JOURNAL OF RESEARCH IN PERSONALITY 31, 319–336 (1997)

ARTICLE NO. RP972186

Measuring Levels of Trust

Laurie L. Couch
Fort Hays State University

and

Warren H. Jones
University of Tennessee

College student respondents involved in romantic relationships (N 5 445) com-

pleted questionnaires containing measures of trust and related constructs. Analyses
focused on two sets of issues. The first concerned the validity of the Trust Inventory,
an innovative self-report measure that partitions trust into separate domains includ-
ing (a) specific relationship partners, called Partner Trust; (b) family and friends,
termed Network Trust; and (c) people-in-general, called Generalized Trust. The
second set of issues involved several previously unanswered questions derived from
the trust literature, specifically: (a) the comparability of competing measures of trust,
(b) the convergence between trust in specific people vs trust in human nature, and
(c) whether trust is more closely related to one’s personality or emotions or to the
quality of one’s relationships. Results generally supported the validity of the Trust
Inventory and its tripartite division of types of trust including the new concept of
Network Trust. In addition, various measures of trust were moderately to strongly
interrelated. However, results also supported the distinction between relational trust
(trust in relationship partners) and global trust (trust in human nature). Measures
of relational trust were significantly more strongly related to relationship quality
and commitment, whereas measures of global trust were slightly more strongly re-
lated to indices of personality and emotion. This latter difference was not signifi-
cant.  1997 Academic Press

Trust has long been an important construct in personality and in under-

standing development, communication, personal relationships, and organiza-
tional behavior. However, the literature on trust contains a diversity of defi-
nitions and conceptualizations, some anomalous findings, and several
unresolved theoretical problems. For example, trust has been defined in di-
verse ways: as a generalized expectancy (Rotter, 1967), as an enduring atti-

Address reprint requests to Dr. Laurie L. Couch, Department of Psychology, Fort Hays
State University, Hays, KS 67601.
319
0092-6566/97 $25.00
Copyright  1997 by Academic Press
All rights of reproduction in any form reserved.
320 COUCH AND JONES

tude or trait (Deutsch, 1958; Giffin, 1967), and as a transitory situational

variable (Driscoll, 1978; Kee & Knox, 1970). Some global measures of trust
are not reliably associated with the trust experienced in specific relationships
(e.g., Holmes, 1991; Tardy, 1988).
Earlier treatments (e.g., Deutsch, 1958; Rotter 1967) often defined trust
broadly as trust in human nature or people-in-general (i.e., termed global
trust), whereas more recent approaches focus on trust in a specific rela-
tional—often romantic—partner (termed relational trust; Lazelere & Hus-
ton, 1980; Rempel, Holmes, & Zanna, 1985). We believe this distinction is
an important conceptual innovation and a potential point of resolution for
some of the theoretical issues involving the concept of trust. A further re-
finement in this line of reasoning is implicit in the Trust Inventory (Couch,
1994; Couch, Adams, & Jones, 1996) which is the first instrument that mea-
sures both global and relational trust simultaneously. In fact, this inventory
divides the construct into three levels: Generalized Trust (i.e., global trust
in human nature), Partner Trust (i.e., trust in specific relationship partners),
and an intermediate level called Network Trust defined as trust in one’s social
network of family and friends. The purpose of the current investigation was
to clarify further the construct of trust by exploring the validity of the tripar-
tite conceptualization contained in the Trust Inventory and by assessing the
convergence among differing definitions and measures of trust (e.g., global
vs relational trust, etc.).
GLOBAL TRUST
Most of the initial theoretical descriptions of trust tended to be broadly
conceived in the sense of seeking to account for an individual’s orientation
toward human nature or people-in-general, referred to as global trust (Rotter,
1967; Wrightsman, 1974). For example, according to Erikson (1964) the
initial task of human development involves learning to trust the primary care-
giver. The infant is deeply influenced by the caregiver’s responsiveness and
sensitivity to the expression of its needs and thus, a strong impression of the
predictability and dependability of the caregiver is formed. The resultant
trust in the caregiver is essential not only for a formation of secure attachment
but also for progression to subsequent tasks of development and the general-
ization of trust to other people. The concept of a global trust is also illustrated
by the work of Rotter (1967, 1971, 1980), who focused on what he called
‘‘interpersonal trust.’’ Rotter defined interpersonal trust as a generalized ex-
pectancy that the promise of an individual or a group can be relied upon.
According to Rotter, trust is not related to any specific experience but derives
from generalizations across experiences that the individual perceives as simi-
lar. Rotter argued that because trusting individuals work on the assumption
that people will respond to them and make their needs known, a sense of
personal control may develop. For example, internal locus of control, or the
 LEVELS OF TRUST 321

feeling that one has some control over the outcome of situations one is in,
has consistently been found to be related to global trust (Rotter, 1980). Fi-
nally, several scholars including Erikson (1950), Wrightsman (1974), and
Rotter (1967) proposed that global trust, conceived in this way, is the essen-
tial and necessary interpersonal component for developing relationships.

RELATIONAL TRUST
By contrast, relatively little attention has been paid to assessing trust in
close relationships directly. This is surprising considering the amount of re-
search recently directed toward the study of personal relationships, and con-
sidering that trust is often seen as a major factor in relationship development
(Hendrick & Hendrick, 1983; Lazelere & Huston, 1980; Rempel, Holmes, &
Zanna, 1985). Relational trust differs from the traditional ideal of global trust
in that it is focused on a specific partner with whom a person has an important
relationship (Holmes, 1991; Holmes & Rempel, 1989). It refers to a person’s
level of confidence in the strength of the relationship and his/her partner’s
positive feelings (caring) toward the person (Rempel et al., 1985). Theoreti-
cal work on the construct of relational trust has generated three main ap-
proaches: (a) developmental theories which are concerned with how trust
emerges and changes in a developing relationship (e.g., Holmes & Rempel,
1989; Lazelere & Huston, 1980); (b) the component model (Rempel et al.,
1985) which focuses on the evolution of trust through specific stages such
as predictability, dependability, and faith; and (c) the appraisal process model
that emphasizes the perceptual and cognitive consequences of one’s level of
trust (Holmes, 1991). Also, with few exceptions the concept of relational
trust has been almost exclusively used to refer to trust in romantic partners.
As is implied by the term relational trust, each of these approaches assumes
that trust is a critical element in any close or important relationship.

