and role-taking
accuracy and at the same time revealed those variables that
might be better predictors. In order to assure the in
dependence of the eleven variables from which the predictions
were made, the variables were entered into the Mesa I
program to derive orthogonal dimensions. l
As a result, the following eleven independent
dimensions were obtained (see Table 7.1). All variables
entered into the program produced orthogonal dimensions
relatively uncontaminated with loadings from other variables.
lIn this step, the undichotomized raw data, includingthe Cornell scores for the attitude scales, were entered, aproduct moment correlation matrix was obtained with l's inthe principal diagonal, and a component analysis wascarried out.
148
The independent variables thus obtained are listed in
Table 7.1, together with the loadings on their orthogonal
factor:
TABLE 7.1 LOADINGS ON ELEVEN ORTHOGONAL FACTORSBY INDEPENDENT VARIABLES FOR THE REGRESSION EQUATIONS
Next highest loadingVariable name Loading on orthogonal factor
PROJEC 90 28 (ROLEAC)SOCIAL 97 -17 rUMANE
)PARTIC 96 -15 EGOSTG)POLKNO 99 11 PARTIC)POLCYN 98 10 EQUAL)BSCORE 97 -26 (PROJEC)POLINT 97 -12 ~PROJEClEGOSTG 97 -16 PARTICEQUAL 96 15 ROLEAC,
HUMANE)HUMANE 96 15 ~EQUAL)ROLEAC 94 32 PROJEC)
Standardized factor scores were obtained for the 71
cases over the eleven orthogonal dimensions of the indepen-
dent variables, and these were entered into a regression
program2 as predictors of advocacy of change (CHANGE),
subjective feeling of success (INFSEL), and objective
success (HELOFF). Held office was a dichotomized attribute,
but it is not inappropriate to predict to a dichotomous
dependent variable, although the results will not be as
reliable. The purpose of the~e regressions was not to in
fer from the results to a population, but to obtain infor
mation on which were the best predictors for this particular
2The BIMD 29, Dec. 1, 1961.
149
group of politically-aware individuals. Independent vari
ables were chosen so as not to be simply tautological. It
is obvious, but uninteresting, that one can predict to
advocacy of change quite well from the verbal-change item.,
No argument is advanced that the independent variables in
the equations precede in time variables and attributes
predicted to. The meaning of prediction is much broader
than that.
The greatest amount of the variance in the attribute
of holding office, the measure of objective success in
politics, was explained by participation, political interest,
sociability, political knowledge, and equality, as can be
seen in Table 7.2. These variables together explain some
34 per cent of the variance. The remainder of independent
predictors explain only six per cent of the remaining
variance. Among the first five variables are none of the
role-taking variables.
Although the 57-case study was not intended to be
compared to the final 71-case analysis, there was a dif
ference in the regression to having held office that
demands explanation, for in the former study role-taking
accuracy was among the first four variables, explaining some
five per cent of the variance. In the 57-case run, assumed
similarity was entered into the factor analysis to obtain
independent dimensions with ten other variables, including
TABLE 7.2 REGRESSION FROM SELECTED FACTOR SCORES TO HELOFFa
(N=7l)
Regression Computed Partial F-value Cmnulated Proportion of2Variable coefficient T-value Cor. coef. each term F-value variance (R )
PARTIC .17 3.66 .43 10.93 10.93 .14POLINT .15 3.19 .38 9.27 10.75 .. 24SOCIAL .09 1. 92 .24 3.50 8.60 .28POLKNO .08 1. 71 .22 2.83 7.33 .31EQUAL -.08 -1.64 -.21 2.67 6.55 .34BSCORE .08 1.60 .20 2.61 6.03 .36HUMANE -.07 -1.41 -.18 2.06 5.55 .38POLCYN -.03 -.71 -.09 .51 4.88 .39EGOSTG .03 .66 .09 .44 4.35 .39PROJEC -.03 -.61 -.08 .37 3.91 .39ROLEAC .03 .57 .07 .32 3.55 .40
aThe results are given complete together with the statistical tests. These testsindicate only the risk involved in accepting the results for this particular group ofpolitically-aware respondents. Where the predictability is very low, the significancetests mean only that the signs of such small figures are beyond statistical guarantees.Signs were alterred in the results according to the direction of scoring of the rawdata and the factor scores of the Mesa I program from which the independent variableswere derived. In these associations, the independent variables can be predicted fromthe dependent variables as well, and no causal sequence is attached to their independentdesignation.
t-"V1o
151
role-taking accuracy. It loaded .55 on the factor with
role-taking accuracy, and, since it did not form an
orthogonal factor of its own but contaminated a number of
other variables. the decision was made to drop it from the
7l-case regression analysis. The effect of such a con
taminated variable, together with 14 additional conservative
respondents, may have alterred the results obtained between
the two computer runs.
In predicting to the second criterion of political
success, self-assessed influence, three variables, partic
ipation, political interest, and ego strength, explain
together some 41 per cent of the variance. The remaining
eight variables add only six per cent of the variance. One
role-taking variable appears fifth in the amount of variance
that it predicts, although the statistical test is in
significant. However, were role-taking accuracy removed,
the explanation for only some two per cent of the variance
would be lost. It is certainly not a very important pre
dictor.
Five predictors explain some 41 per;cent of the
variance in advocacy of change. These are equality, partic
ipation, sociability, political cynicism, and political
interest. None of the role-taking variables appear important
in explaining the variance in advocacy of change, refuting,
on this level of analysis, the hypotheses advanced concerning
them.
TABLE 7.3 REGRESSION FROM SELECTED FACTOR SCORES TO INFSEL (N=71)
Variable Regression Computed Partial F-va1ue Cumulated Proportion of 2name coefficient T-va1ue Cor. coef. each term F-va1ue variance (R )
PARTIC .52 5.28 .57 21.40 21.40 .24
POLINT .39 3.89 .45 13.69 19.51 .36
EGOSTG .24 2.38 .30 5.48 15.69 .41
SOCIAL .16 1.66 .21 2.73 12.75 .44
ROLEAC .16 1.61 .20 2.62 10.98 .46
HUMANE .15 1.55 .20 2.49 9.77 .48
POLCYN -.11 -1.07 -.14 1.19 8.57 .49
EQUAL .07 .75 .10 .59 7.52 .49
PROJEC .07 .73 .09 .55 6.70 .50
BSCORE .06 .57 .07 .33 5.99 .50
POLKNO -.01 -.11 -.01 .01 5.36 .50
I-'VIN
TABLE 7.4 REGRESSION FROM SELECTED FACTOR SCORES TO CHANGE (N=71)
Variable Regression Computed Partial F-va1ue Cumulated Proportion of 2name coefficient T-va1ue cor. coef. each term F-va1ue variance (R )
EQUAL 1.09 4.89 .54 19.50 19.50 .22
PARTIC .63 2.82 .35 7.09 14.16 .29
SOCIAL .52 2.34 .29 5.14 11.73 .34I
POLCYN .43 1.94 .24 3.69 10.07 .38
POLINT .42 1.90 .24 3.68 9.12 .41
BSCORE -.26 -1. 20 -.15 1.48 7.90 .43
HUMANE .26 1.17 .15 1.40 7.02 .44
PROJEC -.25 -1.11 -.14 1.28 6.33 .45
EGOSTG .13 .60 .08 .36 5.61 .45
POLKNO .13 .57 .07 .34 5.02 .46
ROLEAC -.03 -.13 -.02 .02 4.49 .46
r-'VlW
154
Rather than the role-taking variables, it appears that
other more common political variables are far better pre
dictors of the dependent variables. These best predictors
turn out to be participation, political interest, sociabil
ity, political knowledge, ego strength, political cynicism,
and equality. Role-taking accuracy and the behavior score
do not appear to contribute much to an explanation of po
litical success, either subjective or objective.
Projection does not contribute much to the prediction
to advocacy of change, but it may be argued that its place
has been taken by political cynicism instead. In fact,
this is not entirely inimical to the hypothesis that dis
owning projection would relate highly to advocacy of
change, for advocates of change do attribute negative
aspects to the system they intend to alter. But this at
tribution finds expression in political cynicism rather
than a tendency to project.
These results are indicative of the variables that
ought to be used, rather than the role-taking variables, to
predict to success and advocacy of change. While no
attempt was made to put variables together into a causal
sequence, it is possible to suggest a few plausible re
lationships among the dependent and the independent
variables that turned out to be good predictors.
155
It would seem likely that participation and political
interest preceded holding office in time. Persons who put
themselves forward, as evidenced by the importance of
sociability in predicting to this attribute of success,
would be most likely to be eventually chosen for some
political office or another. Persons who make it a point
to participate, which indicates putting themselves forward
(sociability again), and who evince interest in politics
would seem eventually to appear to others to be appropriate
candidates for offices within an organization. The po
litical knowledge may precede or follow holding an office.
Participation and political interest are most likely
to precede in time self-assessed influence as well, for that
dependent variable was apparently based on tangible evi
dences of success. These tangible evidences appear to
precede influence, for it was success in alterring a law
that was the variable unattached to tangible activity.
Influence see~ed to represent accomplished fact, while
success was a measurement of an individual's potential.
Ego strength might precede or follow influence in time;
possibly, like some of- the attitudinal variables, it both
feeds and is fed by political successes.
Advocacy of change is a most interesting case. Here
it would seem that a major incentive for advocating change
is the espousal of certain of one's attitudes, such as
equality. This plunges a person into political activity,
156
into participation as such, but it does not necessarily
cause him to want to become part of the system, as witness
the negative connection between equality and having held
an office. The political cynicism might well follow the
results of advocating great change in a political system
that is quite often adverse to changes. The establishment
espouses equality but successfully prevents many measures
toward real equality among all Americans from reaching
fruition.
In these sequences, unfortunately, the role-taking
variables cannot be said to play even a minor role. What
ever skill is tapped by role-taking accuracy and the
behavior score does not distinguish among the po1itica11y
aware in general. The role-taking variables certainly do
not enter into the ability to predict the variance in the
dependent variables of this study when the entire group is
analyzed.
7.2 POLITICAL PROFILES.
In the final search for the possible influences of
the role-taking variables, the level of analysis was
changed once more, and a set of political-role profiles
was empirically derived from the 71 respondents. The
process by which this was done was somewhat more involved
than the preceding analyses have been.
The dichotomized variables that had provided the phi
matrix and the factor results of Chapter VI were entered
157
into the Mesa I program, and a product-moment correlation
matrix was obtained. Although the phi-coefficient is an
approximation to the product-moment coefficient for
dichotomous data, the matrix, generally the same as the
original phi-matrix, was not exactly the same. 3 The
criteria for the common factor analysis was the same as
for the phi-matrix, all factors with eigenvalues above 0.95
being rotated to a varimax solution. This provided seven
orthogonal factors that were almost identical to those for
the original phi-matrix, which clearly vindicated the use
of this result to obtain factor scores for the 71 cases
over the seven orthogonal political-role factors. 4
A word of caution must be entered concerning the
factor scores. The factor scores obtained from a com-
ponent factor analysis, in which there are as many factors
as variables, are exact estimates according to the equation
A = FS,
where A is the data matrix, F is the factor matrix, and S,
the factor score matrix obtained by algebraic calculations.
3A subroutine to read in the phi-matrix into the Mesa Iprogram (or any other correlation matrix) so that factorscores might be derived on the basis of that matrix was notcompleted and tested in time for this analysis.
4These factor scores in a common factor analysis arenot standardized in the Mesa I program. However, they werestandardized in the subsequent Revised Distance Program.
158
But the common factor analysis produces fewer factors
than variables. The factor scores are therefore estimates
of the true scores. Harman shows that the standard de-
viation of the estimated factor scores is equal to the
multiple correlation of the estimated scores with the
variables of the data matrix. 5 The standard deviations
of the seven factor scores from the product-moment cor
relation matrix input are shown in Table 7.5.
TABLE 7.5 STANDARD DEVIATION OF FACTOR SCORES FORSEVEN FACTORS FROM A COMMON FACTOR ANALYSIS
OF A PRODUCT-MOMENT CORRELATION MATRIX
Factor name
RoleLeft-radicalReal efficacyAgePolitical empathyGroup successChange
Standard deviationof factor scores
.865
.909
.899
.878
.881
.867
.869
Since the communality estimates in the principal
diagonal produced a non-Gramian matrix, negative eigen
values were produced with the result that the factor score
estimates were deflated, and the standard deviations under
estimate the multiple correlation between the estimates for
5See Harry H. Harman, Modern Factor Analysis, Chicagoand London, 1967, pp. 353-354.
159
the factor scores and the variables of the raw data. This
fact gives the following discussion a conservative bias.
A further problem is that the factor estimates are not
unique for the correlation and factor loading estimates.
The minimum correlations between alternative factor score
matrices depends on the multiple correlation of the re
gression estimates in the original data. The equation
2r m = 2R1 - 1,
in which r m is the minimum correlation possible between
alternative factor score matrices and R1 is the multiple
correlation of regression of the factor score estimates
on the m variables of the data matrix (provided by the
standard deviation of the factor scores).
