Copyright of the Academy of Management, all rights reserved.

Contents may not

Academy of Management Review be copied, emailed, posted to a listserv, or otherwise transmitted without the
2007, Vol. 32, No. 2, 344354. copyright holders express written permission. Users may print, download, or email
articles for individual use only.

EDITORS FORUM

AN INTEGRATIVE MODEL OF
ORGANIZATIONAL TRUST: PAST, PRESENT,
AND FUTURE
F. DAVID SCHOORMAN
Purdue University

ROGER C. MAYER
The University of Akron

JAMES H. DAVIS
University of Notre Dame

A considerable amount of research has examined trust since our 1995 publication. We
revisit some of the critical issues that we addressed and provide clarifications and
extensions of the topics of levels of analysis, time, control systems, reciprocity, and
measurement. We also recognize recent research in new areas of trust, such as affect,
emotion, violation and repair, distrust, international and cross-cultural issues, and
context-specific models, and we identify promising avenues for future research.

As we wrote our 1995 paper on trust (Mayer, search that our paper has been cited over 1,100
Davis, & Schoorman, 1995), we were struck by times (according to Google Scholar). In addition
the relative scarcity of research in the main- to management and general business, it has
stream management literature focusing directly been cited in such diverse areas as marketing,
on trust. This led us to several bodies of litera- accounting, finance, economics, information
ture, including management, psychology, phi- systems, industrial engineering, political sci-
losophy, and economics. We found that scholars ence, communication, ethics, law, psychology,
from diverse disciplines were presenting many sociology, health care, and agribusiness. We
insightful views and perspectives on trust but would like to use this opportunity to revisit some
that many of them seemed to talk past one an- of the issues raised by our 1995 paper and re-
other. Our goal was to integrate these perspec- view how the field has dealt with them. We will
tives into a single model. also discuss the new concerns and opportunities
This work came to fruition at about the same for future research on trust.
time as several other works on trust. Papers on
trust by Hosmer (1995) and McAllister (1995) were
also published in Academy of Management CLARIFICATIONS AND EXTENSIONS OF THE
journals that year, followed the next year by a MODEL OF TRUST
book edited by Kramer and Tyler (1996). The con-
fluence of these works, fueled by practical con- Trust As an Aspect of a Relationship
cerns raised by now infamous government and One of the difficult conceptual decisions that
corporate scandals over the next decade, pro- we faced as we developed our definition of trust
duced a groundswell of interest in understand- was to break with the widely accepted ap-
ing this basic and ubiquitous construct. proach, to that point, that trust was dispositional
Since we were drawing perspectives from and trait-like and to argue that trust was an
multiple disciplines as inputs to the model, we aspect of relationships. That meant it varied
wanted to provide a model that was generally within person and across relationships. With
applicable and would be used across multiple some exceptions (e.g., Driscoll, 1978; C. L. Scott,
disciplines. We were gratified to find in a recent 1980), the dominant conceptual and operational
344
2007 Schoorman, Mayer, and Davis 345

