A new scale for the measurement of

interpersonal trust^
Julian B. Rotter, University of Connecticut

One of the most salient factors m the effecdveness of our

present complex social organization is the willmgness of one or
more mdividuals m a social unit to trust others The efficiency,
adjustment, and even survival of any social group depends upon
the presence or absence of such trust.
Interpersonal tmst is defined here as an expectancy held by
an mdividual or a group that the word, promise, verbal or wntten
statement of another individual or group can be relied upon.
This definition clearly departs significandy from Enkson's (1953)
broad use of the concept of basic trust which Enkson descnbes
as a central ingredient m "the healthy personality "
Vanous wnters have already mdicated that a high expectancy
that others can be relied upon is an important variable in the
development of adequate family relationships and of healthy
personahties in children The fadure to tmst others, particularly
representatives of society, such as parents, teachers, and powerful
commumty leaders, has frequently been cited as an important
determmant m delmquency (Redl & Wmeman, 1951) Diffi-
culdes m race relationships and m mmonty group-majonty group
relationships have, likewise, been frequently related to expec-
tancies of one group that the verbal statements of the other
cannot be accepted Many psychotherapists believe interpersonal
tmst is a major determinant in the success of psychotherapy In
fact, an expectancy that others can be believed must be an
important variable m human learmng in general Much of the
formal and informal leaming that human bemgs acquire is based
on the verbal and wntten statements of others, and what they
leam must be significandy affected by the degree to which they
believe their informants without independent evidence.
1 This mvesbgabon was supported by a grant from the Nabonal Insbtute
of Mental Health (MH 11455)
 652 Julian B Rotter

It seems evident that an adequate measure of individual dif-

ferences m mterpersonal trust would be of great value for research
m the areas of social psychology, personahty, and chnical psy-
chology Social scientists have mvestigated some of the condi-
tions relatmg to mterpersonal trust usmg game theory (Deutsch,
1958, i960, Rapaport & Orwant, 1962, and Scodel, 1962). For
the most part these mvestigations have shown that a typical
reaction of two strangers m a two-person non-zero-sum game
situation mvolving trust produces behavior usually mdicative
of competitive rather than cooperative attitudes One might
conclude that Americans at least are a highly suspicious and
extremely competitive group who would give up many benefits
rather than cooperate with someone else. The results of these
studies, however, do not seem consistent with a common-sense
analysis of our own society From the family unit to big busmess,
cooperation seems to mark the everyday behavior of mdividuals
and organizations to a far greater degree than would be antici-
pated from the study of two-person game situations Perhaps
this is the result of special reactions to these laboratory situations
which are highly competitive in nature and are specific to these
situations, or at least have limited generahty. The wnter has
previously published an analysis of some of the factors involvmg
specificity of reaction to test and experimental laboratory situa-
tions which may be apphcable here (Rotter, 1955, i960).
Studies involvmg the commumcation of mformation (Mel-
linger, 1956, Loomis, 1959, Kelley & Rmg, 1961) have several
characteristics in common with game approaches but present
situations somewhat closer to the present study. These investiga-
tions mdicate that people who trust others more are also more
trustworthy, or cooperative. Similar findmgs were obtamed by
Deutsch (i960) using the "game" paradigm
Other recent literature has dedt with trust indirectly Dis-
cussions of Machiavellianism, ie,, the tendency to manipulate
others to gam one's own ends (Christie & Merton, 1958), and
anomie (Merton, 1949), suggest that, at least m part, distrust of
others is dependent upon normlessness m the social organization.
The problem of trust in the present research is bemg viewed
from the perspective of social learning theory (Rotter, 1954).
Measurement of interpersonal trust 653
From this onentation, choice behavior m specific situations de-
pends upon the expectancy that a given behavior will lead to
a particular outcome or remforcement m that situation and the
preference value of that reinforcement for the mdividual m that
situation
It IS a natural lmphcation of social leammg theory that ex-
periences of promised negative or positive reinforcements occur-
rmg would vary for different mdividuals and that, consequendy,
people would develop different expectancies that such rem-
forcements would occur when promised by other people It is
also natural to expect, to some degree, that such expectancies
that promises of other social agents will be kept would generahze
from one social agent to another That is, mdividuals would
differ m a generalized expectancy that the oral or wntten state-
ments of other people can be rehed upon The development of
such a generalized attitude may be leamed direcdy from the
behavior of parents, teachers, peers, etc., and also from verbal
statements regardmg others made by significant people or trusted
sources of commumcation such as newspapers and television.
It IS ironic that we can leam to distrust large groups of people
without personal expenence vahdating such distrust, because
people who are themselves tmsted teach distrust
Previous work on the choice of a smaller immediate reward
versus a more highly valued, delayed reward by Mahrer (1956)
and Mischel (1961a, 1961b) is related to the concept of trust
as defined here These studies sti-ongly suggest that children
who have experienced a higher proportion of promises kept by
parents and authonty figures m the past have a higher generalized
expectancy for mterpersonal trust from other authonty figures