Solving the equation for the lowest value in Table 7.5
gives
r m = 2(.865)2 - 1 = .496
Thus, these factor scores are correlated at least 50 per
cent with a maximally different possible set of factor scores
from the same factors. The use of the highest row corre
lation as the communality estimate has given factor score
estimates that are neither exact nor unique. 6 In the
succeeding analysis, however, where no statistical in
ferences are made, these scores are adequate in the opinion
6The foregoing discussion has been adapted fromRudolph J. Rummel, Ap~lied Factor Ana1~sis, Honolulu,Hawaii, mimeographed raft, i.pr11, 196 , Chapter 16.
160
of this researcher. It is important, however, that un
warranted conclusions are avoided by knowing exactly what
assumptions have gone into the analysis.
These factor scores for 71 cases over seven political
role factors were entered into a program that calculated
the distances between cases by means of the standard dis
tance formula. 7 The raw distance matrix obtained was
tran~formed so that all the values ranged between a minimum
value of 0, indicating the greatest possible distance be-
tween cases, and a maximum value of 1, indicating the
maximum similarity between cases. 8 Since a case is max-
imally similar to itself, this placed l's in the principal
diagonal of the matrix, now a "similarities matrix."
The similarities matrix was entered into the Mesa I
program, and all factors with eigenvalues above 0.95 were
extracted and rotated to a varimax solution. This provided
7The general distance formula is Dil = (~di12)1/2,
where D is the linear distance between the cases i and 1 ina seven-dimensional space and dil is the algebraic dif-
ference between the factor scores of i and 1 on the samedimension; summation is over k dimensions.
8 m•.By the equation n iJ· = 1 - M~] , where m·· is a raw-
L ~J
distance matrix element, ML is the maximum raw distance
element in the matrix and nij is a similarity. Thus, for
an element with the greatest distance, mij = ~, nij = 0,while for an element describing minimum distance, adiagonal element, nij = 1. The computer program for this
161
eight factors on which cases were grouped according to the
distances between them on the original seven factors of
the political-role space. 9
Groups on the eight factors extracted from the distance
(similarities) matrix were designated as those cases with
loadings equal to or above .40 on the same factor. This
held the number of cases appearing on more than a single
factor to a minimum, but there were a number of cases that
did appear on two, rarely more, factors. This is not ad
verse to the analysis of the political-role types, however,
since it is quite possible that a single individual may
1 Ii I , '1 10span severa persona ty or po ~t~ca types,
step was the Revised Distance Program, Carol Hopkins,August 6, 1965.
9These eight groups were clearly demarcated. Theeighth factor rotated had an eigenvalue of 2.735, while thenext highest eigenvalue was 0.543. The eight groups werealso clear in the 57-case analysis; moreover, the formeranalysis provided profiles most of which were similar tothe eight profiles coming from the complete 7l-case analysis.This line of analysis was not pursued in the pilot study.
lOIn the 57-case analysis, groups were first formedby choosing all cases above .30 on a factor. Profiles wereobtained and compared with profiles obtained from groupingsof all cases loading above .45 on each factor. The profiles did not change in pattern, but they were more clearlydelineated in the latter case, where extremes appeared onthe same variables but accentuated.
162
The differences between these groups were hidden in
the abstract analytic realm of mathematical manipulations.
One way to discover how these eight groups differed from
each other was through a computer program that provided a
plot of the means and standard deviations of each empir
ically derived group over either the original seven
orthogonal factors or over certain of the original vari
ables used in the study.ll The latter profile was the one
obtained for the purpose of the analysis in this section,
for the reason that differences between groups differing
on the role-taking variables could be more easily seen and
interpreted.
Figure 7.1 diagrams the eight political-role profiles
on 24 selected variables (none dichotomous). Each profile
depicts a political type, a subgroup that exists among the
entire group of politically-aware respondents from two
organizations in Hawaii. The profiles are drawn according
to the deviations of each subgroup from the mean of the
entire group, which is shown by the straight lines in
Figure 7.1. In only two cases, humanitarianism for Sub-
group VI and participation for Subgroup VII, do the
deviations from the mean of the entire group exceed one
lIThe profile program was devised by Dennis R. Hall.See Research Report No. 14, The Dimensionality of NationsProject, Honolulu, Hawaii, 1968.
PARTIC
IHFSEL
POLINT
ASPlNF
lNFGRP
CHANGE1/ I 1 / \ 1
I-tjH
(f I If , I SUCSEL......... C)
~~~I) " I \ I I SUCGRP~
'-I '-I.~ JI II 11 ~ POLKNOZ
N
" tr1~H / I J 1\ II BSCCREVIC).. ::t:~
Z J I r I II II ROLEACW ""d
" 0~tool.j::--H ( I \ I U 1\ ASSUME.. ~
HZCJ
.j::--,,~ \..1 \. 1 11 1 J ACTUALt-' IVI:;t)~o
I'" I') ( 1 (I PROJECtooltr1
""d 1 J ~ I l. 1\ EGOSTG:;t)0I-tjH I I " 1 ,,/ I RADlCLtooltr1(J) .. - . - ~
FLEXBL
FAlTHP
EQUAL
SOCIAL
lNNOVT
HUl·:ANE
POLCYN
£91 /(\ \ I I POLEFFH H<: H 1-4• . •
PARTIC
INFSEL
POLIN'r
ASPIUF
INFGRP.. IF \ I I It-xjH
-G1 I \ rl \ I \I CHANGE\JlZ ~
1Itr:l..... I J \1 J I 1\ SUCSEL00-.....1. .
.....II f\ , I IIZ SUCGRP
0'\1Itr:l
..... HoG1 I ) ( I { I 1'\ POLKNO
'" ::J::H
Z/' I " f\ I \ BSCORE-.....It"'d
II 0..... L4+,-H
1. JI I \ I J'" H ROLEACH
Z(')
ooll~ q ( I I ) I , ASSUHE\D I"""':;0
0I '\ \. /1 I )L4 ACTUAL
tr:l
t"'dI / I \ l I ("I PROJEC:;0
0t-xjHL4 I i f I P l\ EGOSTGtr:lCJ). I I ~I / I
"(')
RADICL0
~ I } I') \. I 1\ FLEXBLH
~Ir 'l /1 V
FAITHPa
EQUAL
SOCIAL
INNOVT
ntJHANE
\\ I
" I '" I POLCYN
"791 \I I' I' POLEFF<:H <: <: <:H H H .H H •.
165
standard deviation. The tendencies distinguishing groups
are made apparent by differences in patterns rather than
levels of activity. Moreover, since these are subgroups
from among a rather small number of respondents, the N's
are small, and no argument is advanced in the speculation
that follows concerning the generalizability of the pro
files to the politically-aware of the United States or even
of Hawaii. The results are dependent on replication, and
if similar profiles emerge from a number of studies of
politically-active groups, more faith can be placed in
the results.
The most striking result for the purpose of the re
search here is the emergence of two groups, II and IV,
that accentuate the advocacy of change variable. It is in
Subgroup IV that the pattern hypothesized between advocacy
of change and projection is substantiated. Here a high
tendency to advocate change is associated with one of the
highest tendencies to engage in disowning projection. More
over, this subgroup has the lowest tendency to assume
similarity and is deficient in role-taking accuracy. The
behavior score is also low.
These characteristics are associated in Subgroup IV
with relatively high participation and a political knowl
edge score that does not seem deficient with relation to
the entire group. It is this subgroup that is also
particularly high in political cynicism, which substantiates
166
the suggestion offered in the regression analysis that this
scale variable may be associated with the tendency to dis
owning projection and may replace it in politically ex
perienced groups as evidence of frustration.
Subgroups V and VI are distinguished by a higher than
average accuracy in role-taking, and in Subgroup V this ac
curacy is apparent also in the behavior score. In keeping
with the hypothesis concerning the association between ro1e
taking accuracy and success, Subgroup V is also notably high
in self-assessed influence. This association is not so
marked in Subgroup VI. Role-taking accuracy is associated
in Subgroup V with higher than average participation, an
advocacy of change that does not depart from the mean of
the entire group, low projection, low political cynicism,
and few other noteworthy characteristics.
Subgroup VI tends to hover near the mean, although
the success variable is slightly higher than average.
Striking, however, is the low tendency to project, the
conservatism of the subgroup, the tendency to sociability,
and the extremely low rank for humanitarian attitude. It
may be said that both of these subgroups, V and VI,
represent the relatively conservative tendency to work
within the system, learning the political style that
exists about them, and, consequently, achieving a measure
of real success, success probably not closely connected to
issue activity. These are the accurate role-takers, and
167
they look rather similar to what the hypotheses advanced
in Chapter II suggested.
Although these subgroups tend to afford a measure
of evidence in favor of the major hypotheses advanced at
the beginning of this research, a number of subsidiary
hypotheses are not too well substantiated. It was opined,
for instance, that such variables as humanitarianism,
faith-in-people, flexibility, ego strength, and equality
might characterize the more accurate role-takers. But
humanitarianism and equality were found to be attitudes
motivating persons to advocate change, and they are not
found within these subgroups to be associated with accurate
role-taking. In fact, in Subgroup VI, they are found to be
disassociated with role-taking accuracy.
It is among the styles that combine advocacy of change
with low role-taking accuracy or a fairly low behavior
score in which strong tendencies are found for high
humanitarianism and egalitarianism. This is notable in
patterns II, III, and VIII.
Profile I appears to depict a type of conservative
style, in which participation and personal success as well
as the perception of the success and influence of the
group are all low. In the political milieu of the United
States, this profile might represent a moderate and rel
atively inactive conservatism. Perhaps many moderate
conservatives can be placed in such a category, in which
168
assumed similarity, ego strength, faith-in-peop1e, and
sociability are somewhat high, while political cynicism
and efficacy are rather low (but not extremely), and the
radical-left motivations for advocating change, equality
and humanitarianism, are below the mean for the whole group.
This moderate conservative group has not been forced
into a very active defense of the system, in which it ad
vocates relatively few changes, and persons with such a
political style can more or less take things as they are.
It is possible that persons exhibiting such a political
style are not more accurate role-takers than they are
simply because they have not been strongly motivated to
plunge very deeply into the political arena.
Profile II is descriptive of one form of radical
style, to judge from the high tendency to advocate change.
These individuals do not stress their personal success or
influence, tend to participate somewhat more than average
in politics, rate fairly high in political interest, and
espouse causes consonant with the attitudes of humanitarian
ism and egalitarianism. They tend to be somewhat cynical
about politics, although this does not cause them to be
below the average in a feeling of political efficacy.
Notably, their espousal of change has not impaired either
their ability to take the attitudes of the legislator or
their prediction of his behavior in a situation. Their
tendency to project is fairly low. These would appear to
169
be radicals, who, while advocating large change, neverthe
less have accepted the system and are working within it to
bring about change.
It is somewhat anomalous to find a profile such as III
among the politically active individuals of this study, and
it must be kept in mind that all tendencies are relative to
the group being studied. Relative to the entire group of
respondents, therefore, Subgroup III does not participate
much, has little feeling of personal or group success, and
is quite bad in predicting attitudes, although the behavior
score is not below the mean. Among these individuals may be
discerned some evidence that sociability and ego strength
must also be present if one is to put himself forward
enough to learn about the attitudes of others. Thus,
members of Subgroup III may be relatively retiring persons,
whose membership in a political organization may be the re
sult of a mild interest in some causes concerning equality
and humanitarianism.
Subgroup IV has already been delineated. These are
the radical change advocates high in disowtling projection
and political cynicism who disavow the system enough to
suffer impairment of the ability to accurately role-take
or to predict adequately in the behavior area. Quite
possibly, these are the persons more issue-oriented than
concerned with other political values, such as seeking
office. Subgroup V may be deemed the real politicos among
170
the respondents in this study. These are persons ranking
quite high in participation and personal influence and
efficacy. They are not cynical about politics, nor do they
tend to advocate great change. They rate among the highest
in political knowledge, and the amalgam of political
strengths is buttressed by high rank on the role-taking
variables.
Persons who participate least of all among these
respondents compose Subgroup VII. Rather conservative
among the other respondents, these persons do not assess
their personal efficacy and influence very high, but they
are somewhat more sanguine about the success and influence
of the group to which they belong. In spite of their own
rather low efficacy, they are not politically cynical. It
has been suggested previously in the results from the factor
analysis, that there may be some individuals whose own
relative lack of political experience and feeling of strength
in politics may be compensated by joining some organization
in order to participate vicariously in the importance and
success of the organization. These characteristics are
found associated here with a very low tendency to socia?i~-,
ity and ego strength. Such individuals are too shy to put
themselves forward in politics. They may comprise the
followers within the group.
Finally, another moderately radical group completes
the series of political styles derived by the empirical
171
search for groups. Profile VIII unites a role-accuracy
score at the mean with a very low score on predicting
behavior. Moreover, the disowning projection score is
rather high, although the subgroup is not particularly high
on advocacy of change. This group is one of the highest in
political participation, and the very high tendency toward
humanitarianism may explain the moderate radical motivation
for the political activity of these persons.