RESEARCH ISSUES
The development of the concept of relational trust as a theoretical compan-
ion to global trust is, we believe, an advancement in understanding the nature
of trust in experience and most especially the role of trust in close relation-
ships. Even so, several important issues remain unresolved in recent research
and theory. First, defining trust exclusively in either relational (i.e., specific
romantic partners) or global (i.e., people-in-general) terms would appear to
omit a portion of the potential range of the trust domain; namely, one’s will-
ingness to trust the members of one’s social network (i.e., family and
friends). In contrast to ephemeral human nature, one has numerous specific
interactions with the members of one’s social network, and yet those experi-
ences may lack the immediacy, intimacy, and in some cases the importance
of exchanges with one’s romantic partner. Whether trust in family members
322 COUCH AND JONES

and friends functions as it appears to in romantic relationships is not yet

clear and is one of the general questions that initiated the present research.
Second, the statistical relationship between global trust and relational trust
remains unclear. It seems reasonable to expect, and Rotter (1967) specifically
proposed that global trust and trust for relational partners would be corre-
lated. However, it is also reasonable to suppose that in some cases a discrep-
ancy between global and relational trust could occur in which one is sanguine
regarding people-in-general but suspicious of one’s partner. Or the reverse
may hold true; one may be generally mistrusting but have faith in one particu-
lar person or a small group of people. Also, research to date has often failed
to demonstrate a reliable relationship between measures of the two constructs
(Holmes, 1991; Tardy, 1988).
Third, is trust best conceived as a characteristic of a person versus a judg-
ment arising from interactions with a specific other person or people-in-gen-
eral? The former might be termed the personality trait approach and assumes
that individuals vary in trust more-or-less independent of their contemporary
interpersonal experience. One can easily imagine people who might trust
others excessively even when they are in danger and thus at a disadvantage
to do so. Similarly, one can imagine others who exhibit extreme distrust
despite the reliability of their relational partners. Presumably, one’s current
level of trust for specific or generalized others could derive from past rela-
tionship experiences or other mechanisms of personality development. By
contrast, trust may reflect the status or quality of one’s current relationships.
Although difficult to disentangle, these competing characterizations suggest
incompatible origins and consequences of trust. Alternatively, each approach
may reflect differing levels of trust. For example, global trust is likely to be
more closely linked to personality processes whereas, by definition, rela-
tional trust should reflect actual and current relationships and interpersonal
experiences.
Fourth, an important question for any literature for which a variety of
definitions and potentially dissimilar measures have been used concerns the
extent to which established results reflect the constructs underlying theoreti-
cal models or simply the idiosyncrasies of the instruments and definitions
involved. This issue is related to most of the points raised above in that, at
present, the generalizability from studies involving global trust to those fo-
cused on relational trust remain largely unexplored.
The Trust Inventory (Couch, 1994; Couch, et al., 1996) emerged from a
program of research stimulated, in part, by these conceptual and measure-
ment issues. Of special note, in addition to yielding scores for global and
relational trust, this instrument includes a measure of network trust, designed
to assess the confidence, dependence, and trust, in one’s network of friends
and family members. Although it is new, existing research suggests that the
Trust Inventory is internally consistent, reliable over time, and preliminary
 LEVELS OF TRUST 323

evidence supports the validity of scale interpretations. Consequently, one

purpose of the present research was to examine further the validity of the
three conceptualizations of trust as defined by the Trust Inventory scales
using comparisons with both (a) extant measures of global and relational
trust and (b) measures of selected personality traits versus interpersonal and
relational constructs of relevance. In addition, the present research sought
to explore further the distinctions outlined by the tripartite theory of trust
(e.g., among global, network, and relational trust) at the level of the construct
(c) by examining comparisons among extant measures of both global and
relational trust, including the Trust Inventory and by (d) comparing these
measures to the personality and relational constructs used to validate the
Trust Inventory.

METHOD
Participants
Originally, respondents were 552 undergraduate college students. Of this group, 445
(80.6%) reported being involved in a romantic relationship, the mean length of which was
22.9 months. In order to eliminate potential variability in Partner Trust scores deriving from
respondents not involved in a romantic relationship, analyses were conducted on these latter
participants only. The mean age of the participants was 20.2 years, and 89.8% were Caucasian.
These participants completed the Trust Inventory along with one or more additional measures.
All participants were recruited from undergraduate psychology classes at a large public Univer-
sity.

Procedure
Participants completed questionnaires in out-of-class sessions in exchange for nominal
course credit. Each participant completed questionnaires containing: (a) biographical questions
(e.g., age, romantic involvement, etc.), (b) the revised Trust Inventory, (c) alternative measures
of global and relational trust, and (d) measures of various personality, emotional, and relation-
ship constructs included for the purpose of comparisons with the measures of trust.

