25222.:ch

ineq- L \\' WtNit E{£\;::1!'Q‘ L’u‘us' t'uu' it guru.

‘9 at f i ’.
:imsk {02* the Degree <22 2'25. 252,.

wt. 22, 2 «~22: 2222:2222 2w

\‘3 influx} 11:1“: SL515“; Wining—Ssh
“3 ‘ and; . m3;
b\.0 23VLL GELE fl»

1968

u i

mam

JIM/21W '

22227122222222 9 ._ r- ;

Aiming i.» S2: .
3 129310758 0312 ,s M Univwir .

 

 

2,. 2--

/«- .2 2,2
' -M
'200 A319

9.35:; 1 l .LU': aw
‘3
--—, _

3 r4.

_ _.._,__ ...__...'2»_—_—_...a—

ABSTRACT
TRUST, TBUSTNORTHINESS AND AUTHORITY

byrRobert*Gurney/

A sample of 106 38; of‘whom‘37 were female. were
tested‘ontthe fellowing seven variables: California E”
Scale.(39g lack of‘confidence-in authority figures (AP).
tendency to behave according to a norm of reciprocity (B).
trust'(T)w trustworthiness (TW), perceived trust (PT), and
perceived trustworthiness (PTW). AF was measured with a
16 item, moderately reliable (Cronbach's coefficient
alpha = .69), Likert attitude scale constructed by 3.. R,
T, TN. PT and PTW were measured with a paper-and-penciI
adaptation’of the *prisoner's dilemma".

Onrthe basis of ideas found in psychoanalysis and
sociological functionalism, together with empirical find-
ings obtained by Deutsch in 1960, three hypotheses were
formulated. (1) AP‘and R correlate negatively. (2) F and
R correlate negatively. (3) F and AF correlate either
positively or negatively.

Analysis of the data showed no support for these hy-
potheses. (1) AF correlated positively not only with R
but also with T and PT.' (2) F correlated positively with
a and PTW. (3) Although F and AF did not correlate with
one another, a significant negative correlation was found

between F and a subscale of AF, consisting of the reverse

Robert Gurney
scored items of the AF scale. This last finding seemed to
indicate the possible effect of a response bias. Since
the nonsupport of hypothesis (2) differed from Deutsch's
findings, a more detailed comparison of the two studies
was made. This comparison showed significant differences
between the two. Finally. no sex differences were found.

The discussion of this study focussed on the meaning
of F Scale scores. On the basis of a model formulated by
E to describe subjects' responding to Likert scale items.
F was interpreted as an index of ”contact with reality'--
an interpretation which seemed to adequately explain the
data from both the present and Deutsch's study. The
results for hypothesis (1) were interpreted in terms of
T rather than B.2 Finally. a three variable model for in-
vestigating T was proposed. The three variables were
perception of the past, perception of the present and

perception of the future._

l—ev~-~ 2" ' 7‘
”’1' "A“ ’v-r/ "/
AL—fl-"" / ~ ’4”), .2”"‘"‘
Z

53mmittee Chairman

, /’
Date 7 11/" (If / /_ _.

 

_‘..-

Approved

 

TRUST, TRUSTWOBTHINESS AND AUTHORITY

by 2"

x.“ 22X
RoberttGurney

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
MASTER OF ARTS

Department of Psychology
1968

To 136 introductory,.psychology students
who forced me to reconsider

a few hypotheses

iii

ACKNOWLEDGMENTS

Being able to look at sixty or so sheets of paper and
gasp, "My thesis--finishedl' I'probably differ little from
others who have had the same experience and the concomitant
feeling of gratitude to the world. Yet, there exist a few
denizens of this world to whom I feel doubly grateful.

Certainly. foremost among these persons are the three
gentlemen who served as my committee. I was extremely
fortunate to have had as my chairman, Dr. Albert Rabin.
Beth his insightful questions about my ideas and his many
courtesies which encouraged me to seek the answers to
these questions were invaluable stumbling blocks to my 1g-
norance. I am also grateful to Dr. James Phillips who,.in
addition to being a member of my committee, is my academic
adviser. To a large extent, my academic interests of
today are a result of yesterday's help from Dr. Phillips--
a fact for which hopefully I shall always be thankful. To
Dr. John Hurley. my third committeeman, I owe my discovery
of the question which constitutes the foundation of this
thesis. For this and subsequent assistance. I am very
appreciative.

While the virtues of this paper are much the virtues

of its critics..its vices are mine.

111

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS . .. . '. . . 111
LIST OF TABLES . . . . . . . v
LIST OF FIGURES . . . . . . V1
LIST OF APPENDICES . . . . . . vii
INTRODUCTION . . . . . . . 1
PROCEDURE . . . . . . . . 9
Subjects. . . . . . . . 9
Method . . . .‘ . . . . 9
RESULTS . . . . . . . . 1#
DISCUSSION. . . . . . . . 27'
The California F Scale--A detour into theory . 28
The third hypothesis , , . . . 39
The second hypothesis . . . . . #0
The first hypothesis . . . . . #2
Conclusion . . . . . . . nu
REFERENCES . . . . . . . ”6

iv

Table

LIST OF TABLES

Title

Authority Figures Scale and Part-Whole Cor-
relations Between Individual Items and Scores
on the Entire Scale

X2 Test of Sexual Differences in "Prisoner's
Dilemma" Behavior

PrOportions of NYU and MSU Subjects Manifes-
ting Different Types of ”Prisoner's Dilemma"
Behavior

Relationship Between F Scale Scores of NYU
Subjects and Types of "Prisoner's Dilemma"
Behavior

Relationship Between F Scale Scores of MSU
Subjects and Types of “Prisoner's Dilemma"
Behavior

Page

10

16

21

2h

25

Figure
1

LIST OF FIGURES
Title

Payoff matrix for a "prisoner's dilemma"
game.

vf-

Page

11

Appendix

A
B .

LIST OF APPENDICES
Title

RECIPROCITY AND AUTHORITY
QUESTIONNAIRES

vii

Page

#8
50

TRUST. TRUSTWORTHINESS AND AUTHORITY

by Robert Gurney
INTRODUCTION?‘ '

The thesis of this paper is that authority relations
ships and reciprocal relationships are both characterized
by an at least temporary surrender by one actor (Ego) of
some instrument or means for gratifying his own internal
needs and a temporary gain in this power by the other actor
(Alter). An authority relationship is defined as an inter-
personal relationship in which Alter possesses power over
Ego as a result of Ego's voluntary submission to Alter
(Mills, 1961). A reciprocal relationship is defined as a
dyadic relationship in which Ego becomes a creditor to the
debtor Alter (Gouldner, 1963). The thesis of this paper
states that insofar as Ego makes himself dependent upon
Alter's repayment of the debt for satisfaction of his
(Ego's) needs, Ego subordinates himself to Alter.

We may go beyond this initial statement to pose a
question. Does Ego's act of subordinating himself to Alter
imply that Alter possesses complete control of the Ego-Alter

dyad? One way in which to reply, ”No," to this question

 

* An abbreviated version of this introduction may be
found in Appendix A. p. #8.

2
results from postulating two forms of authority. The form
already discussed may be called interactional authority.
This refers to Alter's control over the dyad's activities
and is an effect of Ego's act of subordinating himself to
Alter. The second form may be labeled normative authority.
This refers to Ego's control of Alter as a result of a norm
which places obligations on Alter to treat Ego fairly.
Perhaps,.an example will help to clarify these concepts.
Consider a typical..dyadic, reciprocal relationship-~a
friendship between Alphonse and Cecilia. Suppose Alphonse,
a poverty-stricken, college student, suggests to Cecilia
that she let him take her to dinner. "Just name the place,”
he says. This statement gives Cecilia what has been called
interactional authority-~namely, control over the dyad's
activity. This control is not without limits, however. If?
the young lady were to suggest an extremely costly restau-
rant, Alphonse could certainly complain, 'Is this what our
love means to you?" This ability of Alphonse to appeal to
either a norm or an abstract concept like love which implies
a norm, has been termed normative authority. On the basis
of these two types of authority, we may contend that an in-
dividual subordinates himself to another only if in subor-
dinating himself, he acquires normative authority or, at
least, believes he is acquiring normative authority.

The research hypotheses formulated as guides for ana-
lyzing the above theorizing are, to some extent, based upon

three sources-~sociologica1 functionalism, psychoanalytic“

3
theory and empirical investigations of the "prisoner's di-
lemma'.. The contribution of sociological functionalism is
its analysis of social systems and the relationship of
these systems to personality systems. From this analysis
five propositions have been extracted.

1. According to Pareto social systems strive to main-
tain equilibrium and can be disequilibriated only by forces
external to themselves. They react to disequilibrium by
restoring themselves to equilibrium (1,23,,the original
state of the system). (Martindale, 1960).