It can be seen that Group A and Group B include a
variety of political styles, among which appear the re
lationships sought for in this research. They did not
appear in the cross-tabulations, although the pilot study,
where disowning projection appears distinctly associated
with advocacy of change, may have tapped more salient ex
amples of Subgroup IV, nor prominently in the factor
analysis, for the reason that they were hidden among groups
with other political styles. Although each of the groups
tapped, Group A and Group B, may have a profile describing
the entire group, it is not necessary that such a profile
represents any more than a composite of several distinct
subgroups.
Figure 7.2 illustrates the extent to which these
empirically derived subgroups of the entire group of
respondents can be found scattered within the separate
groups. While Subgroups II and VIII are generally confined
to membership in Group A, and Subgroup I is a conservative
I-
GROUP A GROUP B
FIGURE 7.2 EIGHT POLITICAL-ROLE TYPES FOUND IN TWO VOLUNTARY POLITICALORGANIZATIONS (For a description of types, see profiles inFigure 7.1)
......"-JN
173
prototype, the other subgroups bring together members with
the same political style from both groups in approximately
equal proportions. The divisions of the two groups in
Figure 7.2 are intended to present an idea of the pro
portional size of each of the subgroups.
It is quite possible that Figure 7.2 describes at
least some of the political realities of a mature political
system, where conflicts are not so acute that political
types are confined to either one group or another but over
lap. In any case, it is quite evident why the hypotheses
concerning the role-taking variables and associations could
not be clarified until the level of analysis reached the
subgroups.
Although there is much in the foregoing analysis that
supported the hypotheses presented in Chapter II, caution
must be used in accepting any of the rather imaginative
interpretations without further replications. The slight
differences indicated in Figure 7.1 indicated patterns that
distinguish groups on the basis of a mathematical distance
formula. In only a few instances, do these differences in
pattern exceed a single standard deviation from the mean.
In some instances, the patterns derived appeared to clarify
results obtained from the cross-tabulations and the factor
analysis, while in other cases there may appear to be a
contradiction between the profiles and results obtained,
especially from the regression analysis, which was based on
the entire group of respondents.
174
Nevertheless, allowing some faith in the efficacy of
the empirical technique that isolated the eight groups, it
seems that among these political styles are several that
lend support to the theory that tied the role-taking vari
ables to the political realm. While this is gratifying
and indicative of possible directions for the future,
preferably with the use of larger and more broadly-based
groups of respondents, it does not fully establish the
importance of the role-taking variables for political
analysis. More will be said on this subject in the con
cluding chapter.
CHAPTER VIII
CONCLUSIONS
8.1 RECONSIDERATIONS OF THE HYPOTHESES.
Two major hypotheses stimulated the research on the
role-taking variables. The first, as set out in Chapter II,
related disowning projection to high advocacy of change.
Neither role variable was significantly related to the
variables tapping advocacy of change directly or indirect
lyon the analytic level of the entire group. The factor
analyses provided some moderate associations between the
role variables and some of the political variables. More
over, the oblique factor analysis turned up moderate
correlations between the role factor and the political
empathy and change factors.
It may be argued that the change tapped in this re
search was not extreme enough to merit a test of a prop
osition that had to do with disaffectation from the system.
However, many more respondents than expected responded to
the change item by indicating level 10 as descriptive of
the amount of change they would like to see in the system.
This was in some measure unanticipated, and a review of the
question prior to distributing the questionnaires raised
some doubt as to whether anyone would be motivated to re
spond to the 10-step on the change level, so extreme did
176
the wording of the question appear. But respondents used
the extreme value easily enough. If any difficulty is
apparent, it could only be the lack of respondents from
the radical right-wing of the political spectrum. It is
possible that none of the eight political-role profiles
here would be appropriate for the profiles of such re
spondents.
The pilot study of 67 student respondents had shown a
slight correlation between disowning projection and the
change variable, which in that study was a verbal question
similar to 2.14 in the present questionnaire; and the
factor analysis, a tetrachoric matrix input into the Mesa I
program, showed that disowning projection did indeed load
on the change dimension. It is possible that the effect
predicted by the hypothesis is brought out more strongly
among persons who feel a need for a change and who are too
young to have been issue-active or to have engaged the
system enough to have learned much about its role-encumbents.
Among such persons, there may indeed be a tendency for dis
owning projection, but this tendency did not hold in gen
eral for the politically-aware individual in the final
study, although it was traced to one of the radical sub
groups.
This result indicates the dangers inherent in
selecting respondents from among students only, as is often
done for studies of this type. Generalization beyond the
1771group tested is net warranted, and no conclusion may be
derived concerning even the student group after it has
entered the political arena "for real." In the case of
political variables, students may very well be different
than persons even a few years their senior.
The second major hypothesis related role-taking ac
curacy with both objective and subjective political success.
Disowning projection was to predict to a lack of success.
This hypothesis was also not substantiated by the research
on the between-groups level and when within-groups variance
was studied for the whole group of respondents. The holding
of office, issue activity, and political interest associated
with the role dimension on the factor analysis were moder-
ately low and tended to discourage much speculation that
the role variables might be a vital factor in politicalI
success. The regression analysis made this assessment
quite clear.
Comments on the remaining hypotheses have been made
throughout the course of the discussions on cross-tabulation,
the factor analyses, and the profiles. It is interesting,
that having interacted with politicians, such as legis
lators, appears to have made no difference in one's
IThe same warning holds for the respondents in thisstudy as well, but they are nevertheless closer to thepopulation of adult politically-aware activists than asample of students would have been.
178
role-taking accuracy. Nevertheless, having held office
does load moderately in the positive direction with role
accuracy and assumed similarity. Moreover, at the level of
subgroups, at least one group was isolated in which role
taking accuracy may have been vital enough to have been a
contributor to real success. The empathetic professions,
a variable that may merit more measurement experimentation,
showed no more ability than others in accuracy, although
this is in keeping with social psychological findings.
There was an interesting relationship in the factor
analysis showing actual similarity to load on the political
empathy factor, but this did not contribute to making these
people higher in role-taking accuracy. Motivation appeared
to be related to role-taking accuracy through the small
loading of issue activity and having held office. It also
appeared to have influenced at least one change-oriented
profile to more accurate role-taking in spite of a high
advocacy of change. Obviously, persons may be motivated
to participate in issues, but this motivation does not lead
them necessarily to become better role predictors. It may
also be suggested that it might lead them to be better
behavior predictors, but that variable was even less dis
criminating than role-accuracy.
In general, the hypothesized relationships between the
role variables and political attitudes did not hold when
the entire group was analyzed. The hypothesis suggesting
179
that men would be somewhat better than women in ro1e
accuracy--proposed for the reason that persons were pre
dicting to the attitudes of males (mainly) in an arena
that has traditionally been mascu1ine--fai1ed to hold.
Sex emerged as generally unrelated to role-taking dif
ferences.
The profile analysis uncovered many of the patterns
that had originally been predicted, and they were some
times clearly evident. Even here, equality and human
itarianism in the political context proved to be inimical
to role-taking accuracy. It must be kept in mind that the
research was conducted, first, on group differences and,
second, on politically-aware individuals as an entire
response group. The subgroups on which the profiles were
established are an empirical clustering of individuals who
were not shown in this study to act as political units.
While it may be of interest in describing various political
personalities to attempt to replicate these patterns among
other selections of the politically-aware, it is question
able whether knowledge about these patterns will aid in
political predictions. The political profiles isolated
here apparently operate as within-group distinctions, and
they may perhaps be found in all voluntary groups without
characterizing the activity of the group as a whole.
In studies replicating the search for political-role
profiles, it would be illuminating to produce within-group
180
sociometric diagrams to uncover whether persons of a
particular political type tend to associate more with
others of the same type, especially with regard to voting
and decision-making. Thereby, these profiles of political
style will be made pertinent to the political scientist in
studying pressures on decisions. When this has been done,
the influence of role-accuracy and disowning projection
will be even clearer in their political effects. No
evidence can be presented in this research that political
types within a political organization act in concert.
With qualifications, it may be concluded that role
taking accuracy may be one of the skills cultivated by
certain of the politically successful, especially among the
moderate conservative element. However, there is no evi
dence that disowning projection, an attribute of certain
of the radical subgroups and related to role-taking in
accuracy in these cases, is necessarily conducive to
failure. There would seem to be a decided preoccupation
among such subgroups, however, with the "issues," as opposed
to other political aims.
Besides the positive results from the profiles, there
are at least two other possible qualifications to the con
servative conclusions on the importance of the role-taking
variable array. The first concerns operationalization.
When variables as subtle as the role variables are oper
ationalized, there will always be some question as to
181
whether projection or role sensitivity are really being
tapped. One response to this question is that what is
tapped is defined by the operation used in tapping it.
That is, role-accuracy is the variation tapped by the
respondents' answers to the 35 questions put forth here as
a test of role-accuracy, while disowning projection is the
score manipulation used here. This extreme operationalism
is not intellectually satisfying, however, for the reason
that if role-accuracy exists in the sense in which the
researcher has hypothesized it, then it will be related to
the variables to which it is by theory supposed to be re
lated. If not, then either the operationalization was
inadequately sensitive, or else, the role-variables are
unimportant in the political realm in which they are being
tested.
The question of a possible lack of sensitivity has
been touched on previously, when it was suggested that the
questionnaire might present three response choices: "agree,"
"disagree," and "no consensus." The five nonconsensual
items presented the respondents (see Chapter IV) gave a
firm indication that this would indeed make for a more
sensitive questionnaire. The results of the analysis have
at least been encouraging enough to justify the attempt to
strive toward a more sensitive measure of role-taking ac
curacy.
182
A second qualification arises in the nature of the
two groups tested. There is no doubt that the variables
underwent a stringent examination in this research for the
reason that almost all the respondents were high in polit
ical knowledge, political interest and class. By being in
political organizations in the first place, they were cer
tainly at a high level of political participation.
The reason for conducting a test of the variables at
this level was to determine whether they would be sensitive
to political success and advocacy of change among persons
who might be very difficult to distinguish on the basis of
other attributes.
There are many variables that serve to differentiate
between the politically inactive and uneducated and the
politically active and well-educated. There would be very
little contribution made by the addition of several more
distinguishing variables to the array already available to
study groups far from one another in income, interest,
participation, and so forth. It would be especially
irrelevant when consideration is given to the added effort
required to adequately tap role-taking accuracy differences.
If the variables can provide no important distinctions on a
level on which individuals are otherwise on a par, then
there is relatively little reason to make use of them to
distinguish among groups with large differences.
183
While the regression analysis has suggested the
propriety of adherring to the common measures of political
participation, knowledge, and interest, it also revealed the
importance of attitudes as motivations for political activ
ity. Equality and humanitarianism emerged as vital pre
dictors of advocacy of change, while sociability appeared
a fair predictor of real success in politics. The role
taking variables, however, did not appear viable predictors
on the level of the entire group. It is on this level, and
on the between groups level, that the traditional political
predictors revealed their importance. On the level of
political subgroups, however, the role-taking array came
into its own.
8.2 SUGGESTIONS FOR FURTHER RESEARCH.
The suggestions made in the first section of this
chapter indicate that among large groups and organizations,
it is the variables traditionally put to use in political
science that are the most revealing. The regression
analysis intended to provide alternative courses to the
role-taking variables for the prediction of success and
advocacy of change among whole organizations. It ought to
be a principle of research that the measures used be no more
complicated to use than is needed for maximum predictability.
In such contexts, the role-taking variables fail.
However, the role-taking variables appeared more clear
ly in the profile analysis to provide possible indicators
184
of political activity. In effect, this was the level of
analysis on which such varia9les emerged into importance
among the politically-aware of the Hawaiian political
culture.
At this level, further research is indicated. A
beginning might be to ascertain, as was previously sug
gested, whether or not persons of similar political-role
type associated more with one another within political
organizations. If so, are these subgroup movements
influential in directing the decisions of the entire group?
This may provide important conclusions about the study of
group decision-making and clarify the tactics of political
organizations. Regression predictions within separate sub
groups may reveal that the role-taking variables do become
important.
Comparable studies of groups within underdeveloped
nations might reveal whether, for example, there are small
coteries of leaders with similar profiles from group to
group, with followers having profiles marked by low partic
ipation and role-taking accuracy. Do the political profiles
of leaders, as were some profiles uncovered among the more
radical group in this analysis, lead to change or are they
more conservatively oriented toward other styles of role
accuracy?
A further and important step in analysis of this type
would be the reinterviewing of respondents with particular
185
profiles. Since the respondents in this study were
anonymous, there was no opportunity to recontact those with
profiles of interest to this study, although some additional
knowledge about them may be gleaned from a closer study of
their questionnaires. Most important is the opportunity
offered by a reinterview to validate the measures of role
taking accuracy, disowning projection, and assumed similar
ity used in this research. The behavior variable was used
for this purpose, but, as the profiles and factor analysis
revealed, probably tapped a somewhat different dimension
than the attitude measure.
The problem of linearity, although not crucial in some
respects (for instance, the product-moment component factor
analysis revealed a Gramian matrix) demands attention. If
the results of other analyses do prove fruitful for political
prediction on the basis of the subgroups, it would be best
to provide a measure of assumed similarity that was not
part of a scoring procedure to extract other variables.