Measures
The trust inventory. The Trust Inventory (Couch, 1994; Couch, Adams, & Jones, 1996)
yields three scores reflecting a priori definitions of trust as follows: Partner Trust, defined as
trust or confidence in a romantic partner or in one’s romantic relationship; Network Trust
defined as feelings of confidence and security one has in one’s network of relationships with
family and friends; and, Generalized Trust, defined as the tendency to entertain positive
assumptions about people-in-general, or to attribute positive characteristics to ‘‘human na-
ture.’’
Initial estimates of internal reliability for the current version of the Trust Inventory have
met or exceeded conventional standards of measurements (Couch, 1994; Couch et al., 1996):
Coefficient αs ranged from .87 to .92, and the mean interitem correlations for the scales ranged
from .33 to .40. The Partner Trust scale consists of 20 items such as ‘‘I am sure about how
my partner feels about me,’’ whereas Network Trust (e.g., ‘‘I often worry about the motives
of others in my life’’) and Generalized Trust (e.g., ‘‘Basically I am a trusting person’’) consist
of 10 and 20 items, respectively. An examination of test–retest reliability over a 9-week period
324 COUCH AND JONES

indicated sufficient temporal stability for Partner (r 5 .82), Network (r 5 .74), and Generalized
Trust (r 5 .80). Preliminary validity evidence regarding the Trust Inventory involving compar-
isons with extant measures of trust and relationship status indicated moderate to strong support.
The three scales showed moderate intercorrelations (Couch et al., 1996).
The Partner Trust scale, which measures the faith one has in a romantic partner or in a
specific romantic relationship, has been found to be related to commitment to one’s partner,
relational satisfaction, and passionate love. The Network Trust scale was designed to assess
one’s trust for members of his/her social network (e.g., friends and family). The Network
Trust scale has been found to be positively correlated with friendship potential, a global rating
of friendship satisfaction, and family satisfaction. The Generalized Trust scale was designed
to measure one’s belief that human nature is basically good, and validation studies have indi-
cated that its scores are related to social support, liking others, and trust in human nature
(Couch et al., 1996).
Measures of relational trust. Participants also completed three additional measures of rela-
tional trust. These included the Faith subscale of the Trust Scale (Rempel, Holmes, & Zanna,
1985), 7 items from the Emotional Trust subscale of the Specific Interpersonal Trust Scale
(Johnson-George & Swap, 1982), and the Dyadic Trust Scale (Lazelere & Huston, 1980). The
Trust Scale is a 26-item measure of trust in romantic relationships. Its Faith subscale is a 10-
item measure of the feelings of confidence in a romantic relationship and romantic partner.
Available evidence suggests that it yields reliable scores (coefficient α 5 .80) and there is
substantial evidence to support the validity of the scale. This scale has been found to correlate
with feelings of love for one’s partner and happiness (Rempel et al., 1985). Furthermore, these
items can be distinguished from the Dependability and Predictability subscales of the Trust
Scale along a temporal dimension (i.e., Faith focuses on the present and future whereas De-
pendability and Predictability focus on the past).
The Emotional Trust subscale of the Specific Interpersonal Trust Scale is a 9-item measure
of the trust one individual has for another in situations involving confiding and self-disclosure,
as well as potential criticism and embarrassment. Two of the nine items were gender specific
(i.e. items for women only), so only the remaining seven items from this subscale were used
in the present study. All of the subscales of the Specific Interpersonal Trust Scale have shown
evidence of reliability (αs . .71), including the Emotional Trust subscale. Johnson-George
and Swap (1982) reported evidence suggesting that trust in a specific other person is distinct
from both love and liking. They also presented evidence suggesting both convergent and dis-
criminant validity for this scale.
The Dyadic Trust Scale is an 8-item measure of trust in close relationships; specifically, the
respondent’s judgment of goodwill and honesty of a significant other toward the respondent.
Reliability has been demonstrated (coefficient α 5 .93) and some evidence of construct validity
has been presented. Lazelere and Huston (1980) found dyadic trust to be correlated with mea-
sures of love and intimacy of self-disclosure. Dyadic trust showed greater reciprocity between
partners than either love or self-disclosure. Dyadic trust was also found to vary in association
with the level of commitment toward the relationship. For example, the lowest levels of trust
were observed among ex-partners, whereas the highest trust scores were obtained for engaged
and married couples. This measure has also been found to be operationally distinct from gener-
alized trust.
Measures of global trust. Measures of global trust included the Interpersonal Trust Scale
(Rotter, 1967) and the Trustworthiness subscale of the Philosophies of Human Nature Scale
(Wrightsman, 1974). The Interpersonal Trust Scale is a 25-item measure intended to assess
trust defined as the expected reliability of others’ words or promises regarding a variety of
people and situations. The scale is designed to assess trust in ambiguous, novel, or unstructured
situations, in which one’s generalized expectancy is all one can rely upon. Coefficient α of
.79, a split-half reliability of .76, and test–retest reliability of .58 and .68 have been reported
(Pereira & Austrin, 1980). In addition, the Interpersonal Trust Scale has been widely used to
 LEVELS OF TRUST 325