2. The personality system is distinct from the social
system (Parsons. 1951).

3. The personality system is motivated by a tendency
to maximize gratification of this system's need disposi-
tions (Parsons, 1951).

4. Social norms "constitute . . . the core of the
stabilizing mechanisms of the system of social interaction“
and are essential components of both personality and social
systems (Parsons, 196A).

5. The "norm of reciprocity, in its universal form.
makes two interrelated, minimal demands: (a) people should
help those who have helped them, and (b) peOple should not
injure those who have helped them.” (Gouldner, 1963). i

From prOpositions i. 2 and 3, it follows that disequi-
librium in a social system can be caused by a personality
system's attempt to gratify its needs. If gratificationsof

needs is governed by a social norm, however, the disequilib-

h
rium of the social system will be only temporary (by propo-
sitions 1 and h). More specifically. if an interaction is
governed by a norm of reciprocity, the gratification of
Alter's needs with the assistance of Ego will be accompanied
or followed by the gratification of Ego's needs with the
assistance of Alter (by propositions 1,2h and 5).

From psychoanalytic theory five propositions about the
nature of personality systems have been extracted._

1. The ”principle of stability" states that the "organ-
ism strives to preserve those optimal internal conditions
under which the process of life is possible.” (Alexander.
19MB). This is analogous to Pareto's concept of equilibrium.

2.. Intrapsychic excitation or instability may be caused
by extrapsychic or external stimuli or by the pressure of
internal needs, impulses and wishes (Alexander,.19h8).

3. The usual sources of instability are the internal
needs. impulses and wishes of the organism (Alexander,.19h8).
A. '. . ..the ego's basic function is to maintain
constant conditions in the organism. It is the agent of

the stability principle." (Alexander, 19h8).

5. To optimally satisfy internal motives, the ego
makes an at least tacit contract with society. By this
contract the individual agrees to submit to societal author-
ity on the condition that this submission helps him to
satisfy his internal needs. Within the individual the meter
of this contract's success is the "sense of justice'--the

individual's feeling of being fairly or unfairly treated by

5
the societal authority figures (Alexander & Staub, 1956).

If the five propositions of sociological functionalism
and the five pr0positions of psychoanalysis are juxtaposed,2
two significant differences between social and personality
systems emerge. First, while the social system is disequi-
libriated only from the influence of forces external to
itself (by sociological functional prOposition 1), the per-
sonality system generally loses stability from the influ-
ence of forces within itself (by psychoanalytic propositionl
3). Second. while norms pilot disequilibriated social
systems back to equilibrium. the ego performs this guiding
function in the personality. Since within the latter sys-
tem, some internalized social norms reside in the superego
(Alexander, 1948) rather than in the ego, the principle of
equilibrium motivating the social system need not be com-
patible with the stability principle motivating the per-
sonality system. Hence, there is the possibility of
intersystem conflict.

Alexander and Staub (1956) have outlined the dynamics
of this intersystem conflict for the special case of neur-
otic characters.

The psychological situation in the case when the

sense of justice is injured, is similar to the

case of self-defense; it leads to rebellion and

to the breaking through of primitive anti-social

drives: one feels oneself threatened by the very

authorities whose business it is to uphold the

law. . . . the Superego of adults is more or less

dependent upon the individual's relation to those

in authority; if the latter. by their injustice,

destroy the confidence one has in them, the Ego

at once loses its dependence on the inner repre-
sentative of authority-~the Superego. (p. 83).

6
In other words. as a consequence of feeling himself victim-
ized, rather than helped by the social contract, the indi-
vidual loses confidence in authority figures.

If. however, the norm governing social system equilib-
rium is compatible with the principle.of stability. there
will be no intersystem conflict.

It is theorized that the social norm of reciprocity
promotes intrapsychic stability.. It accomplishes this by
assuring the individual that when he transfers interactional
authority to another individual, he acquires normative
authority (see pp. 1-2). In other words, if the individual
perceives a functioning norm of reciprocity, he probably
also perceives a functioning social contract because, in
both cases, the functional reality of the norm or contract
is dependent upon the efficacy of normative authority.
Then, the extent to which an individual's sense of justice
is unoffended should be related to the extent to which he
perceives the reality of a functioning norm of reciprocity.
We may formalize this as our first hypothesis.
gypothesis 1: There is a negative correlation between lack
of confidence in authority figures (1.3., the extent to
which the individual's sense of justice is offended) and
willingness to participate in a reciprocal relationship.

Conveniently, an experimental situation, the ”prison-
er's dilemma,“ exists for studying the functioning of
reciprocal relationships. Two studies of this situation

are of particular relevance to this paper.

The first (Deutsch, 1960) was an attempt to relate
California F Scale scores to trust (initiation of a recip-
rocal relationship by Ego) and trustworthiness (Alter's
returning the social system and Ego's personality system to
equilibrium and stability by fulfilling the dictates of the
norm of reciprocity). There were two major findings of
this study. First, since subjects who trusted tended to
also be trustworthy while those who did not trust tended to
be untrustworthy, the norm of reciprocity seemed to be a
valuable construct for explaining the data. In other words,
84 per cent of the subjects fell into categories which
could be labeled wholly reciprocal or wholly nonreciprocal.
Of the remaining 16 per cent, 9 per cent were trusting but
untrustworthy and 7 per cent were suspicious but trust-
worthy._ Wholly reciprocal subjects both by initiating a
reciprocal relationship and by fulfilling the dictates of
the norm when the other member of their dyad initiated a
reciprocal relationship, could be said to be perceiving
the reality of a norm of reciprocity. Wholly nonreciprocal
subjects did precisely the opposite.

The second finding of interest was that low F scores
characterized subjects who behaved according to the norm of
reciprocity; relatively high scores characterized subjects
who behaved wholly against the norm. If an F score is, in
fact, an index of an individual's attitude toward authority,
Deutsch's data seem to indicate the worth of further inves-

tigating the relationship between authority and reciprocity

relations.

A modified replication of Deutsch's study casts doubt
upon the stability of his findings. Wrightsman (1966)
found that attitudes toward authority, measured by both the
California F Scale and Chein's Anti-Police Attitudes Scale,
were not significantly related to game behavior. When
trust and trustworthiness were analyzed independently,
“none of the attitude and personality measures was success-
ful" in predicting trustworthiness. In predicting trust
only Wrightsman's own ”PhilOSOphies of Human Nature Scale"
succeeded.

Since there does exist doubt about the relationships
found by Deutsch, it seems worthwhile to attempt to repli-
cate his study. It also seems worthwhile to attempt to
assess the relationship between the California F Scale and
lack of confidence in authority figures. In order to for-
malize the investigation, we may formulate two more hypo-
theses.

Hypothesis g; F Scale scores correlate negatively with

 

willingness to participate in a reciprocal relationship.
Hypothesis 1; There exists a significant correlation,
either positive or negative, between lack of confidence in

authority figures and F Scale scores.

PROCEDURE

Subjects
A sample of 108 introductory psychology students at

Michigan State University served as subjects for this study
during the spring of 1967. All subjects were nonvolunteer
(1,2,, subjects were tested in classrooms during scheduled
class time; no foreknowledge of the substitution of experi-
mental testing for the scheduled classroom instruction was
given to the subjects). Of these subjects, 37 were female.

Also, all subjects were students of the experimenter.

Method

Data were collected with a two part, paper-and-pencil
test (see Appendix B, p. 50). Part One consisted of a ran-
dom intermixture of the 29 items of Form #5 of the Califor-
nia F Scale (Adorno, Frenkel-Brunswik, Levinson, & Sanford,
196A) and 16 items (Authority Figures Scale) designed by the
experimenter to assess the subject's lack of confidence in
authority figures (see Table 1). Each item consists of a
statement attributing adequate or inadequate role perfor-
mance to an authority figure. Whether or not a social role
actor constituted an authority figure and the exact nature
of adequate role performance were decided a’priori by the

test constructor. The items were scored in such a way

9

TABLE 1

Authority Figures Scale and Part-Whole Correlations

10

Between Individual Items and Scores on the Entire Scale

*3.

*10.

11.

12.

*13.

16.

*19.

20.

*23.

26.

*32.

33.

35-

#3.

Item

The typical father rewards his children for good
behavior. '

Host employers underpay their employees.

Most politicians would readily vote against the
best interests of their electors.

Policemen are usually fair and just officers of
the law.

Clergymen are usually more concerned about the
contents of the Sunday collection basket than the
contents of their parishioners' souls.

The typical father does nothing to merit respect
from his children.

Parents usually punish their children only when
punishment is necessary for the child's welfare.

The average policeman misuses his authority.

Most teachers are fair and just graders.

Most teachers would rather be popular with their
students than satisfy their students' educational

needs.

Mothers generally strive more to love than to be
loved.

When the probability of being discovered is small,
the average politician uses taxes primarily for
his own benefit.