The same would be true of the measure of disowning pro
jection, although the behavior score did reveal that not
all of the negative association between PROJEC and role
taking accuracy need necessarily be attributed to scoring
artefact.
In the study of legislatures, the process of obtaining
the consensual items for the further testing of role-taking
accuracy leaves much to be desired. If research is to be
186
conducted on the importance of projection and role-taking
accuracy within the legislature itself, a questionnaire
should be used that not only calls for the attitudes of
the legislator, but determines his guess as to whether or
not his own attitudes are held by most of his fellows. It
is possible that it is not a legislator's attitudes as such,
but his knowledge about how they fit in the context of the
legislative body that is important for political success in
the legislature. Such success should be measured in a
number of ways--a subjective measure, an objective measure,
such as the number of bills sponsored and passed, and
another objective measure, such as the assessment of his
success by his fellow legislators. The present research
did not purport to be a study of the legislature, but the
research indicated in Chapter III can be extended and
intensified.
Finally, the ultimate purpose of this research was to
ascertain how the role-taking variables fit into the
political realm. Of what use is it to tie projection,
role-taking accuracy, and assumed similarity to the po
litical context? From the point of view of the philosophy
of research, such research fills the void often left between
political acts and such background variables as education,
income, and profession. It provides a more subtle picture
for the explanation of political activities. It is very
well, and it is an important first step, to be able to
187
predict, for instance, a person's voting behavior on the
basis of the way in which he voted in the last few elec
tions. It is important to discover the association between
class and political participation. But these pictures of
political reality need to be elaborated for modern political
cultures. It is possible that some of the most ardent ad
vocates of change are not the deprived, but, as the research
here revealed are found among political profiles accentuating
equality and humanitarianism. And the political activity
stimulated by such ideals is further modified by disowning
projection and role-taking accuracy patterns.
Ultimately, of course, the test of the study of role
taking variables and their effect on political subgroups
rests on the question that all new research directions
must heed: Does the study of the role-taking variables in
the political context enable the social scientist to pre
dict behavior more accurately? Although it was the intent
of this study to go beyond the exploratory stage, it cannot
be claimed that this question has received an appropriate
answer. Nevertheless, a foundation for further study and
replication has been laid. -
APPENDIX I
COPY OF COVER LETTER SENT TO LEGISLATORS WITH FIRSTCONSENSUS QUESTIONNAIRE
COPY OF COVER LETTER SENT TO LEGISLATORS WITH DUPLICATECONSENSUS QUESTIONNAIRE IN SECOND ATTEMPT TO OBTAINRESPONSES
COPY OF LEGISLATORS' CONSENSUS QUESTIONNAIRE, MINUS THESEMANTIC DIFFERENTIAL SECTION
UNIVERSITY OF' HAWAII. HONOLULU. HAWAII 96822
OI""'''TMIHT 0,. "OLITICAL .CIIHCI
Dear
Thio is a requeot for your cooperation 1n e research ~roject.
I am a student and an inatructor at the University of Hawaii and amworking toward my doctorate in ~olitical science. The firat phaeeof my dissertation project ~p.kes it neco30ary for me to discoverhow legioletors in Haweii view politics. In order to eccomolishtr.i9, : have ~rep&red the enclosed questionnaire. If you will com?lete the questionnaire for me, a taak th~t will take about 45 minuteD, y~ur coo~eration will be greatly a?preciated.
When you have completed the quest:onnaire, plesse enclose it inthe stamocd, self-addressed envelooe aoc return it to me with thisletter. All returns will be handl~d confidentially by me, and 011individual queationnaireo will remain anonymous. The questionnaireswill be used to obtain figures thbt pertein to the leeislature ofHawaii ss a wholo. The return of this letter with the questionnaire is requested so thut I can keep 8 record of the questionnairee thp.t have been returned, after which the letter will beseparated from the q~estionnaire.
If :'ou hflve inquiries rocording this oroject, you ere welcometo contact me at ~J home (phones 511-776) or at my office at theUniversity of Hawaii (e:<teneions t16T~). When the disgertotiorJio com,lete end accepted, it will be filed at the Univer~ity libraryand in the Dep~rtrnent of Politicol Science. My corr.~ittee chair-mon for the disaertation ie Profe3aor Marshall N. Goldstein.
I am awere that the job of legislator io time-consuminC. Th~re
fore, I thank you 1n advance for :/our kind a~n1at8nce in this 'Jroject.
Sincerely yours,
lli. llR rd D. Ke i III
UNIVERSITY OF HAWAIIDrp:ut/lll'llt of Politi(~al Hdl'llcP
28 Seotember lc67
Dear
In the field of political science, there is 8 new effort to study politicsby meane of empirical evidence, a major source of which io ~ot6rial gatheredthrough questionnaires. This new direction in the studv of Dolitics i8 calledthe -behavioral approach.· Among tho universitie8 eneaged in research of thisnature, tr.e University of Hawaii is among the foremoet. In fect, 80 6 etudontworking toward my Ph.D., 8 reoearch aoaiatent, and a former inntructor in theCollege of General Studies here, I foresee the ti~e when tho University of Hawaiiwill be among the best in the nation in the field of political ocience. Hawaiican well be oroud of the growth of ito univerolty, a growth thet has been trulyphenomenal even in the five years that 1 have been a reoident here.
My own research is 0100 an attempt to ta;> certain sourceD of empiricalevidence. You may recoll a questionnaire thet I mailed to you recently(August 17, 1067). This questionnaire wae an attemot to survey certain attitudesheld by the legiolature of Hawaii AD a whole. If you decided against comoleting the questionnaire at thnt time, may I request that you reconsider yourdecision! The aocuracy of such a study depends on ae complete a sample aspossible. If you have miolald the former questionnaire, I have enclooed aduplicate.
Naturally, in research of this nature, the resoondent is assured that hisreply will be entirely ccnfidential. 1 em not tntereoted in individual reo11esao such, but my research will concern only tho legislature as a ~hole. If youhave any question concerning the nature of this research, please feel free tocontact me at my home (telephonel 511-776).
If you have already returned a queotionnaire, plenee ignore thio letter andaccept my aopreciation fo': your cooperation in thiB project. The percente.geof returns of thene questionnaires hae been very impressive.
Sincerely yo~rD,
W11le rd D. Ke im
2;#) Call1plls RrJlUI· Honolulu, Hawaii 00822
PART ONE: DIRS"CT::::2~:.3. The followine; itet:s ~resent sit'.l3tions that might ariseduring a political career. ?or each of these hypothetical situations,there are given a nu~b~r of possible actions thut ~ight be t8ke~.
Please circle t~e letter for that action which 6?peals to you as thebest choice under the circuostances.
1. I have cade a cac?aign ?rocise to su?~ort e certain bill. This has probablywon quite a few votes for ~e. V~en the bill coces before the legislature,my political party is very cuch opposed to its passage. I will .••
a. keep quiet cnd vote with cy party.b. speak DOre or less in favor of the bill, but vote for a party a~end~ent
that cripples the bill.c. stay away fro~ the legislature on the day of the vote.d. tell the ~e=bers of cy party about cry ca~paign coc~itoent end then vote
in fevor of the bill.e. work vigorously within ~: party to change their stand on the bill and
then, if they decide to vote against it, go along with thee.
2. I have won my first election to the P.a~aii legislature, and the session hasjust opened. A bill is u? for discussion. I 9~ by no means an expert onthe bill, but I have e little ~nowledge and support it. I will
I-'\.0I-'
?rivate dinner ?arty, I find that I have been seated n€xt to a longpolitical ene~y with who~ I have been feucir.~ rece~tly. r will •
gain the floor end deliver a s~eech favoring the bill.offer en amendcent inte~ded to strengthen the bill.infcrcally sug~est to oth~r supporters ttet I might not vote for thebill so the: ca:r be encouraged to offer me socething in return for mysupport.keep quiet and vote for the bill.
B.b.c.
d.
3. At atice
a. qUietly recu~st the hostess to seet ce elsewhere.
page 2
6. A bill of r::ine is before o· COt:i~littee of "r;hicb I ao [:. ';:-;r·:ber. I can foreseethat it ~ill 9robably luck several votes of bein5 8o;roved. 3efore the deyof the vote, a m3~ber of the 09?os:ti0n on the co~~ittee sugge8ts that Iacce?t en aoend~~~t to th~ bill. This e~end~ent will cake the bill lesseffective but will probably allow it to 98SS. I ~i11 •••
a. turn do~n his su~gestion.
b. discuss his amend cent ~ith ot~er interested grou?s end ecc€?t it if mostof the ot~ers go along.
c. accuse ~ie of tr:ring to ~ake t~e bill ineffective.d. acce?t his offer.e. find out what he eight in return for this favor; if I agree, accept his
a~end~ent.
7. !'!y office is being picketed by a voter'!) grou? oec8u:Je I intend to voteagainst a bill they faver. I hapgen to ~now that they heve never given ~e
any support, and so=e of their signs are ?retty in3ulting. I will •••
a.b.c.d.
e.
8. III
a.b.
c.
ignore thee.call the police to see whether they can be legally reooved.call a s~okescen fro~ t~e group into oy offic~ end talk over the ~atter.
tell t~ec they had better cease ?icketin~ or r'11 never sup?ort any oftheir bills in the future.tell t~ec t~ey can ?icket, but they had better re~ove the ~re insultingsigns.
have been in an euto~obi1e accident end a= still wearing band~Jes. Althoughcan get cround with no dan~er, I ee still in sooe pain. In this condition,receive en invitation to s?ea~ before a club. I will
decline the invitation ~ecause of ~y conditi~n.
suggest that I could ~re?ere s 9?~ech for the club to be given by oneof ~y assistants.~rovide the ns=es of sev~ral cth~r s?es~ers who =ight be able to ?resent
~
...0N
......,. -._; .... + _.t:'_".; .n,._,
page 3
Strongly Unde- Stronglyi1grec Agree cidcd Disagr<;e Disagree
3· There is nothin6 ~rong ~ith
conformity g~nerelly• / I i I L/ I I L/. 4. Elections ere too heatedfor my taste. LI L/ LI / / LJ5· It is quite natural to g~t
o~?osing o?inions fro~ the su~e setof facts. 1/ LI LJ LI Li6. You cannot really be sure
whether en opiniQn is true or notunless ?eople are free to argue
LJagainst it. LJ / I / I 17
7. A politician should behighlyeducat-;d. /7 Cl 1/ L/ / /
p. The legisletcr invents-.solutions for ?olitical proble~s. /f r7 / / 17 /-.
---- L-./
~. I feel unco~fortable beinglaughed at. . 1 I LJ C7 Cl" / /
10. It's e good thing that peo?lecan recove their l~gislators even
I I L/ LJ CJ L/for the wrong reasons. ~
\0UJ
11. A ~an ~ho won't co=?ro=iseisn't a good citizen. / I / ! ~ Cl CJL12. It's better to settle issues
page 4
Strongl?Agree A;ree
Unde-cidod Disagree
22. PrivEte ond infor~al discussion so~etioes clears uo difficult
.proble=s that ~ublic debate wouldnot solve. I I / I I / LJ
Stron~l~r
Disagree
LI23. The essence of good politicsis t€e~work. I I 17 LI / / / I24. Il~ al~ost nevsr i~oolite topeo~l€. . I I 17 Cl 11 1125. Socati~€s I choose friends inorder to incr~E3e cy influence. I I 17 17 17 11
26. S06cial interests are usuallyagainst the ?~blic interest. / I 17 f7 /l CJ27. In a better age, ~e will beable to do witr.out ~oliticians. I I f/ f/ 11 11
1/
" I
1/
L:JII
I ,I
C7LJ
LI
1/8rE;u:::ents.I enjoy ~olitical
2g. A cerson who hides b~hind th~
laws ~h~n he is c~e3tion~d sbout hisactivities doesn1t deserve ouchconsideration.
29.
30. A good ~ey to vot~ when indoubt is to :'0110'" the ?art~· hb.~l.11 1/ I I L:J fI
~
\.0~
31. If all =y fri~nds ~ere u~iver
sity graduates, I houlc hide the
42. The general ~ublic is justnot qualified to vote on ~ost oftoday's coo?lex issues.
43. I'e glad to live in t~eseexciting times.
Stron~ly
Agreo
LJ
L/
Agree
oLJ
Undecided
L/
/7
~~se 5
Di se.gree
II
CJ
5~rongly
Disagree
CJ
/7
44. I enjoy being in larg~ groupsof ~eople. / 7
45. If I really wanted a law passed,it would be cowarcly to accept 8 _~8tered-down version of it. L--/
46. The ~olitician is constantlylooking f~r oublicit:'. • /7
47. It's wrong for oeo?l~ to putoressure on le~i9letors to votetheir way. - / 745. Those ~eo?le who hate our wayof life should have a chance tos?eak and be heard. /7
4~. In politics, it's every ~an
for himself. /7
50. Politicians are usuallydbitterly 6ttacking~ or rtreactingViolently. ~
"l_ All mp-\/"l" ~nto,.""t" in nil.,.