study a range of topics. Rotter (1980), for example, found that interpersonal trust was inversely
correlated with telling a lie, cheating, unhappiness, and maladjustment.
The Trustworthiness Subscale of the Philosophies of Human Nature Scale is a 14-item
measure of one’s beliefs that human nature or people-in-general are trustworthy, moral, and
responsible. Coefficient α has been reported to be .79, and the scale correlates .76 with Rotter’s
Interpersonal Trust Scale (Chun & Campbell, 1975). Women, as compared to men, have been
found generally to hold more favorable views of human nature, and this scale also predicts
religious attitudes, attitudes toward teachers, self-esteem, and (inversely) political cynicism,
and Machiavellianism (Wrightsman, 1964).
Measures of relationship constructs. Several scales were selected to represent the domain
of relationship experiences for comparison to the Trust Inventory and other measures of trust.
Measures were chosen to represent both positive (e.g., love, liking, satisfaction, commitment,
and self-disclosure) and negative (e.g. loneliness, betrayal of others, betrayal by others) aspects
of relational behavior. The Love and Liking Scales (Rubin, 1970) consist of 26 items with
13 items designed to measure love and 13 items designed to measure liking. The scale has
proven reliability (α . .84) and is widely used in studies of romantic and dating relationships
(e.g., Rosenman, 1978).
The Love Attitudes Scale (Hendrick & Hendrick, 1986) is a 42-item measure designed to
distinguish between six basic love attitudes as defined by Lee’s (1973) theory of love styles:
Eros (passionate love), Ludus (game-playing love), Storge (friendship love), Pragma (logical
love), Mania (possessive or dependent love), and Agape (selfless love). The love styles ostensi-
bly are independent ways of approaching the love experience and are related to both personality
and attitudes (Raciti & Hendrick, 1992). Each of the six scales shows adequate internal reliabil-
ity (coefficient alpha ranged from .62 to .84 and test–retest reliabilities from .60 to .78), and
the scales have yielded relatively low correlations with each other (Hendrick & Hendrick,
1986).
A revised version of the Marital Satisfaction and Commitment Scale (Jones, Adams, Mon-
roe, & Berry, 1995) was used to assess the level of satisfaction with, and commitment to one’s
relationship. The questionnaire is a 25-item measure, with 10-items measuring commitment
to one’s relationship and partner, and 15-items measuring one’s satisfaction with the romantic
relationship. The inventory was designed to assess satisfaction and commitment in a marriage
relationship. Because the participants from this study are primarily unmarried individuals, all
references to ‘‘spouse’’ or ‘‘marriage’’ in the items was changed to ‘‘partner’’ or ‘‘romantic
relationship’’. The satisfaction and commitment scales have demonstrated validity (e.g., they
discriminate between married and divorced couples) and internal consistency (αs . .75, and
. .80), and show adequate test-retest reliability.
The UCLA Loneliness Scale (Russell, Peplau, & Cutrona, 1980) is a valid and reliable (α
5 .96, test–retest r 5 .73) measure of global dissatisfaction with the number and quality of
one’s social and emotional relationships. The scale is a 20-item instrument designed to measure
subjective feelings of loneliness and the discrepancy between perceived and idealized relation-
ships. The UCLA Loneliness Scale has been widely used in studies of loneliness and relation-
ship problems.
The Interpersonal Betrayal Scale (Jones & Burdette, 1994) contains 30-items measuring
the tendency to betray one’s friends and family members and the corresponding perception
of believing that one has been betrayed by one’s relational partners. Research supports the
internal reliability of the scales (αs . .75). Validity of scale interpretations has been suggested
by available evidence (e.g., alternative measures of betrayal), and evidence further suggests
that scores do not simply reduce to psychopathy or antisocial personality (Jones, Cohn &
Miller, 1991).
A self-disclosure questionnaire (Chaikin & Derlega, 1976) was used to measure the respon-
dent’s willingness to share information about oneself with the romantic partner. The 26-items
present the respondent with types of information previously rated for depth (intimacy) of
326 COUCH AND JONES

disclosure. Respondents were asked to rate on a 5-point scale their willingness to disclosure
the item of information in question to their romantic partner. The self-disclosure score was
the sum of these ratings.
Measures of personality/emotions. Several measures representing the general domain of
personality and emotions were selected for inclusion in this study including measures of shy-
ness, jealousy, guilt, suspicion, and hostility. Each of the constructs was also chosen for its
relevance to interpersonal behaviors and experience in relationships. The Social Reticence
Scale (Jones & Briggs, 1986) is a 20-item measure of shyness. The scale has been shown to
have substantial internal consistency (α 5 .91) and test–retest reliability (r 5 .88 over an 8-
week period and r 5 .78 over 12 weeks). Criterion and construct validity evidence has also
been reported (Jones, Briggs, & Smith, 1986).
The Interpersonal Jealousy Scale (Mathes & Severa, 1981) is a 28-item measure of the
amount of negative emotion resulting from actual or threatened loss of a loved one to a rival
(jealousy). The scale has been found to be internally reliable (α 5 .92) and valid. Discriminant
validity has been demonstrated between jealousy and a liking factor suggesting that the scale
is not just a measure of interpersonal attraction.
The Trait Guilt Scale of the Guilt Inventory (Kugler & Jones, 1992) was designed to measure
a continuing and nonspecific sense of guilt and regret. The scale contains 20 items and has
demonstrated validity and reliability. One estimate of internal reliability was α 5 .89, and
evidence supporting concurrent and discriminant validity has been reported (Jones & Kugler,
1993; Kugler & Jones, 1992).
The Suspicion Subscale of the Buss–Durkee Hostility Inventory (Buss & Durkee, 1957)
was used to measure projection of hostility onto others. The 10-item subscale asks respondents
to indicate their agreement with statements about being distrustful and wary of people, and
beliefs that others are being derogatory or planning harm toward the respondent. The suspicion
subscale has been shown to be a valid and reliable measure of distrustfulness.

RESULTS
The Trust Inventory
The utility of its scales and the tripartite conceptualization of trust of the
Trust Inventory were assessed in two ways: (a) the Trust Inventory scales
were correlated with extant measures of both global trust and relational trust
in order to obtain estimates of convergent and discriminant validity; and
(b) these scales were compared to scores from measures of constructs theo-
retically related to trust.
Concurrent validity. Table 1 presents correlations between scales of the
Trust Inventory and measures of both relational and global trust. As ex-
pected, Partner Trust was more highly correlated with the relational trust
measures of Faith, Dyadic Trust, and Emotional Trust than with the global
trust measures (e.g., Rotter’s Interpersonal Trust Scale) (z 5 7.41, p , .001).
This supports both the convergent and the discriminant validity of the Partner
Trust Scale. By contrast, as is indicated in Table 1, the Generalized Trust
scale was significantly related to Wrightsman’s Trustworthiness and Rotter’s
Interpersonal trust but only slightly more strongly than it was to Faith, Dy-
adic Trust, and Specific Trust (z 5 1.31, ns). Thus, these results support the
validity of the Generalized Trust Scale as a measure of global trust, although
 LEVELS OF TRUST 327

TABLE 1
Correlations between Trust Inventory Scales and Alternative Measures
Variable Partner Network Generalized

Relational trust
Faith .82** .47** .42**
Emotional trust .77** .47** .45**
Dyadic trust .82** .47** .40**
Global trust
Interpersonal trust .41** .59** .56**
Trustworthiness .29** .47** .46**