Most children are well cared for by their mothers.

The average courtroom judge is more interested in
re-election than in being just.

Parents behave as if it were better to punish a
child than to love him.

Most employers overwork their employees.

* This item was scored in reverse.

1‘

.28

.46‘
.38

.b8

.51

~53

.uz

.32

.21

.26

.129

.50

.#2

.52

.50

11
that a high score indicated a lack of confidence in
authority. Finally. unlike the F Scale. six of the
Authority Figures items (3, 10, 13, 19. 23 and 32) were
scored in reverse. This reverse scoring was employed in
order to control for possible acquiescence response set”
tendencies.

Part Two consisted of a story about two fictional
characters involved in a ”prisoner's dilemma” game (see

Appendix B, p. 50). The payoff matrix of this game is

Alter's choices

 

 

I II
I +500,+500 -1000,+1000
Ego's
choices
II +iOOO,-iOOO -500.-500

 

 

 

 

Fig. 1. Payoff matrix for a "prisoner's dilemma" game.
(The first number in each cell represents Ego's payoff.
The second number is Alter's payoff. For Ego choice I

is defined as "trust" while choice II is "suspicion". For
Alter choice I following choice I by Ego is defined as
”trustworthiness" while choice II following choice I by
Ego is "untrustworthiness”.)

shown in Figure 1. Subjects were asked to predict how the
persons playing the game in the story would, in fact, play
it. They were then asked how they themselves would play.
The latter set of reSponses was used as an index of the
subject's trust and trustworthiness. The former set of

answers was used as an index of the subject's reasons for

12
using whichever strategy he employed. The questions con-
cerning the subjects' own behavior were asked in such a
way that the subjects were supposed to imagine them-
selves playing the game with one of the fictional char-
acters. The questions concerning the behavior of the fic-
tional characters would then yield information about the
subjects' perception of the other player's (i.§., the
fictional character's).behavior. This information was then
used for explanatory purposes. For example, suppose a
subject described the fictional characters as trusting and

rustworthy. Suppose, moreover, when asked how he himself
would play the Ego role, he replied with a trusting
response. We could, then, explain his response as based
upon either conformity to the behavior of others (1.2..

the perceived trusting response) or expectation of having
his own trust reciprocated (i.g., the perceived trustworthy
response) or both.

The rationale behind this type of game is relatively
simple. A payoff matrix is presented to the subject. It
is explained to him that the first number in each cell of
the matrix refers to Ego's payoff or winnings. The second
number in each cell is Alter's payoff. The subject is
then asked to assume the role of Ego. As Ego he faces a
choicatbetween playing one of two strategies, “collective
rationality” or "individual rationality” (Rapoport. 1966).
The collectively rational strategy is based upon the re-

alization that if Ego does not trust Alter, the almost in-

 

 

 

 

13
evitable result is that both players lose $500. On the
other hand, if Ego trusts Alter and if Alter proves trust-
worthy, both win $500. Ego's collectively rational strategy
is, then, to trust Alter. Ego's individually rational
strategy, however, instructs him not to trust Alter. The
rationality of this strategy lies in the possibility of
Alter being untrustworthy. If Ego trusts Alter and Alter
is, in fact, not trustworthy, Ego will lose $1000. If he
does not trust Alter, the most he can lose is, of course,
$500. As an index of trust and suspicion, then, the
”prisoner's dilemma" does not lack a logical basis.

Following the subject's choice of strategy as Ego, he
is asked to assume the role of Alter. He is told that the
Ego with whom he is playing has selected the trusting
strategy. He is then asked to choose between the two Alter
alternatives. His choice defines him as either trustworthy
or untrustworthy.

Finally, it should be noted that for purposes of
generalizing beyond the experimental situation, “prisoner's
dilemma“ scores are of dubious validity. Apparently, no
one has conclusively established the relationship between
trust and trustworthiness in game behavior and trust and
trustworthiness as characteristics of other types of be-
havior. Also,.in the present experiment no attempt is
made to assess the reliability of ”prisoner's dilemma"

SCOI‘SS.

RESULTS

Although no sex differences were hypothesized, pre-
liminary checks were made on the effect of this variable
before combining the data. The mean on the F Scale for
females was 3.08: for males, 3.00. Two 2 tests were used
for testing the difference between these means. First, a
t for independent measures (McNemar, 1962) was used. The
F Scale scores for 37 female subjects constituted one
sample while the F Scale scores for 71 male subjects con-
stituted the other sample. The difference between the
means of these two samples was not significant (£3.81;
df=106: p).05, for two tailed test). Second, a _t_ for
correlated measures (McNemar, 1962) was used. Although
the means for this test were identical to the means em-
ployed in the preceding test, the samples were different.
In this case, the female sample consisted of 29 scores--
each score being the mean of all responses given to one of
the 29 items of the F Scale by the 37 female subjects.

The male sample consisted of the same 29 scores, except

for this sample the scores were the means of responses
given by the 71 males. The correlation between the means
per F Scale item for females and males was .94. The dif-
ference between the means of these two samples approached,
but did not reach, significance (§;1.82; df=28; .10)'pj>.05,

for two tailed test).
in

15

On the Authority Figures Scale, the mean for females
was 2.67. For males the mean was 2.69. Again, two £.tests
were used for testing the difference between these means.
First, a‘t for independent measures (McNemar, 1962) was used.
The 37 female, Authority Figures Scale scores constituted
one sample while the 71 scores for males constituted the
other sample.. No significant difference between the
means of these two samples was found (_t_=.38; df=106; 2) .05,
for two tailed test). Second, a £_for correlated measures
(McNemar, 1962) was employed. The means for this test were
identical to the means used in the preceding test. However,
the female and male samples consisted of 16 scores apiece.
Each score was the mean of all responses given to a par-
ticular item of the 16 item Authority Figures Scale by the
subjects of one sex. The correlation between the means
per item for females and the means per item for males was
.69. 3; was not significant (£=.l+8; df=153 p‘>.05).

In order to determine the relationships between sex
and individual items of the Authority Figures Scale, point
biserial correlations (Baggaley, 196#) were computed for
each item. Only the coefficient for item 19 (”Most teach-
ers are fair and just graders.') was significant (£pb=.27:
df=1073 p‘1.05) with females showing less confidence in
teachers than did males.

Finally . a x2

test (Hays, 1963) was used to assess
sex differences in trust and trustworthiness. Because of

a small expected frequency for the “suspicious and trust-

16
worthy” category, the four possible categories were re-
duced to three: (a) reciprocal (subjects who were both
trusting and trustworthy), (b) partially reciprocal (sub-
jects who were either trusting but untrustworthy or sus-
picious but trustworthy), and (c) nonreciprocal (subjects
who were both suspicious and untrustworthy). This data is
presented in Table 2. The evidence indicates no sex dif-

ferences (§?=.666; df=2; 27>.05).

TABLE 2

X2 Test of Sexual Differences

In ”Prisoner's Dilemma“ Behavior

 

 

 

 

 

 

Sex Type of Behavior
Partially
Reciprocal reciprocal Nonreciproca;
Obs. Exp. Obs. Exp. Obs. Exp. Total_
Female 12 10.85 12 11.20 13 14.95 37
Male 19 20.15 20 20.80 30 28.05 69
Total 31 31.00 32 32.00 #3 ”3.00 106

 

§?=.666; df=2:'2j>.05 (not significant)

 

Note.—-'Obs.' refers to observed frequencies. "Exp.”
refers to expected or theoretical frequencies.

On the basis of the above evidence, it is assumed
that sex differences play no significant role in the data
of this study. Only one significant sex difference was

found-~one item of the 16 item Authority Figures Scale

17
was found to be significantly related to sex. Yet, since
its significance level is .05, the relationship seems at-
tributable to chance (1.2., since the probability of this
relationship occurring by chance is i in 20, it is not
surprising to find one such relationship in a sample of
16 relationships).

Before testing the research hypotheses, internal con-
sistency was determined for the F Scale and the Authority
Figures Scale. Cronbach's (1951) “coefficient alpha” was
used for this purpose. Since alpha is'the mean of all
possible split-half estimates of a testis reliability, it
is the best index of internal consistency available.
Coefficient alpha for the F Scale was found to be .83.
For the Authority Figures Scale, alpha was equal to .69.

In order to compare the reliabilities of these two
scales, the Spearman-Brown formula (Baggaley, 196“) was
used to estimate the reliability of the Authority Figures
Scale if it were to be increased to 29 items, the number
of items in the F Scale. It was found that alpha would
probably increase to .80. It seems, then, that the dif-
ference between the reliability of the F Scale and the
reliability of the Authority Figures Scale may be ex-
clusively an effect of the different lengths of the two
scales.