C7
L:J
/7
LJ
Cl
CJ
CJ
II
/ /
Cl
1/
1/
1/
f7
CJ
L:J
r7
r7
1/
f7
;=1
CJ
Cl
r7
r7
r7
I /
1/
~
\.0U1
APPENDIX II
COpy OF COVER LETTER SENT WITH ROLE-TAKING QUESTIONNAIRETO GROUP A
COPY OF COVER LETTER SENT WITH ROLE-TAKING QUESTIONNAIRETO GROUP B
COPY OF ROLE-TAKING QUESTIONNAIRE
1407 Kewalo Street, Apt. 9Honolulu, Hawaii 968229 May 1968
Dear Sir or Madam:
The purpose of the enclosed questionnaire is to determine the politicaloutlooks of persons belonging to certain voluntary organizations in Honolulu.
The results from returned questionnaires will be seen and analyzed onlyby l~self, and they will be presented in a general f~shion as part of aPh.D. dissertation at the University of Hawaii.
In research of this nature, it is necessary that I obtain as manyreturned questionnaires as possible, or else the results will not bereliable. Therefore, I would appreciate your cooperation in thisresearch project.
Thank you for the time spent in answering this questionnaire.
Willard D. Keirn
1407 Kewalo Street, Apt. 9Honolulu, Hawaii 9682228 Hay 1968
Dear Sir or ~mdam:
The purpose of the enclosed questionnaire is to determine the politicaloutlooks of persons belonging to voluntary organizations in Honolulu.
The results from returned questionnaires will be seen and analyzed onlyby myself, and they will be presented in a general fashion as part of aPh.D. dissertation at the University of Hawaii. The opinions of individualswill not be attributed to the organizations to which the respondents belong.In tho text of the dissertation, organizations will be referred to by
letters (A,B,C) or symbols only.
In research of this nature, it is necessary that I obtain as manyreturned que3tionnaires as possible, or else the results will not bereliable. Therefore, I would appreciate your cooperation in this researchproject.
Thank you for the time spent in answering this questionnaire.
Yours sincerely, )
CA-J JJ'(JS~0k~.--Willard D. Keim
Z~~h of the brief questions in thispossible answers. Host of these answersIt is your opinions I am interested in.for the answer that comes closest to how
(This number identifies the questionnaire, notthe respondent.)
INTRODUCTION
This questionnaire has been desi£ncd to determine public opinionon a number of political issues and attitud8S. The results ~lill beanalyzed and used in a doctoral dissertation at the University of lIawaii.
.All returncd qucstionnaircs will be kept in strict confidencc. It is
not nccessary to put your naMe on this questionnaire.
DIRECTlmlS
questionnaire has a number ofarc neithcr "right" nor "wrong."Please place an "X" in the spaceyou feel abou~ the question.
I am interested in your first impressions, so don't bother to spendmuch time on your anS\olers. Try to answer all the ques tions. There areabout 'a half dozen 'qucstions on political knowledge. Host of us can'tanswer all of them, so if you don't know, just guess and go on to thenext questions.
The questionnaire usually takes from 20 to 45 minutes to complete.
I greatly appreciate your cooperation in this research project.
W: KeirnPhone: 511-776
Page 1
PART I. The following questions are about your involvement inpolitics.
1. Have you ever written a letter to the editor of a newspaper or magazinegiving your political opinions?
__________Yes, very often.Yes, a few times.------______Once.
______No, never.
2. Have you ever ~~itten to any public officials giving them your opinionabout something that should be done?
__________Yes, often.Yes, a few tioes.
------Once.________~No, never.
3. Do you associate with any political party officials or people who docampaign work in a political party?
Yes, often.-----~Yes, sometimes.------Seldon.------No, never.------
4. How often do you meet an discuss coonunity affairs with community leaders?
Often.--------Sooetimes.----------Seldom.-----------_____N-cver.
5. How often do you c~ct nnd discuss politic~l affairs with city orcounty officials?
_1* ...... __
Noo
Page 2
10. How much interest would you say you have in political affairs?
A lot.----~_______~More than most people.________...-.;About average.________Less than most people.
Almost none.------..;11. Row much political influence would you like to have?
_____..-:A lot more.More than I have now.----------'About as much as I have now.
------~_______~Don't really care.
12. Row much political influence would you say the ___has?--_._----------
A lot.------~__________More than most other political groups._______~About the same as most other political groups.
Less than most other political groups.------Almost none.------13. Political change may mean either going back to the !lgood old days!l
or moving more rapidly !linto the future." Regardless of your definitionof change, what amount of change would you like to see in the presentpolitical situation in the United States? Suppose the bottom of thescale below (0) indicates no change at all, while the top (10)is the greatest amount of political change you can imagine. Place an"X" at the amount of change you desire:
.,. _~9_.J.. -_.~-_. --j---.a. _
7.._---- ..
NoI-'
PLlge 3
17. Have you ever considered the possibility of running for public officeyourself?
Yes.----~
No.-----Have you been active in any political issue in the past year?
Yes.------:No.
19. Hhat issue (5) was this?
20. HOy1 many Senators are there in the Hawaiian State Legislature?
21. Does the Lieutenant Governor of Hawaii h~ve to be of the same politicalparty as the Governor?
Yes.------;No.-----
22. \~at is the term of office for members of the State House of Representatives?
23. What are the names of Hawaii's representatives in the U.S. Congress(House end Senate)?
NoN
Page 4
PART II. This section contains a number of statements aboutpolitics, people, or yourself. Please place an "X"in the box that comes closest to your agreement ordisagreement. Please don't omit any of the statements,even if your agreement or disagreement is only slight.
Agree DisagreeStrongly A~ree Disagree Strongly
1. I usually maintain my ow~ opinionseven though many other people have adifferent point of view. / /
2. A civil service exam would be a betterway than an election campaign forchoosing our legislators. 1 1
3. The CIA is a very necess3ry agency __of our national gr"V'~'.·:lment. _/ /
4. When I'm in an argument, I usually __get my way. _1 /
5. I often become 50 wrapped up insomething I am doing that I find itdifficult to turn my attention to other __matters. _1 1
6. The businessman and the manufacturerare much more important to society __than the artist and the professor. _/ /
7. Elections are too heat£d for my taste. 1----1
1 /
1 1
1 1
1 I
1 1
/ 1
I I
1 1
1 1
1 1
1 1
1 /
/ /
1 I
1 /
1 I·
1 /
1 1
1 1
/ /
1 1
NoVJ
8. A good rule to follow is never totrust anyone compl~tely. 1 1 1 1 1 1 1 1
Page 5
Agree DisagreeStrongly Agree Disagree Strongly
19. I am often the last one to give uptrying to do a thing. / /
20. A man ~ho does not believe in somegr~at cause does not have a meaninsfullife. / /
21. The government in l~ashington ought tosee to it that everyone who wants a job __can find work. _/ /
22. In a social conversation. I frequentlyhave a definite idea and try to convi~ce
others. / /
23. You cannot really be sure whether anopinion is true or not unl~ss people are __free to argue against it. _/ /
/ /
/ /
/ /
/ /
/ /
/ /
/ /
I /
/ /
/ /
/ /
/ /
/ /
/ / .
/ /
24. I feel uncomfortable being laughedat. / / / / / / / /
25. Those who oppose the draft in theU:lited States should be prosecuted to thelinit of the law. / /
26. People who can't get others to likethem, don't understand how to get alongwith others. / /
27. If you start trying to change things __very much, you usually make them worse. _/ /
28. What the youth needs is strict
/ /
/ /
/ /
/ j
/ /
/ /
/ /
/ /
/ /
No~
Page 6
Agree DisagreeStrongly A~r~e Disagree Strongly
38. Most of our politicians in theco~unity are probably core interested inget".ing known than in serving the needsof ;heir constituents. 1 1
39. It is only human nature to bereluctant about cooperating with others. 1 I
40. ~~·life has generally worked out the __way I wanted it to. _1 1
41. Most people who don't get ahead just __don't have enough will-power. _1 1
1 I
1 /
1 /
1 1
1 1
1 1
1 1
1 1
/ /
/ I
I /
/ /'
42. One of the important lessons of lifeis that people put their own interestsfirst.
43. People like ~e don't have any sayabout what the govcrncent does.
44. It is self-defeating to eobarrassa political opponent needlessly.
45. People ought to cake clear tolegislators how they want the~ to vote.
46. ~ost of those who disagree with mypolitical views are sicply uninfo~ed.
1 1
1 1
1 1
I
1 1
1 1
1 1
1 /
/ 1
/ I
1 /
/ /
/ /
/ 1
1 /
I /
/ I
/ /
/ /
I I
/ / NoVl
47. Even if sooe oinority groups are __treated badly, it is no business of Dine. _1 1
----,- -- .....
1 / / / I . /
Page 7
Agree DisagreeStrongly Agre8 DisQgree Strongly
58. I dislike chancing oy plans in themidst of an undertaking. / / / / / / / /
59. People can be divided into two distinctclasses: the weak and the strong. / /
60. The findings of science may so~~dayshow that many of our most cherished beliefsare wrong. / /
61. I'm almost nev~r impolite to p~ople. / /
/ /
/ /
/ /
/ /
I /
/ /
/ /
/ /
I /
62. Special interests are usuallyagainst the public interest. / / / / / / / /
63. ~fucn a person has a proble~ or worry,it is best for hin not to think about it,but to keep busy with core cheerful things./ /
64. On the whole, I usually like to beby myself rather than with other people. / /
65. The most ~portant things to ~e aremy duties to my fellcw oan and oy job. /----/
66. In a better age, we will be able todo without politicians. / /
67. A person who hides behind the lawswhen he is questioned about his ~ctivities
doesn't deserve much consideration. / I
68. In order to get noninated, cost..... .... _- r ,':_.1 __ ' _.&:~.f,...n h"':":l'1'.' t-n
/ 7
/ /
/ /
/ /
I I
/ /
/ /
1 /
/ j
I I
/ /
/ /
/ /
/---r
I /
No0'
•
Page 8
Agr~e DisagreeStrongly Af.rce Disagree Strongly-
80. I don't mina 2 pOlitician's m~thods
if he get's the right things done. f 1
81. It's good politics to praise youropponents when you can. 1 1
82. In a discussion I often find itnecessary to r~peat nyself several timesto make sure I am being understood. 1 1
83. It is hard for oe to find anythingto talk about when I oect a n~w person. 1 1
84. On a numb~r of local d2cisions, Ihave exerted some influence. 1 1
85. Most politicians are in politicsto be of public service. 1 1
86. I believe that pronptn~ss is a ver.yimportant personality ch~ractcristic. / 1
87. Occupation by a foreign power isbetter than war. 1 1
8S. If you trust people and l~t themknow it, they will very seldom disappoint __you. _1 1
1 /
1 1
1 1
1 1
1 1
/ 1
1 /
/ /
/ /
1 1
1 1
1 1
I I
1 1
I-=-,I
I j
j I
I 1
1 1
1 1
1 1
1 1 •
1 I
1 1
1 1
1 I
I INo'-J
89. I like to be able to plan a setroutine for oy daily wark.
L: __ .. \...._
1 1 I I I -' I. I
Page 9
Agre~ Dis~gree
Strongly Agree Disagree Strongly
100. If w~ could only uncover thee, wewculd find th~re are quite a f~w Co~~u-
nip'.s in th~ national governr.lcnt. I I
101. I don't think public officials caremuch about what people like m~ think. I I
102. It's wrong for people to put pres-sure on legislators to vote their way. I I
103. Those people who hate our way of lifeshould have a chance to speak and be heard.1 I
104. If a local problee that interestedma arose, I would try to influence the .decision. / I
105. In politics, it's every nan forhimself. I I
106. Politicians are usually "bitterlyattacking" or "reacting violently." I I
1 I
1 1
1 1
1 1
1 1
1 "
1 _I
I /
1 1
1 1
I 1
1 1
1 1
1 1
/ /
1 1
I 1
/ I·
1 1
I 1
1 I
107. If it were possible, societyreally ~ould be better off to abolishindustrialization and return to smallto~~s and sinEle-family fares.
108. All ~jor interests in our nationare equally worthy of consideration.
109. The rich deserve to hav~ what they~- '---------- .t... ......... .- •• ..."" .. ., 'h.~,,_"') M1"'\'r";:) ~hi'it.v
I I
1 1
1- 1
1 1
/ 1
1 1
1 1
/ 1
NoO:l
Page 10
PART III. In this section, assuoe for the moment that you are a typicellegislator of the State of Hawaii, a member of the StateSenate or House of Repres~ntatives. You should disregardyour own fe~lin8s, t~npor~rily, and answer these questionsas you assume a typical legislator would. The point of thesequestions is to see whether citizens understand how thetypical legislator sees things.
How do you think a legislator would answer the followingquestions if he were giv~n a questionnaire like this one?