Note. Faith, the Faith subscale of the Trust Scale (Holmes, Rempel, & Zanna, 1985); Emo-
tional Trust, the Emotional Trust subscale of the Specific Interpersonal Trust Scale (Johnson-
George & Swap, 1982); Dyadic Trust, the Dyadic Trust Scale (Lazelere & Huston, 1980);
Interpersonal Trust, the Interpersonal Trust Scale (Rotter, 1967); Trustworthiness, the Trust-
worthiness subscale of the Philosophies of Human Nature Scale (Wrightsman, 1974).
**p , .01.

greater differentiation for this scale between the global and relational do-
mains had been expected. Similarly, the Network Trust scale was related
moderately, but significantly to Faith, Dyadic Trust, and Emotional Trust
and also with Interpersonal Trust and Trustworthiness of Human Nature
(z 5 0.94, p . .05). The pattern of correlations for Network Trust falls
intermediately between that of Partner and Generalized Trust as would be
expected.
Construct validity. Table 2 presents correlations between the Trust In-
ventory Scales and measures of variables theoretically related to trust in an
assessment of construct validity. As may be seen, of the three Trust Inventory
scales Partner Trust showed the strongest correlations with both the Loving
and Liking Scales (Rubin, 1970). Also as is indicated in Table 2, the Eros,
Ludus, and Agape Scales from the Love Attitudes Scale were most strongly
related to Partner Trust as would be expected. By contrast, Pragma and Ma-
nia were most strongly related to Network Trust. Storge was not significantly
related to any of the three trust scales. By contrast, loneliness was inversely
related to all three scales. Also consistent with expectations, self-disclosure
was moderately related to Network Trust and Partner Trust, but was not
significantly correlated with Generalized Trust. Satisfaction with one’s cur-
rent romantic relationship and commitment to that relationship as measured
by the revised Marital Satisfaction and Commitment Scales were both con-
siderably more strongly related to Partner Trust. All three measures were
inversely related to indices of betraying and being betrayed by the members
of one’s social network. More specifically, having been betrayed by others
was most strongly related to Network Trust, whereas the self-reported fre-
328 COUCH AND JONES

TABLE 2
Correlations between Trust Inventory Scales and Measures of
Relationships and Personality

Scale/variable Partner Network Generalized

Relationship measures
Rubin
Love .50** .23** .29**
Liking .64** .37** .41**
Hendrick
Eros .55** .26** .24**
Ludus 2.60** 2.47** 2.38**
Storge .02 2.03 .06
Pragma 2.21** 2.26** 2.17*
Mania 2.23** 2.29** 2.16*
Agape .46** .19** .25**
Loneliness 2.63** 2.64** 2.55**
Self-disclosure .27** .14* .13
Satisfaction .81** .46** .35**
Commitment .69** .38** .33**
Betrayal of others 2.22** 2.23** 2.27**
Betrayal by others 2.32** 2.46** 2.33**
Personality/emotion
Shyness 2.28** 2.38** 2.41**
Jealousy 2.26** 2.29** 2.26**
Suspicion 2.50** 2.72* 2.65**
Guilt 2.25* 2.37** 2.06

Note. Rubin’s Love and Liking, the Rubin Love and Liking Scale (Rubin, 1970); Hendrick &
Hendrick, The Attitudes About Love Scale (Hendrick & Hendrick, 1986); Loneliness, the
UCLA Loneliness Scale (Russell, Peplau, & Cutrona, 1980); Self-disclosure, a self-disclosure
questionnaire (Chaikin & Derlega, 1976); Satisfaction, the revised version of the Marital Satis-
faction and Commitment Scale (Jones, Adams, Monroe, & Berry, 1994); Commitment, the
revised version of the Marital Satisfaction and Commitment Scale (Jones, Adams, Monroe, &
Berry, 1994); Betrayal of Others, the Interpersonal Betrayal Scale (Jones & Burdette, 1994);
the Interpersonal Betrayal Scale (Jones & Burdette, 1994); Shyness, the Social Reticence Scale
(Jones & Briggs, 1986); Jealousy, the Interpersonal Jealousy Scale (Mathes & Severa, 1981);
Suspicion, the Suspicion Scale of the Buss-Durkee Hostility Inventory (Buss & Durkee, 1957);
Guilt, the Guilt Inventory (Kugler & Jones, 1992).
*p , .05; **p , .01.

quency of betraying others was most strongly correlated with Generalized

Trust.
Trust Inventory scores were also found to be related to selected emotional
states and traits as presented in Table 2. Interestingly, shyness was most
strongly correlated with Generalized Trust, followed by Network Trust and
then Partner Trust. Jealousy was also inversely correlated with all these mea-
sures of trust, with Network Trust most strongly related. As expected, the
highest (inverse) correlates of suspicion were Generalized Trust and Network
 LEVELS OF TRUST 329

Trust. Trait Guilt as measured by the Guilt Inventory was related most
strongly to Network Trust followed by Partner Trust. Trait Guilt was not
significantly related to Generalized Trust.
Overall, results indicate that partner trust correlates more strongly with
measures of relationship quality and stability than with global personality
measures (z 5 2.11, p , .05). The network trust scale does not correlate
differentially between the two kinds of measures (z 5 21.03, ns), nor does
the generalized trust scale (z 5 21.22, ns).
Biographics. Although secondary to the purposes of this study, analyses
were conducted to assess variations in Trust Inventory Scores associated with
the biographic variables included. Two patterns of results were statistically
significant. First, female respondents scored higher on the measures of trust
than did male participants. Women scored higher than did men on Partner
Trust [t(444) 5 3.20, p , .01], Network Trust [t(446) 5 3.00, p , .01],
and Generalized trust [t(444) 5 4.22, p , .01) and these findings are consis-
tent with gender differences in trust previously reported in the literature
(Johnson-George & Swap, 1982; Rempel et al., 1985; Rotenberg, 1984).
Second, it was found that the length of one’s romantic relationship was re-
lated to one’s scores on the measures of trust. Those involved in a relation-
ship of 36 months or longer in length compared to those involved in relation-
ships of shorter duration reported higher levels of Partner Trust [t(441) 5
3.57, p , .01], Network Trust [t(443) 5 3.12, p , .01], and Generalized
Trust [t(441) 5 2.65, p , .01]. These findings support one view that trust
evolves over time in a close relationship (Lazelere & Huston, 1980; Holmes,
1991).