CONSTRUCnON OF THE INTERPERSONAL TBUST SCALE

As a first step m die constiriction of the scale a number of items
were wntten using a Likert format An attempt was made to saniple
a wide vanety of social objects so that a subject would be called
upon to express his trust of parents, teachers, physicians, pohtiaans,
classmates, fnends, etc. In other words, the scale was constiiicted
as an addtttve scale in which a high score would show trust for a
great vanety of social objects In addition to the specific items, a
few Items were stated m broader terms presumed to measure a more
general optimism regardmg the society Fmally, a number of iiUer
 654 Julion B Rotter
items, intended to partially disguise the purpose of the scale, were
wntten and mcluded in the first expenmental form
The expenmental form was group-administered to two large classes
of students m an mtroductory psychology course The sample com-
pnsed 248 male and 299 female subjects Along with this scale the
Marlowe-Crowne Social Desirability Scale (1964) of "need for social
approval" was administered
Three cntena were used for mclusion of an item in the final scale
(a) the item had to have a sigmficant correlation with the total of the
other trust items with that item removed, (b) the item had to have
a relatively low correlation with the Marlovs^e-Crowne Social De-
su-abdity Scale score, and (c) endorsement of the item showed rea-
sonable spread over the five Likert categones of (1) strongly agree,
(2) mildly agree, (3) agree and disagree equally, (4) mildly dis-
agree and (5) strongly disagree A final form of the scale was de-
termined by dropping three items from the a pnon scale
In the experimental form of the test, half of the crucial items were
wntten so that an "agree" response would mdicate trust, and half
so that a "disagree" response would mdicate trust In the final form
of this scale the items selected were similarly balanced so that 12
mdicated trust for agreemg and 13 distrust for agreeing Filler items
did not show significant relationships to the trust items but helped
partially obscure the purpose of the test The final form of the
test mcluded 25 items measurmg trust and 15 filler items Some
sample items are presented below
In dealing with strangers one is better oflF to be cautious
until they have provided evidence that they are trustworthy
Parents usually can be rehed upon to keep their promises
Parents and teachers are bkely to say what they beheve
themselves and not just what they think is good for the
child to hear.
Most elected pubhc oflScials are really smcere m their

In addition to the Marlowe-Crowne Social Desirabihty Scale, the

547 subjects completed a personal information questionnaire which
mcluded information on age, soaoeconomic level, father's occupation,
father's and mother's religion and place of birth, and siblmgs, so that
position in the family and family size could be determmed. College
Aptitude scores were also available on these subjects and were ob-
tamed directly from the students' admission records At later dates
several of the students were subjects m other studies mvolvmg the
administration of the same trust scale It was therefore possible to
obtam test-retest rehabihties for long penods of time, where the
 Measurement of interpersonal trust 655

testmg conditions were different for the two administrations. An

analysis of these data is presented below.