Agree Disagrea
1. A civil service exao would be a b~tter way than anelection campaign for choosing our legislators.
2. Elections are too heated for ey taste.
/ /
/ /
1 /
/ /
3. It is quite natural to get opposing opinions from thesame set of facts. / / I /
4. You cannot really ,be sure whether an opinion is true ornot unless people are free to argue against it. /----/
5. I feel uncomfortable being laughed at.
6. A man who won't compromise isn't a good citizen.
/ /
j /
/ /
/ /
/ /
7. It's better to settle issues somehow than to get theesettled absolutely correctly. / / / /
8. It is self-defeating to embarrass a political opponent needlessly.
9. People ought to make clear to legislators how theywant them to vote.
/--,
/ /
/ /
/ /
No\.0
Page 11
Agree Disagree
31. All major interests in our nation are equally worthyof consideration. / /
22. I like to be able to plnn n sct routine for my dailywork. 1 1
26. If I really wanted a law passed, it would be cowardlyto accept a watered-do.vu version of it. 1 1
30. Politicians arc usually "bitterly attacklng ll orlireacting violently." 1 /
1 /
1 1
1 /
/ /
/ 1
1 1
1 1
/ /
/ /
1 /
N
/ /t-->0
/ 1
. .
1 1
1 1
1 1
/ /
1 /
23. All elections ought to be nonpartisan.
24. I get along with persons from all levels of society.
25. l~en I get home from work, I like to forget my job and __relax. _1 /
27. It's wrong for people to put pressure on legislators __to vote their way. _1 1
28. Those people who hate our way of life should have a __chance to spc3k and be heard. ~
tancc.
29. In politics, it's every man for hioself.
~b Th~ T~rnTn~ of le~islators are too little known to the
32. I'd rather stand off and look at events froe a dis-
33. The oajor aio of government is efficiency.
Page 12
2. I have won ny first election to the Hawaii legislature, and the session hasjust opened. A bill is up for discussion. I aD by no neans an expert on thebill, but I have a little knowledge and support it. I will ••
a. gain the floor and deliver a speech favoring the bill.b. offer an aDendment intended to strengthen the bill.c. info~ally suggest to other supporters that I might not vote for
the bill so they ~y be encouraged to offer me something in returnfor ny support.
d. keep quiet and vote for the bill.
3. At a private dinner party, I find that I have been seated next to a longtime political enemy with whom I have been feuding recently. I will ••.•
a. quietly request the hostess to seat me elsewhere.b. sit next to my political enemy and be polite but cold.c. sit elsewhere "by mistake. 1I
d. sit next to my political enemy and use the occasion to get ongood terms with him.
e. sit next to oy political enemy and use the occasion to embarrasshio.
4. A letter comes into my office written in pencil and in very bad graornar.It requests information that will take some ti~e to collect. I will ••••
a. have my secretary return a form letter directing the writer toanother source.
b. have ny secretary obtain the infornation and write a full reply.c. ignore the letter, unless I happen to knotl the person who wrote it.d. put the letter aside; if more come in on the same subject, it
will be worth the time to send out the information all at once.e. forward the letter to the proper government bureau and ask them
to handle it.
5. Mr. Jones is a lobbyist with whom I am friendly ::md tvho has helped me a few
t
""
Nt-'t-'
Page 13
THANK YOU FOR ACTING OUT THE PART OF A TYPICAL LEGISLATOR. Not-l JUSTA FEW QUESTIONS j\BOUT YOURSELF.
1. Hale.Female.----"
. 2. Please indicate the category of education that applies to you.
______Less than high school.____~High school._______College, but not a graduate._______College graduate.
3. How many years have you lived in Hawaii?
4. What is your r~gu1ar occupation? (Please be specific, as insurancesalesman, engin~er, housewife, and so forth.)
5. Haw many brother and sisters were (are) there in your family?
6. Were (are) you the oldest?
Yes.----No.----7. In what age category do you belong?
______20-30 (or below 20)______31-40
t..1-r:;n
Nto-'N
APPENDIX III
VARIABLE AND ATTRIBUTE DEFINITIONS
The following definitions explain the variables andattributes that have been obtained from coding the questionnaires. These definitions cover all the usages in the text.Abbreviations are used in tables and figures and are givenin capitals following the longer variable name.
1. Held office in political organization (HELOFF).Question 1.7.
2. Acquainted with legislator (ACQLEG). Question 1.8.
3. Possibility of running for office (POSOFF).Question 3.17.
4. Issue active (ISSACT). Question 3.18. 1 The numberof issues were not counted for this score, whichwas dichotomous.
5. Sex (SEX). Question 13.1.
6. Years lived in Hawaii (RESIDE). The first computerrun with the respondents was made with the break atthree years and below and over three years, but this
_did not divide the respondents evenly. The finaldivision was less than ten years and ten years ormore.
7. Eldest or only child (ELDONL). This attribute wastaken from Quest~ons 13.5-6.
8. Empathic job (EMPJOB). The occupation question, 13.4,prov~ded no distinctions among the respondents whenit was scored to provide a measure of class distinction.Nearly all the respondents were either professionals,college students, or housewives (often with highereducation). Therefore, the occupations were categorizedby the researcher on the basis of his intuitive notion
1By an error, this question was not numbered on the
questionnaire, but this did not appear to affect theanswers of respondents.
214
as to whether they were "socializing" professions orjobs that dealt in the main with things or ideas. Theformer were called empathic professions. The following list shows the distinctions that were made.Some of the choices may not be happy ones, and thisvariable ought to be experimented with:
Empathic occupations = 1
SecretaryAdministrationNurseAttorneySalesman, customer representative,
representative of a companyPsychiatristSocial workPhysician
Nonempathic occupations = 0
UnemployedDraftsmanEditorStudentProfessorLibrarianInvestment officerTeacherHousewifeEngineerArchitectPharmacistGovernment employeePublisherUnion office workerAccountantFinance company managerBusinessmanAdvertising
9. Ego strength (EGOSTG).
4. When I'm in an argument, I usuallyget my own way.
22. In a social conversation, Ifrequently have a definite ideaand try to convince others.
32 agrees
47
110.
74.
116.
I'd rather stand off and look atevents from a distance. (Reflected)2
I enjoy political arguments.
I'm the kind of person who getshis share of good luck.
53
57
60
215
CR (Coefficient of Reproducibility) = .90MMR (Minimum marginal reproducibility) = .73
10. Radicalism, left (RADICL).
69. I admire Fidel Castro.
10. The United States ought to withdrawimmediately from Vietnam.
20 agrees
31
3. The CIA is a very necessary agencyof our national government. (Reflecte~ 33
90. Socialism might be good for theUnited States. 34
36. A guaranteed annual income for everyfamily in the United States would bea worthwhile experiment. 42
25. Those who oppose the draft in theUnited States should be prosecutedto the limit of the law. (Reflected) 54
100. If we could only uncover them, wewould find there are quite a fewCommunists in the national govern-ment. (Reflected) 55
107. If it were possible, society reallywould be better off to abolish industrialization and return to smalltowns and single-family farms.(Reflected) 61
CR = .92MMR = .67
2That is, the reverse of the ~uestion was enteredinto the scale, so that "disagrees' were actually talleyedas "agrees"; in other words, a "disagree" indicated "egostrength."
216
11. Flexibility (FLEXBL).
86. I believe that promptness is a veryimportant personality characteristic.(Reflected) 20 agrees
89. I like to be able to plan a setroutine for my daily work.(Reflected) 36
31. I usually check more than once tobe sure that I have locked a door,put out the light, or something ofthat sort. (Reflected) 46
76. Sometimes I support principles Idon't entirely agree with. 47
73. There is usually only one best wayto solve most problems. (Reflected) 60
CR = .89MMR = .68
12. Faith-in-people (FAITHP).
49. Very few people are dishonest. 33 agrees
85. Most politicians are in politicsto be of public service. 44
88. If you trust people and let themknow it, they will very seldomdisappoint you. 48
17. Even though people are not alwaysable to help you, most of them meanto be helpful. 58
77. The world just would not work withouttrust. 61
CR = .89MMR = .71
13. Equality (EQUAL).
55. There should be no restriction exceptone's own ability upon the amount ofmoney one may honestly acquire.(Reflected) 22 agrees
217
13. Equality (EQUAL). (continued)
21. The government in Washington oughtto see to it that everyone whowants a job can find work
37. The trouble with democracy is thatmost people don't really know whatis best for them. (Reflected)
109. The rich deserve to have what theydo because they usually have moreability than poor people.(Reflected)
33. If a person is poor in thiscountry it is usually his ownfault. (Reflected)
13. Wealth in the United States isdistributed about as equally asit should be. (Reflected)
CR = .90MMR = .75
14. Sociability (SOCIAL).
93. No matter how hard you try, somepeople just don't like you.
- (Reflected)
46 agrees
48
53
59
62
10 agrees
79. In a group I usually take theresponsibility for getting peopleintroduced. 31
64. On the whole, I usually like to beby myself rather than with otherpeople. (Reflec ted) 54
83. It is hard for me to find anythingto talk about when I meet a newperson. (Reflected) 58
CR = .94MMR = .75
15. Innovativeness (INNOVT).
36. A guaranteed annual income for everyfamily in the United States would bea worthwhile experiment. 42 agrees
15. Innovativeness (INNOVT). (continued)
29. I'd want to know that somethingwould really work before I'd bewilling to take a chance on it.(Reflected)
27. If you start trying to change thingsvery much, you usually make themworse. (Reflected)
32. It is better to stick by what youhave than to be trying new thingsyou really don't know about.(Reflected)
CR = .94MMR = .80
16. Humanitarianism (HUMANE).
87. Occupation by a foreign power isbetter than war.
56. One way to deter men from crime isto make them suffer for theircrime. (Reflected)
71. Compulsory military training inpeacetime is essential for thesurvival of this country.(Reflected)
70. The death penalty is barbaric, andit should be abolished where itstill exists.
CR = .90MMR = .65
218
57 agrees
61
63
16 agrees
36
45
46
17. Political cynicism (POLCYN).
80. I don't mind a politician's methodsif he gets the right things done 10 agrees
85. Most politicians are in politicsto be of public service. (Reflected) 26
30. Money is the most important influenceon public affairs. 28
17. Political cynicism (POLCYN). (continued)
72. The people who really run thiscountry are not even known to thevoters.
-6~ In order to get nominated, most
candidates for political officehave to make basic compromises andundesirable commitments.
CR = .89MMR = .64
18. Political efficacy (POLEFF).
84. On a number of local decisions, Ihave exerted some influence.
9. Sometimes politics and governmentseem so com~licated that a personlike me can t really understandwhat's going on. (Reflected)
101. I don't think public officials caremuch about what people like methink. (Reflected)
11. Voting is the only way people likeme can have any say about how thegovernment runs things. (Reflected)
CR = .91MMR = .72
219
30 agrees
39
35 agrees
52
55
59
19.
20.
8.7.
Participation index ~PARTIC). Questions 1.1-1.6 werescored zero, "never,' to three, "Yes, very often," andthe sum formed a participation index. Scores of 8+3were "high" (=1), those below were "low" (=0) for thedichotomized variable.
Political influencet self (INFSEL). Question 1.9 wasscored from one, "a most none," to five, "a lot." Thedichotomous variable was divided 4+ "high" and below,"low."
3The 8+ means response "8" and responses larger thanThe "low" would then be responses ranging from 0 through
21.
22.
23.
24.
25.
220
Interest in political affairs (POLINT). Question 2.10was scored from one, "almost none," to five, "a lot."The dichotomy was formed by dividing the respondentsat 5+, "high," and below 5, "low."
Political influence one would like (ASPIRE). Question2.11 was scored from one, "don't really care," to four,"a lot more." The dichotomy was formed by dividingrespondents at 3+, "high," and below 3, "low."
Influence of the Hroup (INFGRP~. Question 2.12 wasscored from one, almost none, ' to five, "a lot."For the dichotomous variable, 4+ was "high," below4 was "low."
Advocacy of chan~ (CHANGE). Question 2.13 was scoredtrom zero, "no change," to ten, "the greatest amount ofpolitical change you can imagine." The dichotomousvariable was divided at the integer nearest the mean ofthe combined groups. Scores of 8+ became "high," scoresbelow 8 became "low." Question 2.14 was added in orderto provide a check on the ladder value. Since it wasfound to correlate .73 with Question 2.13, the laddervalue was regarded as having accurately tapped thedegree of change indicated by the verbal question,and the latter was dropped in the further analysis asredundant.
Success-you would have (SUCSEL). Question 2.15 wasscored trom one, "almost no success," to four, "agood deal of success." The dichotomy was formed at3+, "high," below 3, "low."
26. Success group would have (SUCGRP). Question 2.16was scored trom one, "almost no success," to four,"a good deal of success." The dichotomy was formedat 3+, "high," and below 3, "low."
27. Political knowledge index (POLKNO). Questions 3.203.26 were scored one for each completely correctanswer and summed to give a political knowledgescore. The dichotomous variable was formed at 6+,"high," below 6, "low."
28. Behavior-in-situation score (BSCORE). Questions11.1-12.6 were scored correct if the respondentcorrectly indicated the same behavior in the situationthat had been determined by the consensus of thelegislators. The dichotomous variable was formed at4+, "high," and below 4, "low."
29.
30.
31.
32.
221
Ro1e-takin~ accurac~ (ROLEAC). Questions 10.1-11.35were summe for eac respondent according to the numberof res~onses he checked that coincided with the legislators consensus. The dichotomous variable wasformed by dividing the entire group of respondentsnear the mean: 28+, "high," below 28, "low."