In order to further explore the internal consistency
of the Authority Figures Scale, part-whole correlations

were computed. For each item the subjects' responses to

18

the item were correlated with subjects' scores on the en-
tire scale. These correlations are presented in Table 1
(p. 10). Fifteen of the part-whole correlations were
positive and significant at the .01 level. The other cor-
relation was positive and significant at the .05 level.
The mean of these correlations was .44. The range of the
correlations extended from .21 for item 20 ("Most teachers
would rather be popular with their students than satisfy
their students' educational needs.") to .64 for item 16
(”The average policeman misuses his authority.”).

The first hypothesis of this study states that there
is a negative correlation between Authority Figures Scale
scores and willingness to participate in a situation
governed by the norm of reciprocity. In order to test this
a point biserial correlation was~calculated, comparing the
mean of Authority Figures scores for subjects who were
both trusting and trustworthy with the mean scores for
subjects who were neither trusting nor trustworthy. The
hypothesized negative correlation was found to be +.24
(df=73; p‘<.05), a significant positive correlation.

In order to better understand the above correlation,
the relations between lack of confidence in authority
figures and a variety of variables presented by the game
situation were analyzed with biserial correlations. Four
biserials were computed--each relating scores on the
Authority Figures Scale to a dichotomized game variable.

The four game variable dichotomies included trust-suspicion,

19
trustworthiness--untrustworthiness, perceived trust--per-
ceived suspicion, and perceived trustworthiness--perceived
untrustworthiness. Since two of these correlations were
virtually zero, let us concentrate only upon the two sig-
nificant ones. First, it was found that subjects who
described themselves as trusting tended to have higher
scores on the Authority Figures Scale than did suspicious
subjects (£b=.31; df=105; p«<.01). Since high scores on
the Authority Figures Scale indicate lack of confidence in
authority, this finding may be restated as: trusting sub-
jects tend to have less confidence in authority figures
than do suspicious subjects. The second significant corre-
lation indicates that subjects who perceived the game behav-
ior of fictional characters to be trusting tended to have
higher scores on the Authority Figures Scale than did
subjects who perceived suspicious behavior (£b=.26;df=105;
p,<.01). In other words, subjects who perceive others as
trusting have less confidence in authority figures than do
subjects who perceive others to be suspicious.

Hypothesis 2 states that there is a negative correla-
tion between F Scale scores and willingness to participate
in a situation governed by the norm of reciprocity. In
order to test this, a point biserial was calculated com-
paring the mean of F Scale scores for subjects who were
both trusting and trustworthy with the mean of F Scale
scores for subjects who were both suspicious and untrust-

worthy. The hypothesized negative correlation was found

20
to be +.3O (df=73; 2;.01). Since this is diametrically
opposed to Deutsch's (1960) findings, a more comprehensive
comparison between the data of the present study and that
obtained by Deutsch was made.

It will be recalled that the norm of reciprocity
adequately characterized his data (1.2., most subjects were
either both trusting and trustworthy or both suspicious and
untrustworthy). Table 3 compares the findings from the two
studies in terms of proportions of subjects falling into
each of the four categories of game behavior. The findings
for suspicious subjects are virtually identical in the two
studies. The difference lies entirely in the greater propor-
tion of trusting subjects who are not trustworthy in the
present study. Consequently, the construct of a norm of
reciprocity fits Deutsch's data slightly better than it does
the present data. While 84 per cent of the NYU students
were either both suspicious and untrustworthy or both
trusting and trustworthy, only 70 per cent of the MSU students
fell into these two categories. The difference between these
proportions approaches significance (£;1.69; df=159;

.10>p> .05, for two-tailed test).

Deutsch also found that most of his subjects behaved in
accordance with their expectancy of the other player's behav—
ior. Although he reported no percentages, he contended
that most subjects who trusted expected the other player to
be trustworthy; subjects who were suspicious expected the

other player to be untrustworthy. In the present study, two

Proportions of NYU and MSU Subjects Manifesting
Different Types of "Prisoner's Dilemma“ Behavior

TABLE 3

21

 

Type of behavior

Trusting

Suspicious

Trustworthy

Untrustworthy

Trusting and

trustworthy

Trusting but
untrustworthy

Suspicious but
trustworthy

Suspicious and
untrustworthy

PrOportion of
NYU subjects

.527

.473

.509

.491

.436

.091

.073

.400

.518

.481

.367

.632

.292

.226

.075

.406

Proportion of
MSU subjects

 

22
constructs were available for explaining the data. The
first was Deutsch's expectation construct. A subject's be-
havior could be explained by this construct if either (1)
he behaved trustingly and perceived the other player as
trustworthy or (2) was suspicious and.perceived the other
player as untrustworthy. A second construct may be called
"conformity". A subject's behavior could be explained with
this construct if he played the game in exactly the same way
as he perceived the fictional characters' game behavior.
There are four possible patterns of responding which may be
described as conforming: (1) a subject may be trusting and
trustworthy and perceive others as trusting and trustworthy;
(2) a subject may be trusting and untrustworthy and perceive
others as trusting and untrustworthy; (3) a subject may be
suspicious and trustworthy and perceive others as suspicious
and trustworthy; and (4) a subject may be suspicious and un-
trustworthy and perceive others as suspicious and untrust-
worthy. In the present study,.while the expectation construct
described the behavior of 62 per cent of the subjects, con-
formity accounted for the behavior of 65 per cent of the
subjects. The difference between these two pr0portions is
considerably more interesting if we consider the chance prob-
abilities of people behaving according to one or the other
construct. By chance alone, expectation should describe the
behavior of 50 per cent of the subjects. Conformity, on the
other hand, should describe the behavior of only 25 per cent

of the subjects. It seems, then, that in the present study,

23
conformity provides a more satisfactory description of the
data than does expectation. The two constructs together
describe 78 per cent of the data. Since Deutsch reported
no percentages of subjects behaving according to either of
these two constructs, however, no realistic comparison of
the two studies can be made.

Finally, Tables 4 and 5 present a comparison between“‘
the relationship of F Scale scores to game behavior for the
NYU subjects (used by Deutsch) and the MSU subjects (used in
the present study).respectively. It is clear that the
findings of the two studies differ. Whereas Deutsch found
relatively clearcut relationships between F and game behavior,
the present study does not reveal these relationships.

In order to further clarify the nature of the present
study's findings, biserial correlations were calculated in
an attempt to assess the relationship between F Scale scores
and each of four dichotomous game variables.~ The game
variables included trust--suspicion, trustworthiness--
untrustworthiness..perceived trust-~perceived suspicion,
and perceived trustworthiness--perceived untrustworthiness.
Three of these correlations were virtually zero. Only the
correlation between F Scale scores and perceived trust-
worthiness--perceived untrustworthiness was significant
(gb=.30; df=105; pg<.01). Verbally, this result states
that subjects who perceive others as trustworthy tend to
have higher F Scale scores than do subjects who perceive

others to be untrustworthy.

TABLE 4

24

Relationship Between F Scale Scores of NYU Subjects
And Types of “Prisoner's Dilemma" Behavior

 

Game behavior

Trusting and
trustworthy

Suspicious and
untrustworthy

Suspicious but
trustworthy

Trusting but
untrustworthy

Low
(102‘-202)

.50

.00

.00

.40

F Scale score

Medium
(203-'303)

.42

~59

1.00

.60’

Total
High
(3.4--4.4)
.08 1.00
.41 1.00
.00 1.00
.00 1.00

 

Note.--25 per cent of the subjects had "Low‘ F Scale scores;
55 per cent had "Medium" scores: and 20 per cent had ”High“
scores. For the prOportions of subjects manifesting each

type of game behavior, see Table 3..

TABLE 5

25

Relationship Between F Scale Scores of MSU Subjects

And Types of "Prisoner's Dilemma" Behavior

 

Game behavior

 

Low
(1.0-2.79)

Trusting and .26
trustworthy

Suspicious and .26
untrustworthy

Suspicious but .12
trustworthy

Trusting but .29
untrustworthy

F Scale score*

Medium
(2.8-3.41)

.55

.58

.63

.46

High

(3.45-4.0)

.19

.16

.25

.25

Total

1.00

1.00

1.00

1.00

 

Note.—-26 per cent of the subjects had ”Low” F Scale

scores; 54 per cent had ”Medium“ scores; and 20 per cent
had "High" scores. For the pr0portions of subjects mani-
festing each type of game behavior,,see Table 3.

26

From the above we may conclude that Deutsch's results
were not successfully replicated. The findings of the
present study indicate only that there seems to be a posi-
tive relationship between F Scale scores and an individual's
willingness to participate in a reciprocal relationship and
a positive relationship between F and the tendency to per-
ceive others as trustworthy..

The third hypothesis states that there is a relation-
ship between F Scale scores and lack of confidence in1
authority figures. The obtained correlation between the F
Scale and the Authority Figures Scale is .O4--clear1y not
significant.