TEST CHARACTERISTICS

Internal Consistency and Test-Retest Reliability

Table x below provides means and standard deviations of the
248 male and 299 female college student subjects Internal con-
sistency based on spht-half reliability, corrected by the Spearman-
Brown formula, are also provided While these consistencies are
not high for objective type tests, it should be remembered that
these are additive scales samplmg a vanety of different social
objects rather than a measure of mtensity lmuted to a narrow
area of behavior. Regarded m this hght these internal consis-
tencies are reasonably high The difference m mean scores for
males and females is not statistically significant and distributions
of scores for both sexes are similar.
Table 1 Test data for the Interpersonal Trust Scale,
Group N Mean SD

Males 248 73 01 1023

Females 299 71 91 9 95
Total 547 72 41 1090

Split half reliability*

N r p

Males 248 77 <001

299 75 <001

Total 547 76 <001

'Corrected by the Spearman-Brown Prophecy Formula

Two estimates of test-retest rehabihty are available The first

of these involves 24 subjects, 10 male and 14 female, who took
the test ongmally in a large group-testmg situation and repeated
the test as part of a sociometnc study to be descnbed later. The
average length of time between first and second tests was ap-
proximately seven months. The conelation was ,56, (p < 01),
The second measure of test-retest rehabihty was obtained on
students who had also taken the test ongmally m a large group
situation Then- second test was part of an experiment m which
656 Julian B. Rotter
the tmst scale was given in groups of from two to 13 with two
other tests appearmg equally often m first, second, and third
positions m order of admmistration. There were 34 males and
eight females m tbis group, and the approximate average tune
between tests was three months The correlation was 68, (p <
.01). Considermg the important differences m administration
procedures and the relatively long penods of time between tests,
these test-retest coefficients mdicate surpnsing stability of test

For the 248 male subjects the correlation with the Marlowe-
Crowne S-D Scale was .21; for the 299 females, .38 The overall
correlation was 29. All correladons were stadstically significant.
These results suggest that trust is regarded as a socially desirable
trait but that the total amoimt of variance m the trust scale ac-
counted for by the social approval modve is reladvely small To
determine the relationship, if any, with general abihty, 100 male
and 100 female subj'ects were selected at random and their
tmst scores correlated with the college entrance (SAT) scores.
The correlation for the 100 females was — i6 and for the 100 males
— 06. At least for this sample of college students, abihty has no
sigmficant mfluence on tmst scale scores

Demographic Characteristics of High and Low Trust Individuals

From the personal information sheet filled out by all of the
547 students who took the Interpersonal Tmst Scale, analyses
of variance were computed for the variables of ordmal position,
family size, religion, socioeconomic status, age, and number of
semesters in college. In addidon, the subjects were grouped
mto two categories based on whether or not the reported religions
for both parents were the same or different. Since male and
female subjects were essentially similar throughout this analysis,
data were combined for the sexes.
There were no significant differences m test scores for subjects
of different ages or according to the number of semesters of
college attended However, the range for both of these vanables
was extremely narrow The data of family size were dichotomized
into three children or less, or more than three children Students
from larger families did not show sigmficandy different trust
Measurement of interpersonal trust

Table 2. Demographic data for the Interpersonal Trust Scale,

Significant differences among groups

Variable and group N Mean (p < 05)

Ordinai position
Oniy 52 73 08
Oldest 195 72 21
Middle 129 73 02 Youngest
Youngest 171 71 97 Middie

ANOV over all F=37^1, p < 05

iteiigion
Left biank 12 71 50 Jewish, none

Jewish 85 74 65 All groups except Miscellaneous

Protestant 197 73 31 Jewish, Catholic, none

Cathoiic 203 71 33 Jewish, Protestant, none. Miscellaneous

Misceilaneoui 17 73 82
c, atheist 33 67 48 All groups

ANOV overall F = 4 6 7 8 , p < 001

«eligious Differences

ill F = 7 5 06, p < 001

Socioeconomic Ievei
N Mean groups ( p < 05)

25 73 48 V
Wamer Group i 117 73 45 iil,V
Warner Group 11 150 72 70 V
91 71 81 1
Warner Group IV 64 72 48
Warner Group V 100 70 97 No information, 1,11