Assumed simi1arita (ASSUME). Questions 10.1-11.35were also include among the scale questions, pages4-9 of the questionnaire, where they were answeredby the respondent for himself. The sum of theanswers for the scale-located questions that matchedthe answers the respondent gave as if he were thelegislator became the assumed similarity score. Thedichotomous variable was formed by dividing the entiregroup of respondents near the mean: 27+, "high,"below 27, "low."
Actual similarity (ACTUAL). The scale-located rolequestions that were answered in the same direction asthe legislators' own consensual responses were summedto give an actual similarity score. The dichotomousvariable was formed by dividing the group near themean: 28+, "high," below 28, "low."
Disownin -assimilative ro·ection score (PROJEC).T e ro e-ta ~ng ~tems t at were ~ncorrect were recordedfor each respondent. Wrong guesses in which therespondent indicated similarity between himself and thelegislator were scored as assimilative projection,while those in which the respondent indicated a difference between himself and the legislator were scoredas disowning ?rojection. A projection score wasobtained by subtracting the assimilative projectionscore from the disowning projection score and adding12 in order to provide positive scores for all individuals. The dichotomous variable was formed bydividing the scores at the integer nearest the mean:14+, "high," below 14, "low."
The following example of the scoring should makethe procedure clear:
222
Respondent X
AccuracyAssumed similarityActual similarity
231521
Incorrectguesses
Similar toLegislator
Different fromLegislator
Item 456
101719202527293334
Total:
1
11
3
111111
1
11
9
33.
34.
35.
Disowning projection 9Assimilative projection 3
Final PROJEC score: 9 - 3 = 6; 6 + 12 = 18.
tge (AGE). ~uestion 13.7. The dichotomy was formedy age 41+, 'high," below 41, "low," or young.
resent influence and desire for influencePG P. T e raw score or uest~on . , present
political influence, was subtracted from Question 2.11,desire for political influence, and if the remainderwas a positive integer, ASPGAP was "high," while anegative integer or zero was "low."
Index for authoritarianism (FINDEX). The items forauthoritarianism scaled for the pilot study, but therespondents for the final study were simply notauthoritarian enough to provide an appropriate Guttmanscale, even though CR = .89 and MMR = .77. Therefore,respondents who agreed with two or more items of eight(Questionnaire items 14, 28, 41, 50, 51, 67, 23reflected, and 25-reflected) were scored "high," those
223
agreeing with one or none of the items were scored"low" in authoritarianism.
36. Index for apolitical tendency (APOLIT). In an attemptto scale some of the items among the role questionstogether, two tentative scales were formed, APOLIT andtwo others which were called "political toughness" and"political flexibility." High CR's were obtained because of the large consensus among respondents forthese items. Therefore, two tentative scales weredropped, and the items of APOLIT (Questionnaire items66, 74-reflected, 76-reflected, and 110) were scoredfor each respondent. Any respondent with a singleagree was scored "high" for APOLIT.
BIBLIOGRAPHY
Almond, Gabriel A., The Appeals of Communism, Princeton,N. J., Princeton University Press, 1954.
and Sidney Verba, The Civic Culture, Princeton,N. J., Princeton University Press, 1963.
Anikeeff, Alexis M., "Reciprocal Empathy: Mutual Understanding among Conflict Groups," Studies in HigherEducation, Purdue University, Division of EducationalReference, No. LXXVII (August 1951), 1-48.
Astin, Helen S., "Assessment of Empathic Ability by Meansof a Situational Test," Journal of Counselin~Psychology, Vol. 14, No. 1 (January 1967), 5 -60.
Bates, Frederick L., "Position, Role, and Status: AReformulation of Concepts," Social Forces, Vol. 34,No.4 (May 1956), 313-321.
Bell, Graham B. and Harry E. Hall, Jr., "The Relationshipbetween Leadership and Empathy," The Journal ofAbnormal and Social Psychology, Vol. 49, No. 1 (January1954), 156-157.
and Rhoda Stolper, "An Attempt at Va1idati.on of---"""th"-e- Empathy test," The Journal of ApE1ied Psychology,
Vol. 39, No.6 (December 1955), 442-443.
Bender, I. E. and A. H. Hastorf, "On Measuring GeneralizedEmpathic Ability (Social Sensitivity)," The Journal ofAbnormal and Social pstcho10gy , Vol. 48, No. 4(October 1953), 503-50 .
and , "The Perception of Persons: Fore-casting Another Person's Responses on ThreePersonality Scales," The Journal of Abnormal and SocialPsychology, Vol. 45, No. 3 (July 1950), 556-561.
Berkowitz, Leonard, "The Judgmental Process in PersonalityFunctioning," Psnchologica1 Review, Vol. 67, No.2(March 1960), 13 -142.
Ber1ew, David E., "Interpersonal Sensitivity and MotiveStrength," The Journal of Abnormal and SocialPsychology, Vol. 63, No. 2 (September 1961), 390-394.
225
Bieri, James, "Cognitive Complexity--Simplicity andPredictive Behavior," The Journal of Abnormal andSocial Psychology, Vol. 51, No. 2 (November 1955),263-268.
Blalock, Hubert M., Social Statistics, New York, McGrawHill, 1960.
Blanchard, William A., "Assimilation and Contrast in Interpersonal prediction with control for the interactionof real similarity and differential accuracy," Journalof personalit~ and Social Psychology, Vol. 3, No. 5(May 1966), 5 7-573.
Bronfenbrenner, Urie, John Harding and Mary Gallwey,"The Measurement of Skill in Social Perception,"David C. McClelland, Alfred L. Baldwin, UrieBronfenbrenner, and Fred L. Strodtbeck, Talent andSociety, Princeton, N. J., D. Van Nostrand Company,1958, pp. 29-111.
Brmvn, James C., "An Experiment in Role-taking," AmericanSociological Review, Vol. 17, No.5 (October 1952),581-597.
Buchheihler, Arnold, "The Development of Ideas AboutEmpathy," Journal of Counseling Psychology, Vol. 10,No.1 (Spr1ng 1963), 61-70.
Buros, Oscar Krisen, The Fifth Mental Measurements Yearbook,Highland Park, N. J., 1959.
Cameron, Norman, The Psychology of Behavior Disorders,Boston, New York, Chicago, ballas, Atlanta, SanFrancisco, Houghton Mifflin Company, 1947.
Campbell, Donald T., "An Error in some Demonstrations ofthe Superior Social Perceptiveness of Leaders," TheJournal of Abnormal and Social Psychology, Vol. ~No. 3 (November 1955), 694-695.
Cantril, Hadley and William Buchanan, How Nations See EachOther, Urbana, Ill., University of Illinois Press,1953. 443-469.
Cattell, Raymond B., "The Dimensions of Cultural Patternsby Factorization of National Characters," The Journalof Abnormal and Social Psychology, Vol. 44, No. 4(October 1949).
_____~__, The Scientific Analysis of Personality, Baltimore,Md., Penguin Books, 1965.
226
Chamber of Commerce, Hawaii, Who's Who in Government,State of Hawaii, Honolulu, Hawaii, 1967.
Chance, June E. and Wilson Meaders, "Needs and Interpersonal Perception," Journal of Personality, Vol. 28,No.2 (June 1960), 200-209.
Chin, Robert, "The Utility of System Models and Developmental Models," Jason L. Finkle and Richard W. Gable,Political Development and Social Change, New York,John ~iley and Sons, 1966, pp. 7-18.
Cline, Victor B., and James M. Richards, Jr., "Accuracyof Interpersonal Perception--a General Trait?"Journal of Abnormal and Social Psychology, Vol. 60,No. 1 (January 1960), 1-7.
and , "Components of Accuracy of Inter-personal Perception Scores and the Clinical andStatistical Prediction Controversy," The Ps~chological
Record, Vol. 12, No.4 (October 1962), 373- 81.
Cohent
Edwin, liThe Methodology of Notcutt and Silva'sKnowledge of other People': A Critique," The Journal
of Abnormal and Social Psychology, Vol. 48, No. 1(January 1953), 155.
Comrey, Andrew L. and Edward Levonian, "A Comparison ofThree Point Coefficients in Factor Analyses of MMPItems," Educational and Ps!chological Measurement,Vol. XVllI, No. 4 (Winter 958), 739-755.
Cottrell, Leonard S., Jr., and Rosalind F. Dymond, "TheEmpathic Responses, A Neglected Field of Research,"Psychiatry, Vol. 12, No.4 (November 1949), 355-359.
Couch, Carl J., "Self-Attitudes and Degree of Agreementwith Immediate Others," The American Journal ofSociology, Vol. LXIII, No. 5 (March 1958), 491-496.
Coutu, Walter, "Role-playing vs. Role-taking: an Appeal forClarification," American Sociological Review, Vol. 16,No.2 (April 1951), 180-187.
Cowden, Richard C., "Empathy or Projection?" Journal ofClinical Psychology, Vol. XI, No.2 (April 1955),188-190.
Crockett, W. H., and Meidinger, Thomas, "Authoritarianismand Interpersonal Perception," The Journal of Abnormaland Social Psychology, Vol. 53, No. 3 (November 1956),378-380.
227
Cronbach, Lee J., "Processes Affecting Scores on 'Understanding of Others' and 'Assumed Similarity, '"Ps~chological Bulletin, Vol. 52, No.3 (May 1955),17 -193.
and Goldine C. GIeser, "Assessing Similaritybetween Profiles," The Pst:chological Bulletin, Vol. 50,No.6 (November 1953), 45 -473.
Crow, Wayman J. and Kenneth R. Hanunond, "The Generalityof Accuracy and Response Sets in InterpersonalPerception," The Journal of Abnormal and SocialPsychology, Vol. 54, No. 3 (May 1957), 384-390.
Daane, Calvin J. and Louis G. Schmidt, "Empathy andPersonality Variables," Journal of EducationalResearch, Vol. 51, No.2 (October 1957), 129-135.
Dymond, Rosalind F., ,~ Preliminary Investigation of theRelation of Insight and Empathy," Journal of ConsUltin~ Psychology, Vol. XII, No. 4 (july-August 1948),228-23 .
, "A Scale for the Measurement of Empathic Ability,"----~j-ou-rnal of Consulting Psychology, Vol. 13, No.2
(April 1949), 127-133.
, "Personality and Empathy," Journal of Consulting----~P~sy-chology, Vol. 14, No.5 (October 1950), 343-350.
, Anne S. Hughes, and Virfiinia L. Raabe, "Measurable-----=Ch~a-nges in Empathy with Age, I Journal of Consulting
Psychology, Vol. 15, No.3 (June 1952), 202-206.
Edwards, Allen L., "Four Dimensions in Political Stereotypes,"The Journal of Abnormal and Social Psychology, Vol. 35,No. 4 (October 1940), 566-572.
Festinger, Leon, "A Theory of Social Comparison Processes,"Human Relations, Vol. VII, No.2 (1954), 117-140.
Fiedler, Fred E., "The Psychological-Distance Dimension inInterpersonal Relations," Journal of Personality,Vol. 22, No.1 (September 1953), 142-150.
228
Fleishman, Edwin A., and James A. Salter, "Relationbetween the Leader's Behavior and his Empathy towardSubordinates," Journal of Industrial Psychology,Vol. 1, No.3 (September 1963), 79-84.
Foa, Uriel G., "Empathy or Behavioral Transparency?"The Journal of Abnormal and Social Psychology, Vol.56, No. 1 (January 1958), 62-66.
Friedman, C. Jack, and John W. Gladden, "ObjectiveMeasurement of Social Role Concepts via The SemanticDifferential," Ps~cholo~ical Reports, Vol. 14, No.1(February 1964), 39-24.
Gage, N. L., "Explorations in the Understanding of Others,"Educational and Psychological Measurement, Vol. 13,No. 1 (Spring, 1953), 14-26.
and Lee J. Cronbach, "Conceptual and Methodological--.....P~r~o-.blems in Interpersonal Perception," Psychological
Review, Vol. 62, No.6 (November 1955), 411-422.
, George S. Leavitt, and George C. Stone, "The-----=I-n~termediarr. Key in the Analysis of Interpersonal
Perception, ' Psychological Bulletin, Vol. 53, No.3(May 1956), 258-266.
Galtung, Johan, Theorl and Methods of Social Research,New York, Co1umb~a University Press, 1967.
Goffman, Erving, Encounters, Indianapolis, Ind., The BobbsMerrill Company, 1961.
Gornpertz, Kenneth, "The Relation of Empathy to EffectiveCorranunication," Journalism Quarterly, Vol. 37, No.4(Autumn 1960), 533-546.
229
Gullahorn, Jeanne E. and John T. Gullahorn, "The Utilityof Applying both Guttman and Factor Analysis toSurvey Data," Sociometry, Vol. 31, N(J. 2 (June 1968),213-218.
Hall, Dennis R., Research Report No. 14, The Dimensionalityof Nations Project, Honolulu, Hawaii, 1968.
Halpern, Howard W'i
"Empathy, Similarity, and SelfSatisfaction, ' Journal of Consulting Psychology,Vol. 19, No.6 (December 1955), 449-452.