Further investigation of the third hypothesis seemed
in order. It will be remembered that the Authority Figures
Scale consisted of two types of items-~10 items scored in“
the same way as all F Scale items were scored'and 6 items
scored in reverse. It was feasible” therefore, to divide
each subject's Authority Figures score into two separate
scores--one based on the 10 positively scored items and the
other based on the 6 reverse scored items. Two correlations
were then computed. Between F and the positively scored
Authority Figures subscale, the correlation was found to be
+.14 (df=107: not significant). Between F and the reverse
scored Authority Figures subscale, the correlation was -.26

(df=107: p< .01).

DISCUSSION

The following commentary on the research just pre-
sented is so lengthy that a brief introduction may be of
assistance to the reader. It will be noted that the main
focus of the discussion is the meaning of California F’
Scale scores. The criterion for selecting this focus was
twofold. First” two of the three hypotheses tested were
based upon the assumption that F Scale scores are an index
of attitudes toward authority. However, since the F Scale
literature casts doubt upon this assumption (2.5,,.Peabody,
1966), a reformulation of the meaning of F Scale scores
seems in order. Second, since the other two sources of
data (1.2,, Authority Figures Scale and paper-and-pencil
form of the "prisoner's dilemma”) were constructed for and
used only in the present study, little is knOwn about their
value.~ The Authority Figures Scale is, of course, of a con-
ventional type of Likert attitude scale--consisting of
repeated measures of attitude toward an object. Hence, its
value is probably greater than the value of the rather un-
conventional "prisoner's dilemma” test, which lacks repeated
measures. Yet, in neither of my own tests do I place a
great deal of confidence. Consequently, I plan to devote
relatively little space to the discussion of variables

measured by these instruments. Only when I have collected

27

28
data based on different estimates of these variables will I
feel competent to adequately discuss the relationships
between these variables.

The following, then, focusses on the question of what
the California F Scale measures. 1y answer to this question
is based upon a theory of attitudes which is a generaliza-
tion from two sources. First, preliminary to the research
reported herein, I carried out several pilot studies. Two
consisted of only four subjects each. The small size of
these samples was particularly conducive to discussion,
subsequent to the testing. In the process of these discus-
sions, it became apparent that my subjects considered many
of the F Scale statements to be ”screwy“ and nonsensical.
This was not true of the Authority Figures Scale. Often,
subjects spontaneously said,."It was hard to know what you
were looking for,“ or something of that sort. In addition
to these discussions, the other source for the theory was
myself--attempting to answer the question, what would I

have done if I were a subject in this experiment?

The California F Scale-~A detour into theory

I found the most surprising result of this study to be
the virtually zero correlation between the F Scale scores
and lack of confidence in authority figures. To some extent,
this may be explained as a result of response bias. The
correlation between F and the reverse scored Authority
Figures subscale was significantly negative. Yet, the mag-

nitude of this correlation is so small that response bias

29
seems to provide only part of the answer.

There is one study (Rudin,.1961), however, which helps
to clarify the problem. Rudin constructed a true--false
scale much like the Authority Figures Scale. He designed
this scale to measure the subjects' acceptance of rational
authority--as defined by Erich Fromm. Assuming that the
F Scale measures attitude toward irrational authority, he
hypothesized that the two scales would not be related.
Between the two scales he obtained a correlation of .04.

Attempting to explain my own findings, I have elected
to disagree with Hudin's assumption that the F Scale directly
assesses attitude toward irrational authority. It may be
true that response to an item like, ”Some day it will probably
be shown that astrology can explain a lot of things,"
correlates with attitude toward irrational authority. Yet,
is this what the item pg; s2 assesses? Restricting attention
to the item's content, it seems that a respOnse to it is
simply an index of an individual's attitude toward the
future of astrology. Content analyses of other F Scale
items reveals the foci of these items to be exotic objects
like people as germ carriers, the ethics of childhood.
earthquakes and floods as panaceas for social problems, 33
cetera.

In addition to the objects contained in these state-
ments, there is at least one other salient characteristic.

If we view an attitude statement as an expressed relation-

ship between objects, then certainly the nature of the

30
relationship is also salient. In F Scale items this rela-
tionship is generally one of 'oughtness" or futurity.
Subjects are faced with the task of agreeing or disagreeing
with statements of the type, "A will be B"--or more specifi-
cally, ”Astrology will be a powerful explanatory tool.”

Let us generalize the above considerations to a theory.
This theory is based on the premise that an attitude state-
ment is relevant to a subject if and only if both the objects
and the relationships are relevant. Before proceeding, let
me attempt to clarify what I mean by a "relevant" statement.
Heuristically, a statement is relevant to a subject if that
subject can use the statement as a testable hypothesis. The
data for testing the hypothesis is an element of the sub-
ject's set of retrievable experiences. For example, Rudin's
scale has an item which reads, ”Most policemen are fair and
honest." If this item is relevant to a subject, he may
recall one or more experiences in which a policeman did or
did not behave fairly or honestly or both. His response,
then, will be based upon the retrieved experiences. On the
other hand,.there are at least two types of statement which
are.not simply testable hypotheses. The first is an hypo-
thesis which is theoretically testable but for which a
given subject has no data. For example, "Wag Egg 22333 is
a long novel,P may be untestable for Ezra McCoy in the back
hills of Tennessee. Since in his memory store, there is no
cell labeled, Wag'ggg_§gagg, the statement is irrelevant to

Ezra. The second type of irrelevant statement is one which

31
is not a hypothesis at all. Instead,.it is a description~
of a state of affairs which ought to or will exist but which
presently does not. Let us label these three types of at-
titude statement Type A, Type B and Type C; respectively.

In terms of the preceding, it seems that the Authority
Figures Scale is composed primarily of Type A statements
while the F Scale consists primarily of Type C statements.

As Type C'statements, then, F Scale items are defined
as irrelevant to a subject. The question now is, how can
irrelevant statements be made relevant to a subject? That
they are, indeed, made relevant may be inferred from the
fact that subjects respond to the items. In accord with our
theory, let us say that these Type C statements are, in some
way, transformed to Type A statements. When a subject is
unable to make such a transformation, he will either not
respond at all or, if a Likert scale with a neutral category
is used, he will respond with neutrality.

Up to this point we have merely said that since F Scale
items may be described as Type C statements, subjects
reSpond to them only after they transform them to Type A
statements. We now face the problem of analyzing this
transformation process. Before doing so, however, let me
give the reader an idea of the direction in which this
argument is tending. In what follows, I intend to show
that it is reasonable to contend that F Scale items are
transformed to Type A statements by prefacing each item

with the phrase, "The experimenter believes . ...9 From

32
this point, I shall attempt to show that it is also reason-
able to interpret F Scale scores as indices of 'contactt
with reality.”

To begin, then, there seem to be at least three ways
in which a Type C statement may be transformed to Type A:
translation, ideational referral, and authoritarian referral.
Since I intend to assume that the subjects in both my own
experiment and Deutsch's (1960) experiment used authoritar-
ian referral, let me merely define the first two transforma-
tion techniques. By "translation” I mean the process of
restating the contents of an item to eliminate the ”will"
or 'ought"or "should be“ and replace it with an ”is”. For
example, let us assume the F Scale item, ”Every person should
have complete faith in some supernatural power whose decisions
he obeys without question,“ to be Type C. As Type A it
might read, "Every person has complete faith in some super-
natural power whose decisions he obeys without question.“
This statement could then be tested with the subject's
retrievable experiences.

By "ideational referral” I mean the process of testing‘
a statement against some other statements which the subject
uses as shorthand summaries of many experiences. This
process is analogous to the use of Aristotelian syllogisms..
Suppose an individual's experience leads him to conclude
that people desire happiness. Suppose, moreover, that the
people he knows who believe in an authoritarian God are

unhappy. The subject could then institute a syllogism.

33
The major and minor premises would be the two experientially
based statements just mentioned. The F Scale item on belief
in an authoritarian God would constitute the conclusion to
be tested. The subject in question would probably conclude
that the F Scale item is incompatible with his experience
and reject it as an hypothesized conclusion.

In a Comtean sense, if translation is the positivistic~
technique for transformation and ideational referral is the
metaphysical technique, authoritarian referral is the theo-
logical approach. Authoritarian referral may also be
viewed as a specific type of ideational referral. It is
based upon the subject's idea that if he wishes to succeed
at the task of filling out an attitude questionnaire, he
should refer each item to that set of ideas which he thinks
characterizes the thinking of some person other than him-
self. For example, he might restate the F Scale item dis-
cussed in the preceding two paragraphs as, "Daddy Grace
believes that every person should have complete faith in
some supernatural power whose decisions he obeys without
question." In interpreting the findings of the present
study and the findings of Deutsch's study, I am assuming
that this technique of authoritarian referral was the pre-
dominant technique used by the subjects. I shall also as.
sume that the ”authority” to whom the items were referred
was the experimenter.