ANOV overall F = 8 63, p < 01

scores from those with three or less children Significant dif-

ferences were obtained on all the other vanables.
Table 2 presents mean scores for the vanous breakdowns of
subjects for the vanables of ordmal position, rehgion, religious
differences, and socioeconomic status A multiple-comparisons
test for a single degree of freedom contrast (Myers, 1966) was
made to determme differences among means It should be noted
that m many cases there are significant differences because of
658 Julian B Rotter
the large number of subjects m the sample, but actual mean
differences are relatively small
Inspection of the findmgs for birth order reveals one signifi-^
cant difference, but the actual mean differences are suflBciently
small to suggest no important psychological vanability in this
group However, m a separate smaller sample, Geller (1966)
also found youngest children to be the least trustmg and sig-
nificantly dierent from all other ordmal positions While this
findmg cannot be mterpreted without additional data, it is
possible that the youngest child has less interaction with his
parents and has the least acceptance of the adult mterpretation
of the venties of our society
The data on rehgion are more clear-cut Students who fill
out the blank by stating any rehgion tend to be more trusting
than those who state they are agnostics, atheists, or who write
"none " Smce it is clear that such students are already expressmg
less faith m one currently accepted mstitution, it is not surpnsmg
that they show a generalized lower degree of trust m others.
Perhaps most mterestmg are the lower tirust scores for sub-
jects with religious differences between parents In any case
where the student mdicated a religious difference for the
two parents, mcludmg one parent bemg atheistic and the other
not, the subject was put mto the rehgious difference cate-
gory This group mcludes all subjects who listed different
rehgions for the parents, regardless of the religion stated for the
subject himself Only nme of the 100 subjects so classified gave
their own rehgion as "none," "atheist," or "agnostic," so that this
group of subjects has only a mmimum overlap with the students
who were classified as nonrehgious m the previous analysis It
seems reasonable that a child subjected to two different kmds of
adult mterpretations m such an important area as rehgion would
grow up to be more cymcal of the verbal communications of
authonty figures
Fmally, the data on socioeconomic status more or less fol-
low the expected progression for more trust at the highest eco-
nomic level to less trust at the lowest economic level. For this
analysis, subjects were classified accordmg to Wamer's system
based upon father's occupation. The interpretation agam seems
 Measurement of interpersonal trust 659
to be consistent with the general notion that those students who
had least reason to accept the status quo as defined and defended
by the authonties m the social system tended to show the least
trust of those authonties It should be reiterated here, though,
that the differences are agam small and the overlap among groups
is very great