Trust as a Construct
In reviewing the literature, several issues regarding the nature of trust were
raised including: (a) the relationship between global trust and relational trust;
(b) the convergence between conceptualizations of trust as a dimension of
personality versus an indicator of relationship quality; and (c) the correla-
tions among differing measures of relational and global trust. These issues
were examined further by calculating correlations among and between the
group of trust measures classified as global or relational, and between these
groups and the group of measures classified as relationship experiences ver-
sus personality (or emotional) traits. Specifically, the intercorrelations of all
the relational trust measures were computed and a mean correlation for each
scale with all other measures of relational trust was obtained. These measures
included the Partner Trust Scale, the Faith Subscale of the Trust Scale, the
Emotional Trust Subscale of the Specific Interpersonal Trust Scale, and the
Dyadic Trust Scale. The mean correlation was obtained by first converting
each Pearson r to Fisher’s z, calculating the mean and reconverting to r. The
330 COUCH AND JONES

TABLE 3
Average Correlations between Measures of Trust
and Groups of Trust Measures

Variable Relational Global

Partner trust .81 .49

Faith .82 .36
Emotional trust .80 .37
Dyadic trust .81 .36
Mean .81 .40
Interpersonal trust .35 .49
Network trust .52 .60
Generalized trust .45 .47
Trustworthiness .24 .55
Mean .39 .53
Note. Partner Trust, the Partner Trust Scale of the
Trust Inventory; Faith, the Faith subscale of the
Trust Scale (Holmes, Rempel, & Zanna, 1985); Dy-
adic Trust, the Dyadic Trust Scale (Lazelere & Hus-
ton, 1980); Emotional Trust, the Emotional Trust
subscale of the Specific Interpersonal Trust (John-
son-George & Swap, 1982); Trustworthiness, the
Trustworthiness subscale of the Philosophies of Hu-
man Nature Scale (Wrightsman, 1974); Interper-
sonal Trust, the Interpersonal Trust Scale (Rotter,
1967); Network Trust, the Network Trust Scale of
the Trust Inventory; Generalized Trust, the General-
ized Trust Scale of the Trust Inventory.

same procedure was used to form a group of global trust measures consisting
of the Interpersonal Trust Scale, the Trustworthiness Subscale, and the Net-
work Trust Scale, and the Generalized Trust Scale of the Trust Inventory.
The Network Trust Scale was included as a measure of global trust because
no comparable measures of this category or type of trust were available and
because the items refer to friends and family relationships collectively rather
than individually.
Results from the average correlation comparisons of each scale with the
measures of the relational trust group and the global trust group are presented
in Table 3. For these comparisons, the individual scale was first removed
from its respective group of measures before computations were made. As
is indicated, each scale in the relational trust group on average correlated
very highly with the others in the relational trust group. This is evidence of
the concurrent validity of the measures of relational trust. These comparisons
also demonstrated that each of the relational trust measures were better mea-
sures of relational trust than they were measures of global trust. Thus, these
findings also supported the discriminant validity of each of the relational trust
 LEVELS OF TRUST 331

scales included. No significant differences were found between the average

correlations within the group. Also, as a group, the overall mean correlation
of the relational trust measures with one another was significantly higher
than their mean correlation with the global trust measures (z 5 10.10, p ,
.01), thereby providing empirical support for the theoretical distinction be-
tween the two types of trust.
A similar pattern of correlations emerged in comparisons involving the
group of global trust measures. Among these measures, the Network Trust
Scale yielded the highest average correlation with the relational trust mea-
sures, but the scale showed a strong average correlation with global trust
measures. The other scales in the group also correlated highly with each
other, and no significant differences within the group were observed among
the measures of global trust. Finally, the mean correlation within all global
trust measures included in the study was higher than the mean correlation
between global and relational trust measures and this difference was signifi-
cant (z 5 2.62, p , .05).
In addition, this procedure was also used to compare each scale’s average
correlation with a group of relationship measures (i.e., the Rubin Loving and
Liking Scales, the Self-Disclosure Test, the revised Marital Satisfaction and
Commitment Scales, the UCLA Loneliness Scale, and the Interpersonal Be-
trayal Scales) versus a group of inventories assessing personality trait and
emotion constructs (i.e., the Social Reticence Scale, the Interpersonal Jeal-
ousy Scale, the Trait Guilt Scale of the Guilt Inventory and the Suspicion
Subscale of the Buss-Durkee Hostility Inventory). The six Love Style Scales
(Hendrick & Hendrick, 1986) were omitted from these analyses because
these scales have divergent, often diametrically opposed, definitions and con-
ceptualizations. Results from comparisons of the group of relational trust
measures and the group of global trust measures with measures of relation-
ships and measures of personality/emotion traits are presented in Table 4.
As is consistent with the definitions of relational vs. global trust, the average
correlation of relational trust measures with the group of relationship scales
was significantly higher than its average correlations with the group of trait
and emotion measures (z 5 3.47, p , .01). This finding provided convergent
and discriminant validity evidence for the measures of relational trust and
further support for the distinction between the types of trust. Finally, al-
though global trust measures appeared to be more highly related to measures
of personality traits and emotions than to measures of relationships, this dif-
ference was not significant (z 5 1.71, p . .05).