Harman, Harry H., Modern Factor Analysis, Chicago andLondon, University of Chicago Press, 1967.
Hastorf, A. H. and I. E. Bender, "A Caution Respectingthe Measurement of Empathic Ability," The Journal ofAbnormal and Social Ps3cholo,t' Vol. 4/, No. 2(Supplement) (April 19 2), 5 ~-576.
, , and D. J. Weintraub, "The Influence---o""£'""":Response Patterns on the 'Refined Empathy Score, '"
The Journal of Abnormal and Social Psychology, Vol. 51.,(September 1955), 341-343.
Hawkes, Glenn R. and Robert L. Egbert, "Personal Valuesand the Empathic Response: Their Inter-relationships,"The Journal of Educational ps~chology, Vol. 45,No. 7 (November 1954), 469-47 .
Hero, Alfred 0., Americans in World Affairs, Boston,The World Peace Foundations, 1955.
Hobart, Charles W. and Nancy Fahlberg, "The Measurementof Empathy," The American Journal of Sociology,Vol. LXV, No. 5 (March 1965), 595-603.
Howard, Kenneth I., and Hermann I. Diesenhaus, "Directionof Measurement and Profile Similarity," MultivariateBehavioral Research, Vol. 2, No.2 (April 1967),225-237.
Hyman, Herbert, Survey Design and Analysis, New York,The Free Press, 1955.
230
Katz, Robert L., Empathy: Its Nature and Uses, Glencoe,Ill., The Free Press of Glencoe, 1963.
Kerlinger, Fred N., Foundations of Behavioral Research,New York, Holt, Rinehart, and Winston, 1964.
Kerr, W. A. and B. J. Speroff, The Em~athY Test, Chicago,Ill., Psychometric Affiliates, 1 54.
Lerner, Daniel, The Passing of Traditional societ~,
Glencoe, Ill., Collier-Macmillan Lim~ted, 19 8.
Leventhal, Howardt
"Cognitive Processes and InterpersonalPredictions, ' The Journal of Abnormal and SocialPsychology, Vol. 55, No. 2 (September 1957), 176-180.
Levinson, Daniel J., "Role, Personality, and SocialStructure in the Organizational Setting," The Journalof Abnormal and Social Psychology, Vol. 58, No. 2(March 1959), 170-180.
Lindesmith, Alfred R., and Anselm L. Strauss, SocialPsychology, New York, The Dryden Press, 1956.
Lindgren, Henry Clay and Jacqueline Robinson, "AnEvaluation of Dymond's Test of Insight and Empathy,"Journal of Consulting Psychology, Vol. 17, No.2(April 1953), 172-17 .
Longsbaugh, Richard, "The Structure of InterpersonalBehavior," Sociometry, Vol. 29, No.4 (December 1966),441-460.
McCall, George J. and J. L. Simmons, Identities andInteractions, New York, The Free Press, 1966.
McClelland, William A., "A Preliminary Test of RolePlaying Ability," Journal of Consulting Psychology,Vol. 15, No.2 (April 1951), 102-108.
McClosky, Herbert, "Consensus and Ideology in AmericanPolitics," The American Political Science Review,Vol. LVIII, No. 2 (June 1964), 361-382.
McConaughy, John B., "Certain Personality Factors ofState Legislators in South Carolina," The AmericanPolitical Science Review, Vol. XLIV, No. 4(December 1950), 897-903.
231
Mahoney, Stanley C. and Charles A. Auston, "The EmpathyTest and Self-Awareness of Kuder Interest Pattern,"Ps~cho1o~ica1 Reports, Vol. 4, No. 3 (September19 8), 4 2.
Marwe11, Gerald, "Problems of Operational Definitions of'Empathy,' 'Identification, r and Related Concepts,"The Journal of Social Psychology, Vol. 63, FirstHalf (June 1964), 87-102.
Matthews, Donald R., "The Folkways of the United StatesSenate: Conformity to Group Norms and LegislativeEffectiveness," The American Political ScienceReview, Vol. LIII, No. 4 (December 1959), 1064-1089.
, U.S. Senators and Their World, Chapel Hill,----~Un~iversity of North Carolina Press, 1960.
Mead, George H., Mind, Self, and Society, ed. by CharlesW. Morris, Chicago, The University of Chicago Press,1934.
Merton, Robert K., "Bureaucratic Structure and Personality,"Social Forces, Vol. 18, No.4 (May 1940), 560-568.
Miller, Clem, Member of the Housel edited by John W.Baker, New York, Charles Scr1bner's Sons, i962.
Murstein, Bernard I., "Some Cormnents on the Measurementof Projection and Empathy," Journal of ConsultingPsychology, Vol. 21, No.1 (February 1957), 81-82.
Newcomb, Theodore M. Personality and Social Change,New York, Dryden Press, 1943.
________ , Social Psychology, New York, Dryden Press, 1950.
Norman, Ralph D., "The Interrelationships among AcceptanceRejection, Self-other Identity, Insight into Self,and Realistic Perception of Others," The Journal ofSocial Psychology, Vol. 37, Second Half (May 1953),205-235.
and Patricia Ainsworth, "The Relationships amongProjection, Empathy, Reality, and Adjustment,Operationally Defined," Journal of consu1tin~Psychology, Vol. 18, No. 1 (February 1954),3-54.
232
Norman, Ralph D., and Wa1deman C. Leiding, "Relationship between Measures of Individual and MassEmpathy," Journal of Consulting Psychology, Vol. 20,No. I (February 1956), 79-82.
Notcutt, R. and A. L. M. Silva, "Knowledge of OtherPeople," The Journal of Abnormal and Social Psychology,Vol. 46, No. 1 (January 1951), 30-37.
Pace, C. Robert, "A Situations Test to Measure Socia1Political-Economic Attitudes," The Journal of SocialPsychology, Vol. 10, No.3 (August 1939), 331-344.
Patterson, C. H., "A Note on the Construct Validity ofthe Concept of Empathy," The Personnel and GuidanceJournal, Vol. XL, No.9 (May 1962), 803-806.
Powell, Reed M. and Associates, "An Experimental Studyof Role-taking, Group Status, and Group Formation,"Sociology and Social Research, Vol. 40, No. 3(January-February 1956), 159-165.
Empathy andInnovation,
Remmers, H. H., nA Quantitative Index of SocialPsychological Empathy," American Journal of Orthopsychiatry, Vol. XX, No. 1 (January 1950), 161-165.
________, Introduction to Opinion and Attitude Measurement, New York, Harper and Brothers, 1954.
Robinson, William S., "Ecological Correlations and theBehavior of Individuals," American SociologicalReview, Vol. 15, No.3 (June 1950), 351-357.
Rodgers, David A., "Personality Correlates of SuccessfulRole Behavior," The Journal of Social Psychology,Vol. 46, first half (August 1957), 111-117.
Rogers, David A., "Relationship between Real Similarityand Assumed Similarity with Favorability Controlled,"The Journal of Abnormal and Social Psychology, Vol.59, No. 3 (November 1959), 431-433.
Rokeach, Milton, The Open and Closed Mind, New York,Basic Books, Inc., 1960.
233
Rosander, A. C., "An Attitude Scale Based upon BehaviorSituations," The Journal of Social Psychology, Vol.8, No.1 (February 1937), 3-15.
Rose, Arnold M., Human Behavior and Social Processes,Boston, Houghton Mifflin Company, 1962.
Rummel, Rudolph J., Applied Factor Analysis (forthcoming,1969).
Russett, Bruce M., Community and Contention: Britain andAmerica in the Twentieth Century, Cambridge, Mass.,M.l.T. Press, 1963.
Sarbin, Theodore R., "Role Theory," in Gardner Lindzey(editor), Handbook of Social Psychology, Reading,Mass., Addison-Wesley PUblishing Company, 1954,Vol. I, pp. 223-258.
---v;o........ ' "The Concept of Role-taking," Sociometry,Vol. VI, No.3 (August 1943), 273-285.
, and Donal S. Jones, "An Experimental Analysis---o""f-'Role Behavior," Eleanor E. Maccoby, Theodore
M. Newcomb, and Eugene L. Hartley, Readings in SocialPsychology, New York, Henry Holt and Company, 1958,pp. 465-lf:72.
Scheff, Thomas J., "Toward a Sociological Model ofConsensus," American Sociological Review, Vol. 32,No.1 (February 1967), 32-46.
Schelling, Thomas C., The Strateg~ of Conflict, New York,Oxford University Press, 19 .
Scheuch, Erwin K., "Cross-National Comparisons UsingAggregate Data: Some Substantive and MethodologicalProblems," Richard L. Merritt and Stein Rokkan,editors, Comearing Nations, New Haven and London,Yale Univers1ty Press, 1966, pp. 131-167.
Scode1, Alvin and Paul Mussen, "Social Perceptions ofAuthoritarians and Non-authoritarians," The Journalof Abnormal and Social Psychology, Vol. 48, No. 2(April 1953), 181-184.
Selvin, Hanan C., "A Critique of Tests of Significancein Survey Research," American Socio10,ica1 Review,Vol. 22, No.5 (October 1957), 519-52 .
234
Siegel, Arthur I., "An Experimental Evaluation of theSensitivity of the Empathy Test," The Journal ofA~~lied Psychology, Vol. 38, No.4 (June 1954),2 -223.
Smith, Henry Clay, Sensitivity to People, New York,McGraw-Hill Book Company, 1966.
Smith, Thomas Vernor, The Legislative wa~ of Life,Chicago, The University of Chicago ress, 1940.
Smouse, Albert D., Morris Aderman, and Charles Van'Buskirk, "Three Empathy Measures as Correlates ofTest and Rating Criteria," Psr;chological Reports,Vol. 12, No.3 (June 1963), 8 3-809.
Spilka, Bernard and Marvin Lewis, "Empathy, AssimilativeProjection, and Disowning Projection," The ps~cholog
ical Record, Vol. 9, No.3 (July 1959), 99-10 .
Stark, Stanley, "Role-taking, Empathic Imagination, andRorschach Human Movement Responses: A Review of TwoLiteratures," Perceptual and Motor Skills, Vol. 23,No.1 (August 1966), 243-256.
Steiner, Ivan D., "Interpersonal Behavior as Influencedby Accuracy of Social Perception," PSlichologicalReview, Vol. 62, No.4 (July 1955), 2 8-273.
Stewart, David A., Preface to Empathy, New York,Philosophical Library, 1956.
Stone, G. C., N. L. Gage, and G. S. Leavitt, "Two Kindsof Accuracy in Predicting Another's Responses,"The Journal of Social Psychology, Vol. 45, SecondHalf (May 1957), 245-254.
Stotland, Ezra, and Robert E. Dunn, "Empathy, SelfEsteem, and Birth-order," The Journal of Abnormaland Social Psychology, Vol. 66, No. 6 (May 1963),532-540.
Stryker, Sheldon, "Conditions of Accurate Role-Taking:A Test of Mead's Theory," in Arnold M. Rose, HumanBehavior and Social Processes t Boston, HoughtonMifflin Company, 1962, pp. 41-62.
, "Role-taking Accuracy and Adjustment,"-----S~oc-iometry, Vol. 20, No.4 (December 1957), 286-296.
235
Taft, Ronald, "Accuracy of Empathic Judgments ofAcquaintances and Strangers," Journal of Personalityand Social Psychology, Vol. 3, No. 5 (May 1966),600-604.
__-=........ ' "The Ability to Judge People," PsychologicalBulletin, Vol. 52, No.1 (January 1955), 1-23. .
Thurstone, L. L., Multiple-Factor Analysis, Chicago andLondon, The University of Chicago Press, 1947.
and E. J. Chave, The Measurement of Attitude,Chicago, Ill., The university of Chicago Press,1929.
Turner, Ralph H., "Role-Taking: Process Versus Conformity,"Arnold M. Ross, Human Behavior and Social Processes,Boston, Houghton M~ff1in Company, 1962, pp. 20-40.
, "Role-Taking, Role Standpoint, and Reference---G'='r-o-up Behavior," The American Journal of Sociology,
Vol. LXI, No.4 (January 1956), 316-328.
Vinacke, W. Edgar, The PSyChOlO,y of Thinking, New York,McGraw-Hill Book Company, nc., 1952.
Wahlke, John C and Heinz Eulau, editors, Le~islative
Behavior, The Free Press of Glencoe, I 1., 1959,pp. 294-298.
Wallen, Richard, "Individuals' Estimates of GroupOpinion," The Journal of Social Psychology, Vol. 17,Second Half (May 1943), 269-274.
Wang, Charles R. A., "sug*ested Criteria for WritingAttitude Statements, ' The Journal of Social Psychology,Vol. III, No.3 (August 1932), 367-373.
Warren, Roland L., "Cultural, Personal, and SituationalRoles," Sociology and Social Research, Vol. 34,No.2 (Nov - Dec 1949), 104-111.
White, Leonard and Thomas Vernor Smith, Politics and PublicService, New York, Harper and Brothers publishers, 1939.
Yinger, J. Milton, Toward a Field Theory of Behavior,New York, McGraw-Hill Book Company, 1965.