OF SCALE
In order to assess the vahdity of the Interpersonal Trust Scale
It would be optimal to obtam one or more natural life cntenon
situations The two-person non-zero-sum game seems like a
face-valid procedure to mvestigate mterpersonal trust However,
the results of these studies suggest that the situation is reacted to
by many if not all subjects as a competitive game, often regard-
less of special mstructions For the reasons cited earher it was
decided to test the validity of the scale agamst observations of
everyday behavior by a sociometric techmque Two fraternities
(N = 35, N = 38) and two soronties (N = 41, N = 42) on the
Umversity of Connecticut campus were asked to cooperate m
the study Lump sum payments were provided to each of the
four organizations if they could promise that all members would
be available for a smgle evenmg and all would agree to take
the sociometnc ratmg of trust and two bnef tests. However,
members would only be used m the study if they had hved m
the house for a penod of at least six months pnor to the date of
testmg The data was collected by the author and, m each case,
a research assistant of the same sex as the subjects ^ In addition
to askmg the subjects to nommate members of the group who
were the highest and lowest m mterpersonal trust, subjects were
also asked to nommate others high and low on dependency,
guUibihty, and trustworthmess As control variables, scales were
^ o mcluded for humor, populanty, and fnendship Fmally the
subjects were asked to make a self-ratmg of trust on a four-point
scale of (1) much more than the average college student, (2)
more than the average college student, (3) less than the average
college student and (4) much less than the average college
student
2 Grateful acknowledgment is made to Ray Muhy and Lmda Yuccas who
assisted in this research
 To avoid halo effect, elaborate instmcbons were given asking
each subject to pay special attention to the different character-
isbcs required for each sociometnc descnpbon Confidentiality
was assured as well as the fact that we were not mterested m
mdividuals and that we would elumnate the use of names as
soon as the data were obtained, substitutmg numbers for each
individual. To avoid stereotypmg, no labels were used for the
sociometnc scales, but rather descriptions of typical behaviors.
In each group the order of presentabon was first the trust scale,
second the sociometnc scales, and last the Marlowe-Crowne
Social Desn-ability Scale. However, the Marlowe-Crowne S-D
Scale was not given to the first group
One other difference occurred in the procedure for the first of
the four groups In this group, a soronty, each subject was
asked to nominate the five highest and five lowest persons on
each sociometnc scale These data were analyzed usmg four
methods of sconng In the first method the highest was weighted
5, the next highest 4, the next 3, and so on, and the negative
nommabons were similarly weighted —5, —4, —3, etc The sec-
ond was also a weightmg method but using only the first three
nommabons for the negabve and posibve ends of each scale
The third method involved no weightmg but gave a score of -f 1
for each mention, ublizmg all five nommabons The last method
gave a score of + 1 only for the top three and the bottom three
nommabons Intercorrelabons of the four methods mdicated no
substantial differences among them Smce subjects met difiB-
culty m findmg five names for the top and bottom of each scale,
subsequent groups were asked to nommate only the top three
and bottom three. Each menbon was then scored either ± 1 to
give an overall score on that scale
Instmcbons for the sociometric rating of trust are given below
as are the descriptions for the tmst variable. The order of
presentation was (1) dependency, (2) tmst, (3) humor, (4)
gulhbility, (5) trustworthmess, (6) popularity, (7) friendship,
and (8) self-ratmg of trust Dunng the administrabon of the
sociometnc scale a strong attempt to keep a serious atmosphere
was more or less successful, success bemg greater in the soron-
ties than in the fratemibes
Measurement of mterpersonal trust 661
Sociometric Instructions
On the followmg pages you will be asked to nommate
some people m your group who fit vanous descnptions
Please do so as thoughtfully as possible, paymg special at-
tention to the different characteristics called for m each
descnption Agam let me assure you the results are confi-
dential and we have no mterest m you as mdividuals The
data from these questionnaires will be placed on IBM cards
identified only by numbers, not names
On the next seven pages vanous kinds of people will be
descnbed Place the name of the person who most closely
fits the descnption after the ( l ) , next most closely after
the ( 2 ) , and so on until you have listed the three people m
the group who most closely fit the descnption List only the
names of people who are here in the group now Do not
list any members who are not present
Do each page m order Do not look at the page ahead
until you have finished the one you are workmg on You
may wish as you go along to use some of the same names on
different descnptions.
You may find the task difScult but we hope you will take
it senousiy and do the best you can We feel we are domg
important research and hope you wiU cooperate with us to
the fullest
Description of Trust Variable
This person expects others to be honest. She is not sus-
picious of other people's mtentions, she expects others to be
open and that they can be rehed upon to do what they say
they will do
This person is cymcal She thinks other people are out
to get as much as they can for themselves She has htde
faith m human nature and m the promises or statements of
other people.
T h e correlations to b e reported below are combined for t h e
four groups. They were obtamed by calculatmg separate cor-
reladons for each group, transforming to z scores, finding the
average z score, and then transforming to an f for the entire
group. Before testmg the vahdity of the trust scale against the
sociometric scale it was necessary to determme whether or
not the sociometnc ratings were reliable. This was done by
dividmg each group mto random, equivalent halves and ob-
taming the sociometric score on each vanable for each person
662 Julian B, Rotter
Table 3 Spht-half reliabiLties of sociometnc scores, combmed groups