In order to justify this last assumption, let us

direct attention to the possible choices of authority.

34
These choices may be divided into persons within the exper-
imental situation and persons outside this situation. Since'
the persons outside the situation are innumerable and virtu-
ally impossible to specify, let us focus attention upon those
choices who exist within the experimental situation prOper.
There are at least three such candidates for the role of
'authority'--the individual subject's ideal self, the whole
group of subjects taking the test, and the experimenter.
It is my own belief that any one of these three would
constitute a reasonable choice.

Suppose we first choose the individual's self as the
"authority". Then a subject's responses will reflect what.
he thinks he ought to say-~but ”ought to say” to whom?

Since only two people have access to the responses, the
subject is deceiving either himself, the experimenter, or
both. If he is deceiving the experimenter, his responses
will probably reflect his perception of the experimenter's
own responses to the F Scale items. In this case, the
experimenter becomes the ”authority". Suppose, on the
other hand, we theorize that the subject is not concerned
with the experimenter's evaluation of his responses. As I
see it,.such a contention leads us into a position which is
extremely difficult to maintain. In short, how do we ex-
plain the "double agreement” phenomenon (McBride & Moran,
1967; Peabody, 1966)--the fact that an ominously large pro-
portion of subjects who agree to any specific F Scale item

also agree to a reversal of that item? A reasonable reply

35
would be that a subject double agrees because his ideal
self tells him to be an agreeable fellow. It seems more
reasonable, however, to contend that double agreement is
based upon a subject's belief that since the experimenter
is using the test, the experimenter himself agrees with the
items. To anyone familiar with this type of research, such
a belief seems ridiculous; yet, we must remember that sub-
jects who agree with F Scale items tend to be in lower
social classes, have little education, have low IQ scores,
33 cetera--in short, are not "socially sophisticated" sub-
jects (Christie & Cook, 1958). Such an interpretation is
also in accord with Peabody's (1966) contention that double
agreement is behavior characteristic of relatively ”simple
minded" subjects.

Finally, contending that double agreement is not
founded upon a subject's ideal self instructing him to be
agreeable is also based upon the absence of a double dis-
agreement phenomenon. Why are there no "scientific" ideal
selves which instruct subjects to doubt the validity of any
statement and consequently disagree with both original and
reversed items? Defining the experimenter as the subjects'
authority, however,.we may suggest that double disagreement
has not been found because subjects who score low on F are
relatively socially SOphisticated. This enables these low
F scorers to know that psychological tests do not neces-
sarily contain statements of the experimenter's own

beliefs or disbeliefs.

36

Let us now suppose that subjects are responding in
terms of the whole group of subjects' beliefs. This is,
again, a fairly reasonable interpretation. Yet, since the
questionnaires were seen only by the experimenter and the
subject himself, it seems more plausible to use the ideal
self or experimenter as the choice of "authority”.

From my perspective, then, it seems most reasonable
to select the experimenter as the authority to whom the items
were referred. Having sought empirical evidence for this
contention, I must admit that no such evidence seems to
exist. For the moment, then, let us assume the validity of:
the contention-—hoping that someday empirical investigations
(perhaps through studying the relationship between different
types of attitude scales and the magnitude of experimenter
effect) will substantiate our claim.

To summarize, we have said four things. First, F
Scale items are Type C statements. Second, since they are
Type C, subjects transform them to Type A. Third, the
technique used for transforming the statements was author-
itarian referral. Finally, the authority to whom the
items were referred was the experimenter.

We are now in a position to propose an answer to the
question of what the California F Scale measures. We have
hypothesized that the subjects prefaced each F Scale item
with the phrase, ”The experimenter believes . . .' As far
as the subject himself is concerned, he is testing a hypo-

thesis which phrased as a question will read, "Does the

37
experimenter believe . ... ?" Let us join the subject in
asking this question. Yet, in this case, we shall provide
an answer which is independent of the answer given by the
subjects. Our answer will be based upon introspection and
a reasonable guess about Deutsch and his associates. The
answer based upon introspection is that the experimenter is
a very low F scorer. Our guess is that the same is true of
Deutsch and company. Having independently answered the same
question asked and answered by the subjects, we are in an
interesting situation. We now have two sets of answers to
the same question. If we assume that our own answer is cor-
rect--that the researchers in question are extremely low F
scorers-~we are in possession of a criterion against which
the subjects' F Scale scores may be validated. In short,.
the magnitude of a subject's score may be interpreted as a
deviation from fact. On the basis of the preCeding,.I am
proposing that the magnitude of a subject'sF Scale score
is an index of his awareness of his immediate environment.

In a sense, this conclusion is similar to Christie
and Cook's (1958) belief that F Scale scores are an index
of social sophistication. The answer which I am proposing,.
however, applies to a different realm of data. Given
several samples from pOpulations with different socio-cul-
tural milieux, the mean F for each sample may allow us to
classify samples as more or less socially s0phisticated.
Yet, given any one sample the F for each individual seems

to be more an index of the subject's contact with reality

38
than an index of social SOphistication. Borrowing analysis
of variance terminology, social sophistication is a source
of between groups variance while contact with reality is a
source of within groups variance.

Before proceeding, let me attempt to clarify what I
mean by ”contact with reality". In order to be as precise
as possible, it is necessary to first define "reality" and,
subsequently, specify the process by which an human being-
“contacts" this reality. Neither of these tasks are per-
formed with facility--especially when the performer is
neither a metaphysicist nor an epistemologist. Noting these
limitations, we may make an initial distinction between
self and nonself. By "reality" I mean that portion of the
nonself which exists independently of the substance of self.
Essentially, this is merely a restatement of classical
realism--that self and nonself or I and it or I and Thou
are metaphysically distinct. By "contacting” this reality
I mean the process whereby a self relates to the nonself.
Any specific relationship may be defined along a continuum
ranging from correct to incorrect. For example, show a
subject a horse. Then ask the subject to state the word
conventionally assigned to the object. If he says "horse",
he is correct. If he says ”cow", he is incorrect. He is
even more incorrect, however, if he replies ”my father".

I am suggesting, then, that for a given sample of sub-
jects--in our case, college freshmen--F scores reflect

contact with reality. This means that when a subject

39

refers an item on the F Scale to the experimenter, a cor-
rect response, in general, is “strongly disagree" or
”disagree”. Applying this reasoning to the entire scale,

we may conclude that within any sample which is homogeneously
socially SOphisticated, the lower a subject's F Scale score,
the better is his contact with reality.

The strong assumption being made, of course, is that a
sample of college freshmen is homogeneously socially sophis-
ticated. However, since all subjects in the present exper-
iment had spent eight weeks as students of the experimenter,
it seems that the relevant aspect of the subjects' milieux
(1.9., the experimenter's behavior) was the same for all
subjects. Hence, the assumption of homogeneous social
sophistication seems reasonable. The differences between:
subjects may be attributed, then, to individual differences
in perceiving ”reality“ rather than to differences in the

"reality” itself.

The third hypothesis

Let us now move from the realm of theory back to the
findings of this study. We may recall that the obtained
correlation between F and lack of confidence in authority
was effectively zero. We also saw that to a slight extent,
response bias contributed to this finding. Above and beyond
this bias effect, however, let us reconsider our third
hypothesis in light of our reformulation of the meaning of
F. It may be noted that we have eliminated the assumption--

that F measures attitude toward authority--underlying the

4O
earlier hypothesis. Let us merely note, then, that if F
does assess an individual's contact with reality, for our
subjects contact with reality seems to be unrelated to lack

of confidence in authority.

The second hypothesis

Relevant to the F Scale findings, we know that there
is a positive relationship between F and the tendency to
perceive trustworthiness. We also know that reciprocity as
a genuinely functioning norm is used more by high F scorers
than by low F scorers. In order to explain these findings,
let us again use our new definition of P. On the basis of
this, it can be hypothesized that if the majority of persons
in an environment are trustworthy, then low F scorers in
that environment should evidence a greater tendency to per-
ceive that trustworthiness than do high F scorers. On the
other hand, if most peOple in an environment are untrust-
worthy, low F scorers should tend to perceive untrustworthi-
ness. Consequently, if we define the environment of a
”typical" MSU introductory psychology student as other MSU
introductory psychology students, we need an estimate of
the trustworthiness of these students. If we use our own
data on game behavior, we can estimate that approximately
63 per cent of these MSU students are untrustworthy while
only 37 per cent are trustworthy (see Table 3, p. 21).
Hence, we should expect low F scorers to evidence a tendency
to perceive untrustworthiness--which is precisely what we

found.