DISCUSSION
The Trust Inventory
The results of this study suggested three broad conclusions regarding the
Trust Inventory. First, in general these data supported the concurrent validity
332 COUCH AND JONES

TABLE 4
Average Correlations between Trust Measures and
Groups of Relational and Trait/Emotion Measures

Trait/
Variable Relational Emotional

Partner trust .54 .35

Faith .55 .19
Emotional trust .49 .21
Dyadic trust .38 .23
Mean .49 .25
Interpersonal trust .25 .41
Network trust .37 .49
Generalized trust .35 .46
Trustworthiness .19 .28
Mean .29 .41

Note. Partner Trust, the Partner Trust Scale of the

Trust Inventory; Faith, the Faith subscale of the
Trust Scale (Holmes, Rempel, & Zanna, 1985); Dy-
adic Trust, the Dyadic Trust Scale (Lazelere & Hus-
ton, 1980); Emotional Trust, the Emotional Trust
subscale of the Specific Interpersonal Trust Scale
(Johnson-George & Swap, 1982); Trustworthiness,
the Trustworthiness subscale of the Philosophies of
Human Nature Scale (Wrightsman, 1974); Interper-
sonal Trust, the Interpersonal Trust Scale (Rotter,
1967); Network Trust, the Network Trust Scale of
the Trust Inventory; Generalized Trust, the General-
ized Trust Scale of the Trust Inventory.

of the interpretations of the Trust Inventory Scales and its tripartite classifi-
cation. This is particularly so for Partner Trust which yielded the strongest
pattern of both convergent and discriminant validity coefficients and the most
consistent and interpretable pattern of construct validity evidence. Specifi-
cally, concurrent and discriminant validity of the Partner Trust Scale was
supported in that the scale correlated more strongly with other measures of
relational trust than with measures of global trust. This seems to indicate
that, as conceptualized, Partner Trust scores reflect respondents’ relation-
ships with a specific person rather than a generalized orientation toward peo-
ple or human nature. Analyses involving the Generalized Trust Scale con-
firmed expectations also although not as clearly as was the case for Partner
Trust. The Generalized Trust Scale showed higher correlations with other
global trust measures than with relational trust measures. This, too, supports
the conceptualization of Generalized Trust as a distinct type of trust. These
data then supported the utility of the Partner Trust and Generalized Trust
 LEVELS OF TRUST 333

Scales of the Trust Inventory. Thus, the Trust Inventory is the first measure
available which includes valid scales reflecting both of the predominate mod-
els of trust described in the literature, namely, global trust and relational
trust.
Second, although there were fewer opportunities to confirm the utility of
Network Trust because of the uniqueness of its conceptualization, some sup-
portive evidence was observed here as well. Moderate correlations were ex-
pected between the Network Trust Scale and both relational trust measures
and global trust measures. This was based on the assumption that network
trust occupied an intermediate position on a hypothetical continuum from
intimate partners to human nature. In fact, the Network Trust Scale showed
significant, but moderate correlations with measures of both relational and
global trust. This finding supported the utility of the Network Trust Scale.
However, additional research is needed to determine whether Network Trust
is distinct from Partner and Generalized Trust and also whether the indepen-
dent assessment of network trust adds substantive information regarding the
nature and importance of trust in relationships and interpersonal processes.
Third, the three scales of the Trust Inventory correlated differentially, but
in predictable ways, with measures of relationship experiences versus mea-
sures of personality and emotions and this also supports their validity. Spe-
cifically, it was expected that Partner Trust would be most strongly related
to indices of relationship satisfaction and experience, whereas Generalized
Trust would be more strongly related to trait/emotion measures and this pat-
tern of results was generally observed. Also, it seemed a reasonable expecta-
tion that Generalized Trust would differentiate among correlates of trust less
strongly than Partner Trust because it is presumed to be a measure of a broad
and global trait, unrelated to any specific situation nor relationship.
The Trust Construct
In addition, these results supported three conclusions with respect to the
construct of trust independent of the specific scale used to measure it. First,
relational and global trust are related, but distinct constructs. This distinction
had been cited in the literature previously and is further substantiated by
comparing scores among a sample of measures representing the relational—
global distinction. These analyses also found that measures of relational trust
were considerably more strongly related to each other on average than on
average to measures of global trust. Correspondingly, measures of global
trust were more strongly related to one another on average than their average
statistical relationship with measures of relational trust, although the perfor-
mance of the global trust measures was less impressive in this regard.
Second, the average correlations of each of the relational trust measures
was more closely correlated with the relationship variables than to the mea-
sures of traits and emotions. Conversely, global trust measures appeared to
334 COUCH AND JONES

be more closely related to the average of the trait/emotion measures, but

this difference failed to satisfy the criterion for significance. The differential
pattern of correlations for global and relational trust was expected because
global trust frequently has been conceptualized as a personality trait, whereas
relational trust often has been defined as an outcome of a specific relation-
ship. Taken together, these findings at least partially supported the need for
differentiating between the different types or levels of trust. Additional re-
search is needed, however, to explore the extent to which, for example, rela-
tional trust reflects a tendency or expectation of a relational partner that ex-
isted prior to a given relationship or the extent to which global trust is
enhanced or inhibited by specific experiences with other people.
Third, within categories there was little evidence favoring one measure
over an alternative. On the contrary, separate measures of relational trust
appeared to be virtually interchangeable, whereas measures of global trust
were less strongly intercorrelated. The implications that result from differing
studies using divergent measures of relational trust can be combined in litera-
ture reviews and that one may generalize from one study to the next. Evi-
dence is less convincing that studies utilizing various measures of global
trust may be combined. On the other hand, there are advantages to using the
Trust Inventory relative to other measures; specifically, it is the only measure
incorporating scores reflecting both relational and global trust; and it is the
only measure that assesses the domain of network trust.