Variable

Dependency 88

Humor 93
Gullibility 93
Trustworthiness 89
Popularity 95
82

•^=21 forp < 01

Table 4 Combined mtercorrelations of sociometnc and test scores,

combined groups (N =: 156)
Varioble 10*

1 interpersonoi Trust Scoie - 2 3 37 09 - 0 3 31 20 19 29 13

, Sociometric Dependency - 0 7 - 3 6 78 - 4 5 - 4 6 - 5 3 - 0 6 -05
Sociometric Trust 34 13 62 43 42 39 02
Sociometric Humor - 3 3 26 61 66 14 -0«
; Sociometric Guiiibility - 2 4 - 4 3 - 6 0 01 0)
oociomctric Tru$Twortnin6$s 57 50 24 01
' Sociometric Popuiarity 83 05

} Self-Rating of Trust •31

10 Mariowe-Crowne S-D Scoie*

>rrelations involving the S-D Scale

r = 21 forp < 01 (N = 156)
r = 16 forp < 0 5 ( N = 156)
r = 18 forp < 05(N==114)

m the two subgroups The resultmg correlations shown m Table

3 mdicate the degree to which the members of the group are
likely to see each other m a similar way It can be seen that the
conelations are unusually high, suggestmg not only good co-
operation but also that the members of the groups were basmg
their ratmgs on a common core of observations
The mtercorrelations for the 10 vanables are presented m
Table 4 This mcludes the Trust Scale, the seven sociometric
ratmgs, the self-ratmg of trust, and the Social Desirabihty Scale
It can be seen from Table 4 that the Interpersonal Trust Scale
was significantly related to the sociometnc trust score Indi-
vidual correlations m the four groups range from 23 to 55, The
overall correlation of 37 is significantly higher than that for the
 Measurement of interpersonal trust 663
control vanables of humor, populanty, and friendship, mdicatmg
that the sociometnc ratmg for trust was measurmg an mde-
pendent vanable and was not merely the result of halo effect
Both the tmst scale and the sociometiic ratmg of trust conelated
sigmficandy vnth tiustworthmess, providmg strong support for
the belief that people who tmst others are regarded themselves
as bemg dependable
It IS of considerable mterest that no significant relationship
was found between gullibihty which was defined on the socio-
metnc scale as "naive and easily fooled m contrast to sophisti-
cated, expenenced, etc," and tmst as measured by the socio-
metric scale or the Interpersonal Tmst Scale While it is some-
what difficult conceptually to entirely separate gullibihty from
mterpersonal tmst, it is clear that m practice the mdividuals m
our sample made such separation and saw the two traits as
mdependent.
The other significant relationships with the trust scale were
for the self-rating of trust and the negative relationship with
dependency. The trustmg individual is seen as less dependent on
others (makmg decisions, seekmg advice and help) than the
mdividual rated as low in trust But dependency is seen as a
clearly negative trait correlatmg — 46 with popularity and — 53
with fnendship Some of this relationship may be negative halo
smce it IS clear that there is a significant although quite low
positive relationship between trust and fnendship and popular-
ity The correlation between self-ratmg of trust and the socio-
metric rating of trust (39) is also mdicative of the cooperation
and senousness with which the subjects completed the socio-
metnc task It may be surpnsmg to some that the self-rating
showed a relatively high relationship ( 39) with the ratmg of
trust made by others It should be remembered, however, that
the self-ratmg came at the end of the sociometnc task and all of
the subjects knew that they were bemg rated on the same trait
by others, providmg pressure on them toward honesty Similarly,
the relationship between the trust scale and the self-ratmg of
trust ( 29) might not have been so high if the knowledge that
others had just rated them had not influenced the self-rating
The insignificant correlation between the Trust Scale and the
 664 Julian B. Rotter
S-D Scale may appear surprismg in Ught of the correlation of
.29 found in the large sample. However, the S-D Scale was given
in this case after a sociometnc task in which each subject knew
he was bemg rated by others on a number of vanables As a
result mean scores for the S-D test were significantly depressed
in the direction of greater honesty The mean S-D score for the
sociometric study was 12 4, for the earlier study it was 14 3.
While trust and trustworthiness showed a sigmficant rela-
tionship, some evidence that they are also regarded somewhat
differently can be found m the correlations of both vanables,
measured sociometncaUy, with popularity and friendship Trust-
worthmess is clearly seen as the more desirable trait with a
significantly higher relationship to popularity
In summary, sociometnc analysis reveals relatively good
construct and discnmmant validity for the Interpersonal Trust
Scale as agamst observed behavior m groups who have had
ample opportunity and a long tune to observe each other. Tmst
as measured sociometncaUy was negatively related to depen-
dency, not significantly related to gidhbihty, and positively re-
lated to humor, fnendship, populanty, and especially trust-
worthmess.