41

In attempting to explain the second finding, we encoun-
ter an interesting contradiction. While the data for MSU
subjects indicate that the norm of reciprocity is used
less by low F scorers than by high F scorers, Deutsch's
data for NYU students indicate that the norm of reciprocity
is used relatively 2933 by low F scorers. An initial resolu-
tion of this contradiction is provided by Wrightsman's (1966)
finding of no relationship between either trusting and F or
trustworthiness and F. Hence, it seems that F and game
behavior is certainly not stable across samples.

Let us, however, attempt to be more specific. As a
result of our redefinition of the meaning of Fr we can hy-
pothesize that low F scorers will behave according to a
norm of reciprocity only to the extent that reciprocity is
a functioning norm in the subjects' environment. On the
basis of the empirical evidence, this last statement would
lead us to suspect that the cultural environment of the
MSU subjects is less reciprocal than is the environment of
NYU subjects. Is this reasonable? If we look at, perhaps,
a major difference between these two environments, it does
seem a reasonable contention. We may observe that MSU sub-
jects (primarily freshmen) exist in a world relatively new
to them. They are in a process of forming new friendships
away from the probably stable atmosphere of family and high
school friends. NYU, on the other hand, is an urban univer-
sity.. Certainly, NYU subjects may also be in a process of

forming new friendships, but it seems likely that this

42

process occurs in a more stable atmosphere created by the
presence of family and precollege friendships. Since
stability within a social system implies the existence of
an adequately functioning norm 0? reciprocity (Gouldner,
1963), the contradictory findings for_MSU and NYU may be
only empirically contradictory. Theoretically, both sets
of data may support the same assumptions about the meaning
of F and the relationship between reciprocity and social
system stability.

The contentions of the preceding paragraph are, of
course, guesses. Consequently, let us return to the haven
of the data available to us. If our guessing has been ac-
curate, our data should show a greater prevalence of recip-
rocal behavior among NYU subjects than among MSU subjects.
Table 3 (p. 21) indicates that while 44 per cent of the NYU
sample behaved reciprocally, only 29 per cent of the MSU
sample did so. We seem, therefore, to be in a position of
attributing the differences between the two studies under
consideration to the differences in the environments in

which the subjects of those studies exist.

The first hypothesis

Relevant to the relationship between game playing be-
havior and Authority Figures scores, there are three statis-
tically significant correlations. First, it seems that
subjects who behave according to a norm of reciprocity have
less confidence in authority figures than do subjects who

are nonreciprocal. Second, subjects who exhibit trusting

43
behavior tend to have less confidence in authority figures
than do suspicious subjects. Third, subjects who perceive
others as trusting tend to have less confidence in authority
than do subjects who perceive others as suspicious. Two
other findings which are relevant to the following discus-
sion are (1) that subjects who trusted tended to perceive
others as trusting and subjects who were suspicious tended
to perceive others as suspicious. Also, (2) subjects tended
to be trusting if others were perceived as trustworthy but‘
to be suspicious if others were perceived as untrustworthy.

Before presenting a model of these findings, it is
necessary to justify the dismissal of one finding. I have
decided that the point biserial correlation between recipro-
city and Authority Figures scores is deceptive. Although
statistically significant its significance is exclusively a
result of computing the statistic on a subsample of the
entire sample. The deceptiveness of this statistic becomes
apparent if we look at the mean Authority Figures scores
for each category of game playing behavior. The means for
the two possible trusting categories (i,g.,.trusting and
trustworthy or trusting and untrustworthy) are virtually
identical--2.77 and 2.76, respectively. For the suspicious
categories, the untrustworthy subjects have a higher mean
(2.53) than do the trustworthy (2.38)., In other words, it
seems that Authority Figures correlates with reciprocity
only because trust correlates with Authority Figures.

Having dismissed this point biserial, we may build a

44
relatively simple model of trusting behavior. The basic
premise of this model is that the probability of a subject
exhibiting trust is a function of three variables: percep-
tion of past, perception of present and perception of future.
Corresponding to each of these perceptions are three empir-
ical measures. To the perception of past, corresponds a
subject's lack of confidence in authority. To the perception
of present, corresponds perception of others as trusting or
suspicious. Finally, corresponding to perception of future
is Deutsch's expectancy index--the perception of others
as trustworthy or untrustworthy. A formal statement of the
nature of each of these as they relate to game behavior
would be premature. Yet, from the data of this study,.it

does appear that each plays a significant role.

COnclusion
It seems tenable to conclude that the variables dis-
cussed at the beginning of this paper are valuable tools
for investigating game behavior. The hypothesized relation-
ships between these variables, however, seem to be untenable.
There is one other result of this study which seems to
merit discussion. This result is the virtual absence of
sex differences. The nonsignificant sex difference in game
behavior seems to be a typical finding for ”prisoner's
dilemma” studies similar to the one presented in this paper
(Gallo & McClintock, 1965: Kanouse & Wiest, 1967; RapOport
& Chammah, 1965). The findings for the Authority Figures

and F Scales, on the other hand, may be dependent upon the

45
specific sample of subjects.

Finally, it should be noted that the explanation sug-
gested for the findings of this study is only one of many
plausible explanations. For example, it is possible that
F does not measure ”contact with reality”. In selecting
this interpretation, I attempted to steer a middle course
between the feuding "authoritarian" versus "response
set” researchers--a course which seemed reasonable and con-
sistent with both my own data and selected data of other
researchers. Other points of departure for divergent ex-
planations surely ought to include the questionable validity
of the Authority Figures Scale and possible differences
between my paper-and-pencil form of the "prisoner's dilemma”
and Deutsch's ”behavioral" form. In short, the validity of‘
my explanation is heavily dependent upon the validity of a
large number of assumptions. The fruitfulness of this ex-
planation, then, must await the collection of data for

testing these assumptions.

REFERENCES

Adorno, T. W., Frenkel-Brunswik, Else, Levinson, D. J., &
Sanford, R. N. The authoritarian personality. Science
Editions. New York: John Wiley_& Sons, Inc., 1964.

Alexander, F. Fundamentals g: psychoanalysis. New York:
W. W. Norton & Company, Inc.,.1948.

 

Alexander, F., & Staub, H. The criminal, the judge, and
the public: A psychological analysis, trans. G. Zil-
boorg. Glencoe, Illinois: The Free Press, 1956.

Baggaley, A. R. Intermediate gorrelational methods. New
York: John Wiley & Sons, Inc.,.1964.

Christie, R., & Cook, P. A guide to published literature
relating to the authoritarian personality through

1956.. l. Psychol., 1958, 4 , 171-199.

Cronbach, L. J. Coefficient alpha and the internal struc-
ture of tests. Psychometrika, 1951, 16, 297-334.

Deutsch,.M. Trust, trustworthiness, and the F Scale. g.
abnorm. soc. Psychol., 1960, 61, 138-140.

Gallo, P. S.,.& McClintock, C. G. Cooperative and competi-
tive behavior in mixed-motive games. i. Conflict
Resolution, 1965, 2, 68-78.

Gouldner, A. W. The norm of reciprocity: A preliminary
statement. In E. P. Hollander & R. G. Hunt (Eds.),
Current perspectives 13 social psychology. New York:
Oxford University Press, 1963. Pp. 269-280.

Hays, W. L. Statistics for psychologists. New York: Holt,
Rinehart and Winston, 1963.

Kanouse, D. E., & Wiest, W. M. Some factors affecting
choice in the Prisoner's Dilemma. i. Conflict Resolu-
tion, 196?, 11, 206-213.

McBride, L.,.& Moran, G. Double agreement as a function of
item ambiguity and susceptibility to demand implica-
tions of the psychological situation. g, Pers. soc.

Psychol., 1967, 6, 115-118.

46

4?

McNemar, Q. Psychological statistics. (3rd ed.) New
York: John Wiley & Sons, Inc., 1962.

Martindale, D. The nature and pypes g: sociological theory.
Boston: Houghton Mifflin Company, 1960.

Mills, C. W. The sociolo ical imagination. New York:
Grove Press, Inc., 1961.

Parsons, T. The social system. New York: The Free Press,
1951. (1964

Parsons,.T.. Social structure and personality. Glencoe,
Illinois: The Free Press, 1964..

Peabody,.D. Authoritarianism scales and response bias.
Psych..Bull., 1966, 6 , 11-23.

Rapoport, A. Two person game theory: The essential ideas.
Ann Arbor Science Paperbacks. Ann Arbor, Michigan:
The University of Michigan Press, 1966.

Rapoport, A., & Chammah, A. M. Sex differences in factors
contributing to the level of cooperation in the Prison-
er's Dilemma game. g, Pers. soc. Psychol., 1965, g,
831-838.

Rudin,,S. A. The relationship between rational and irration-
al authoritarianism. g. Psychol.,.1961, 52, 179-183.

Wrightsman, L. S. Personality and attitudinal correlates
of trusting and trustworthy behaviors in a two-person
game. 1, Pers. soc. Psychol., 1966, 4, 328-332.