REFERENCES
Buss, A. H., & Durkee, A. (1957). An inventory for assessing different kinds of hostility.
Journal of Consulting Psychology, 21, 343–349.
Chaikin, A. L., & Derlega, V. J. (1976). Self-disclosure. In J. W. Thibaut, J. T. Spence, &
R. C. Carson (Eds.), Contemporary topics in social psychology. Morristown, NJ: General
Learning Press.
Chun, K., & Campbell, J. B. (1975). Notes on the internal structure of Wrightsman’s measure
of trustworthiness. Psychological Reports, 37, 323–330.
Couch, L. (1994). The development of the Trust Inventory. Unpublished master’s thesis, Uni-
versity of Tennessee, Knoxville, TN.
Couch, L. L., Adams, J. M., & Jones, W. H. (1996). The Assessment of Trust Orientation.
Journal of Personality Assessment, 67, 305–323.
Deutsch, M. (1958). Trust and suspicion. Conflict resolution, 2, 265–279.
Driscoll, J. W. (1978). Trust and participation in organizational decision making as predictors
of satisfaction. Academy of Management Journal, 21, 44–56.
Erikson, E. H. (1950). Childhood and society. New York: Norton.
Erikson, E. H. (1964). Childhood and society. (2nd ed.). New York: Norton.
Giffin, K. (1967). The contribution of studies of source credibility to a theory of interpersonal
trust in the communication process. Psychological Bulletin, 68, 104–120.
Hendrick, C., & Hendrick, S. (1983). Liking, loving and relating. Monterey, CA: Brooks/
Cole.
Hendrick, C., & Hendrick, S. (1986). A theory and method of love. Journal of Personality
and Social Psychology, 50, 392–402.
 LEVELS OF TRUST 335

Holmes, J. G. (1991). Trust and the appraisal process in close relationships. In W. H. Jones &
D. Perlman (Eds.), Advances in personal relationships (pp. 57–106). London: Jessica
Kingsley Publishers.
Holmes, J. G., & Rempel, J. K. (1989). Trust in Close Relationships. In C. Hendrick (Ed.),
Close relationships (pp. 187–220). Newbury Park, CA: Sage.
Johnson-George, C., & Swap, W. (1982). Measurement of specific interpersonal trust: Con-
struction and validation of a scale to assess trust in a specific order. Journal of Personality
and Social Psychology, 43, 1306–1317.
Jones, W. H., Adams, J. M., Monroe, P. R., & Berry, J. O. (1995). A psychometric exploration
of marital satisfaction and commitment. Journal of Social Behavior and Personality, 10,
923–932.
Jones, W. H., & Briggs, S. R. (1986). Manual for the social reticence scale. Palo Alto, CA:
Consulting Psychologists Press.
Jones, W. H., Briggs, S. R., & Smith, T. G. (1986). Shyness: Conceptualization and measure-
ment. Journal of Personality and Social Psychology, 51, 629–639.
Jones, W. H., & Burdette, M. P. (1994). Betrayal in close relationships. In A. L. Weber &
J. Harvey (Eds.), Perspectives on close relationships. New York: Allyn & Bacon.
Jones, W. H., Cohn, M. G., & Miller, C. E. (1991). Betrayal among children and adults. In
K. J. Rotenberg (Ed.), Children’s interpersonal trust: Sensitivity to lying, deception, and
promise violations (pp. 118–134). New York: Springer-Verlag.
Jones, W. H., & Kugler, K. (1993). Interpersonal correlates of the Guilt Inventory. Journal
of Personality Assessment, 61, 246–258.
Kee, R., & Knox, R. (1970). Conceptual and methodological considerations in the study of
trust and suspicion. Journal of Conflict Resolution, 14, 357–366.
Kugler, K., & Jones, W. H. (1992). On conceptualizing and assessing guilt. Journal of Person-
ality and Social Psychology, 62, 318–327.
Lazelere, R. E., & Huston, T. L. (1980). The dyadic trust scale: Toward understanding interper-
sonal trust in close relationships. Journal of Marriage and the Family, 42, 595–604.
Lee, J. A. (1973). The colors of love: An exploration of the ways of loving. Don Mills, Ontario:
New Press.
Mathes, E. W., & Severa, N. (1981). Jealousy, romantic love, and liking: Theoretical consider-
ations and preliminary scale development. Psychological Reports, 49, 23–31.
Pereira, M. J., & Austrin, H. R. (1980). Interpersonal trust as a predictor of suggestibility.
Psychological Reports, 47, 1031–1034.
Raciti, M., & Hendrick, S. S. (1992). Relationships between eating disorder characteristics
and love and sex attitudes. Sex Roles, 27, 553–564.
Rempel, J. K., Holmes, J. G., & Zanna, M. P. (1985). Trust in close relationships. Journal
of Personality and Social Psychology, 49, 95–112.
Rotenberg, K. J. (1984). Sex differences in children’s trust in peers. Sex Roles, 11, 953–957.
Rotter, J. B. (1967). A new scale for the measurement of interpersonal trust. Journal of Person-
ality, 35, 651–665.
Rotter, J. B. (1971). Generalized expectancies for interpersonal trust. American Psychologist,
26, 443–452.
Rotter, J. B. (1980). Interpersonal trust, trustworthiness, and gullibility. American Psycholo-
gist, 35, 1–7.
Rubin, Z. (1970). Measurement of romantic love. Journal of Personality and Social Psychol-
ogy, 16, 265–273.
Russell, D., Peplau, L., & Cutrona, C. (1980). The revised UCLA loneliness scale: Concurrent
and discriminant validity evidence. Journal of Personality and Social Psychology, 39,
472–480.
Tardy, C. H. (1988). Interpersonal Evaluations: Measuring attraction and trust. In C. H. Tardy
336 COUCH AND JONES

(Ed.), For the study of human communication: Methods and instruments for observing,
measuring, and assessing communication processes (pp. 269–284). Norwood, N. J.:
Ablex Publishing.
Wright, T. L., & Kirmani, A. (1977). Interpersonal trust, trustworthiness, and shoplifting in
high school. Psychological Reports, 41, 1165–1166.
Wrightsman, L. S. (1964). Measurement of philosophies of human nature. Psychological Re-
ports, 14, 743–751.
Wrightsman, L. S. (1974). Assumptions about human nature: A social–psychological ap-
proach. Monterey, CA: Brooks Cole.