SUMMABY

Interpersonal tmst, defined as a generalized expectancy that

the verbal statements of others can be rehed upon, appears po-
tentially to be a frmtful vanable for investigation in several fields
of psycholo^ A new, Likert-type scale was developed and
refined on the basis of item analysis of mtemal consistency, rela-
tive mdependence of social desirabihty, and item spread Overall
mtemal consistency and test-retest rehabihty appear satisfac-
tory Demographic data were exammed for 547 college students.
Tmst scale scores are related significantly to position m the
family, socioeconomic level, rehgion, and rehgious differences be-
tween parents. A first assessment of constmct and discnmmant
vahdity was attempted by a sociometnc study of two fratermties
and two sororities. Results mdicate both good constmct and
discnimnant vahdity for the Interpersonal Trust Scale
 Measurement of interpersonal trust 665

REFEBENCES
Chnstie, R , & Merton, R K Procedures for the sociological study of the
values climate of medical schools / med Educ, 1958, 33, 125-133
Crowne, D P , fit Marlowe, D The approval motive Studies in evaluative
dependence New York Wiley, 1964
Deutsch, M Trust and suspicion. J Conflict Resdut, 1958, 2, 265-279
Deutsch, M Trust, trustworthmess, and the F scale / abnorm soc Psychol,
i960, 81, 138-140
Enkson, E H Crowth and cnses of the "Healthy personality" In C Kluck-
hohn and H Murray (Eds ), Persondtty m nature, society, and culture
(2nd ed ) New York Knopf, 1953
Geller, J D Some personal and situational determinants of mterpersonal trust
Unpublished doctoral dissertation, Umver of Connecticut, 1966
Kelley, H , & Rmg, K Some effects of "suspiciousness" versus "trustmg" trammg
schedules / cAmorm soc Psychol, 1961, $3, 294-301
Loomis, J L Communication, the development of trust, and cooperative be-
havior Hum Relat, 1959, " , 305-315
Mahrer, R R The role of expectancy m delayed remforcement / exp Psychol,
1956, 5X, 101-105
Mellmger, G D Inteipersonal trust as a factor m commumcabon / abnorm
soc Psychol, 1956, SZ, 304-309.
Merton, R Social theory and social structure Glencoe, UI Free Press, 1949
Mischel, W Father-absence and delay of gratification Cross-cultural com-
parisons / abnorm soc Psychol, 1961, S3, 116-124 (a)
Mischel, W Preference for delayed remforcement and social responsibihty
7 abnorm soc Psychd, 1961, U, 1-7 (b)
Rapaport, A , & Orwant, Carol Expenmental games A review Behao Sci,
1962, 7, 1-37
Redl, F , & Wmeman, A Children who hate Glencoe, 111 Free Press, 1951
Rotter, J B Social learning and clinical psychology Englewood Chffs, N J
Prentice-Hall, 1954
Rotter, J B The role of the psychological situabon m determmmg the direcbon
of human behavior In M R Jones (Ed ), Nebraska symposium on moti-
vation Lmcob Umver of Nebraska Press, 1955
Rotter, J B Some imphcabons of a social leammg theory for the predicbon of
goal directed behavior from testmg procedures Psychol Rev, i960, 87,
301-316
Scodel, A Induced collaborabon m some non-zero-sum games / Confiict Reso-
hit, 1962, 8, 335-340

Manuscript received February 27, 1967