APPENDIX A: RECIPROCITY AND AUTHORITY

The following discussion may be substituted for the
"Introduction" to the preceding paper.

In this study three hypotheses were formulated.
Hypothesis 3: There exists a negative correlation between~
lack of confidence in authority figures and willingness to
participate in a reciprocal relationship.

Hypothesis g: There exists a negative correlation between
California F Scale scores and willingness to participate
in a reciprocal relationship.

Hypothesis 1: There exists a correlation (either positive
or negative) between California F Scale scores and lack of
confidence in authority figures.

The rationale for these hypotheses is both theoretical
and empirical. Hypothesis 1 is based upon the assumption‘
that if an individual lacks confidence in authority figures,
he does so because his past experience in authority relation-
ships has led him to conclude that he gains no gratifica-
tion and perhaps loses gratification by entering such rela-
tionships. The first hypothesis states then that if an
individual has been unsuccessful in obtaining gratification:
in authority relationships,.through a generalization process
he will avoid participating in any relationship similar to

an authority relationship. The similarity between authority
48‘

#9
and reciprocal relationships is twofold. First, the in-
dividual who initiates either relationship places control
of the social situation in the hands of another member of
the group» Second, the initiator of the relationship has
control over the other member only if a social norm of
"fair play" is functional. In an authority relationship
this norm may be called a ”social contract”. In a recip-
rocal relationship, the norm may be called a ”norm of
reciprocity".

The second hypothesis is based upon findings in a
study by Deutsch (1960).

The third hypothesis is based upon the supposed
validity criteria of the California F Scale and the scale
(constructed for the present study) measuring lack of con-
fidence in authority figures. Theoretically,.this is a
crude hypothesis. It simply states that if both of these
scales measure attitudes toward authority, they should be

related to one another.

APPENDIX B: QUESTIONNAIRES

PUBLIC OPINION QUESTIONNAIRE

The following statements refer to opinions regarding
a number of topics about which some people agree and
others disagree. Please mark each statement in the left.
hand margin according to your agreement or disagreement,
as follows:

+1: Tend to agree -1: Tend to disagree
+2: Agree -2: Disagree '
+3: Strongly agree -3: Strongly disagree

1.. SOme day it will probably be shown that astrology:
can explain a lot of things.

2. Nowadays when so many different kinds of people
move around and mix together so much,.a person
has to protect himself especially carefully
against catching an infection or disease from‘
them.

3. The typical father rewards his children for good
behavior. .

4. Obedience and respect for authority are the most
important virtues children should learn.

5. Every person should have complete faith in some
supernatural power whose decisions he obeys
without question.

6. Most employers underpay their employees.

7. No weakness or difficulty can hold us back if we
have enough will power.

8. Most politicians would readily vote against the
best interests of their electors.

9. Nobody ever learned anything really important
except through suffering.

10. Policemen are usually fair and just officers of

the 13W.
50

11.

12.

13.

14.

15c

16.

17.

18.

19.

20.

21.

22.

23.

 

24.

25.

51

Clergymen are usually more concerned about the
contents of the Sunday collection basket than
the contents of their parishioners' souls.

The typical father does nothing to merit respect
from his children.

Parents usually punish their children only when
punishment is necessary for the child's welfare.

What the youth needs most is strict discipline,
rugged determination, and the will to work and
fight for family and country.

Sex crimes, such as rape and attacks on children,
deserve more than mere imprisonment; such crimi-
nals ought to be publicly whipped, or worse.

The average policeman misuses his authority.

What this country needs most, more than laws and
political programs, is a few courageous, tireless,
devoted leaders in whom the people can put their
faith.

There is hardly anything lower than a person who
does not feel a great love, gratitude, and respect
for his parents.

Most teachers are fair and just graders.

Most teachers would rather be popular with their
students than satisfy their students' educational
needs..

Homosexuals are hardly better than criminals and
ought to be severely punished.

PeOple can be divided into two distinct classes:
the weak and the strong.

Mothers generally strive more to love than to be
loved.

When a person has a problem or worry, it is best
for him not to think about it, but to keep busy
with more cheerful things.

The wild sex life of the old Greeks and Romans
was tame compared to some of the goings-on in
this country, even in places where people might
least expect it.

26.

 

27.

 

28.

 

40.

52

When the probability of being discovered is
small, the average politician uses taxes primar-
ily for his own benefit..

Young people sometimes get rebellious ideas, but
as they grow up they ought to get over them and
settle down.

Science has its place, but there are many impor-
tant things that can never possibly be understood
by the human mind.

Most peOple don't realize how much our lives are
controlled by plots hatched in secret places.

Some people are born with an urge to jump from
high places.

Most of our social problems would be solved if we
could somehow get rid of the immoral, crooked and
feebleminded people.

Most children are well cared for by their mothers.

The average courtroom judge is more interested in
re-election than in being just..

Nowadays more and more people are prying into
matters that should remain personal and private.

Parents behave as if it were better to punish a
child than to love him.

If people would talk less and work more, every-
body would be better off.

No sane, normal, decent person could ever think
of hurting a close friend or relative.

The business man and the manufacturer are much
more important to society than the artist and
the professor.

A person who has bad manners, habits, and breed-
ing can hardly eXPeet to get along with decent
people.

Wars and social troubles may Someday be ended by
an earthquake or flood that will destroy the
whole world.

53

41. Human nature being what it is, there will always
be war and conflict.

42. It is best to use some prewar authorities in Ger-
many to keep order and prevent chaos.

43. Most employers overwork their employees.
44. An insult to our honor should always be punished.

45. Familiarity breeds contempt.

54

SOCIAL INSIGHT TEST

The following is a test of your ability to predict
human behavior. Read the following story carefully and
answer the questions at the end to the best of your
ability.

Three wealthy young men, Tom,.Dick and Harry, had just-
been introduced to one another. Having discovered during
the course of their conversation that they shared an inter-
est in gambling, they decided to participate in a game of
chance about which Tom had heard from a friend.

The game is designed for three people, Player I, Player
II and a Spectator. By drawing straws it was determined
that Tom would be Player I, Dick would be Player II and
Harry would be the Spectator. Tom as Player I made the
first play. Tom had in his possession a key which would
unlock a box in Dick's possession. The first play consis-
ted of Tom either giving Dick the key or not giving Dick
the key 0

Dick as Player II made the second play. Dick possessed
a glass of water. The second play consisted of Dick either
giving or not giving Tom the glass of water. The game then
ended.

In other words, four possible events could have oc-
curred:

(1) Tom gave Dick the key and Dick gave Tom the glass
of water.

(2) Tom gave Dick the key but Dick did p93 give Tom
the glass of water.

(3) Tom did g9; give Dick the key but Dick did give
Tom the glass of water.

(4) Tom did 393 give Dick the key and Dick did 223
give Tom the glass of water.

In order to make the game interesting, a system of
payoffs was devised. If event (1) occurred Harry payed Tom
and Dick $500 apiece. If event (2) occurred Tom payed Dick
$1000. If event (3) occurred Dick payed Tom 31000. If
event (4) occurred,.Tom payed Harry $500 and Dick payed
Harry $500.

55

This may be schematically represented as follows:

EVENT WINS AND LOSSES OF PLAYERS
Tom Dick Harry
(1) Tom gave Dick the key: Win Win Lose
Dick gave Tom the water. #500 $500 $1000
(2) Tom gave Dick the key; Lose Win Neither win
Dick did not give Tom the water$1000 $1000 nor lose
(3) Tom did not give Dick the key; Win Lose Neither win
Dick gave Tom the water. $1000 $1000 nor lose

(4) Tom did not give Dick the key; Lose Lose Win
Dick did not give Tom the water$500 $500 $1000

Question #1: Tom made the first play. What did he probably

do? Circle either (a) or (b).
(a) Tom gave Dick the key.
(b) Tom did not give Dick the key.

Question #2: If Tom gave Dick the key, what did Dick do?
Circle either (a) or (b).

(a) Dick gave Tom the glass of water.

(b) Dick did not give Tom the glass of water.

Question #3: If Tom did not give Dick the key, what did
Dick do? Circle either (a) or (b).

(a) Dick gave Tom the glass of water.)

(b) Dick did not give Tom the glass of water.

Question #4: If you were Tom what would you do? Circle
either (a) or (b).

(a) I would give Dick the key.

(b) I would not give Dick the key.

Question #5: If you were Dick and Tom had given you the
key, what would you do? Circle either (a) or (b).

(a) I would give Tom the glass of water.

(b) I would not give Tom the glass of water.

Question #6: If you were Dick and Tom had not given you
the key, what would you do? Circle either (a) or (b).
(a) I would give Tom the glass of water.
(b) I would not give Tom the glass of water.

"k ..th

ICHIGRN STQTE UNIV. LIBRRRI IES

| 2lllll ”758