definition of trust in the literature was Rotters research points out that trust should be exam-
(1967). We then went the next step and included ined at both the macro and micro levels within
ability as an antecedent of trust that allowed a an organization (McEvily, Perrone, & Zaheer,
partys trust to vary within a given trustee but 2003; Zaheer, McEvily, & Perrone, 1998). While
across domains. The dispositional aspects of the need to understand trust has been noted in
trust considered by Rotter are contained in the areas of study both within and between organi-
construct of propensity to trust in our model. The zations, methodological difficulties can arise in
literature that has followed our model has not the absence of a clear multilevel conceptual
questioned this decision and has accepted the model (Mossholder & Bedeian, 1983; Rousseau,
view that trust is based in relationships. 1985).
Just as perceptions about an individuals abil-
ity, benevolence, and integrity will have an im-
Application Across Levels of Analysis
pact on how much trust the individual can gar-
The importance of multilevel and cross-level ner, these perceptions also affect the extent to
perspectives is gaining increasing attention in which an organization will be trusted. We de-
organizational research. This has led to a call fined each of these trustworthiness dimensions
for examining trust across levels of organization- so that it could be applied to interpersonal, in-
al analysis (e.g., Rousseau, Sitkin, Burt, & Cam- tergroup, or interorganizational levels of analy-
erer, 1998). There is a need to understand trust sis.
both within and between organizations because At higher levels of analysis, such as between
methodological difficulties can arise in the ab- organizations, viewing the trustee in terms of
sence of a clear multilevel conceptual model ability and integrity seems to be well accepted.
(Currall & Inkpen, 2002; Mossholder & Bedeian, At macro levels of analysis, however, benevo-
1983; Rousseau, 1985). lence has received little attention. We defined
We have heard from a number of scholars that benevolence as the extent to which a party is
the 1995 framework is fairly robust across levels believed to want to do good for the trusting
of analysis. A bit of history on the development party, aside from an egocentric profit motive.
of our theory may shed light on this issue. Early Does the other company hold the focal compa-
drafts of our paper developed our trust model nys best interests as highly important? While
across multiple levels of analysis. One of our we may be able to identify situations, such as
initial goals was to develop a theory that would sole proprietorships, where the owners have
be applicable across levels of analysis. We were strong bonds that display significant benevo-
careful to develop constructs that would cross lence toward one another, the more traditional
levels of analysis, and we developed examples mode is probably one wherein each company is
of how cross-level applications of the model motivated primarily by its own financial inter-
would work. Perhaps it was fortunate that early ests. If this is indeed the norm, benevolence is
reviewers of our paper made the accurate obser- not likely to be the most important factor in the
vation that the paper was very cumbersome development of interorganizational trust. How-
(and long) because it developed the multilevel ever, acts of benevolence (e.g., allowing bench-
model. They recommended that we restrict our marking) from a potential partner in a joint ven-
paper to a single level. The fact that our initial ture would help to build trust.
goal was to develop a multilevel theory is prob- We contend that all three factors of ability,
ably why the model works as well as it does benevolence, and integrity can contribute to
across levels, but we do agree with those who trust in a group or organization. Consider, for
argue that one of the weaknesses in much of the instance, a supplier-buyer relationship. The
current trust research is that it is limited to re- buyer may believe that a supplier is able to
lationships at a single level of analysis, consid- provide a quality product in a timely fashion.
ering either dyadic trust relationships within However, this only assures that the supplier
organizations or trust between organizations. could perform. This does not mean that it will
Several authors have recognized differences perform, and, therefore, the supplier will not
in trust for single referents at different hierar- necessarily be trusted. The perception that the
chical levels within an organization (e.g., Cook supplier has integrity suggests that it will fulfill
& Wall, 1980; Driscoll, 1978; D. Scott, 1980). Recent agreements as promised. Yet even if there is an
346 Academy of Management Review April

agreement, if the suppliers ability to deliver is time as the relationship between the parties de-
questionable, it will not be trusted. If the sup- velops (1995: 722).
plier is perceived as benevolent, it will have a Despite these assertions, in many empirical
strong desire to serve this particular buyers studies researchers have raised questions about
needs. If the suppliers integrity is suspect be- the high observed correlation between benevo-
cause, for instance, its track record with other lence and integrity and have questioned the in-
firms is inconsistent with its stated policies, dependence of these variables. In a discussion
trust will again be lacking. As the perception of of several empirical studies, Schoorman (2002)
each of these factors increases, we would expect observed that the findings as a whole were com-
an increase in willingness to take a risk in the pletely consistent with the model. Those studies
relationship. conducted in laboratory settings were more
The trust of either the dominant coalition or likely to show a high correlation between be-
the management team is critical to understand- nevolence and integrity because the relation-
ing organizational trust, since it is this level of ships had not had time to develop any real data
trust that will govern the strategic actions of the about benevolence. In field samples where the
organization (Cyert & March, 1963; Simon, 1957). parties had longer relationships, benevolence
As with individuals, we propose that some or- and integrity were more likely to be separable
ganizations develop greater propensities to factors. We continue to find this pattern to be
trust than do others. For organizations, these consistent in our research. We think it would be
propensities develop from geographic, industry, interesting for future research to establish more
and economic histories. A series of previous in- specifically the process and time frames in
teractions with other organizations that resulted which each of the variables contributes to trust.
in, for example, lawsuits or monetary losses
would lower an organizations propensity to
Trust, Risk, and Control Systems
trust. Conversely, a series of such experiences
as mutual benchmarking with various organiza- In our model we argued that trust would lead
tions that significantly improved the quality to risk taking in a relationship (see Proposition
processes for an organization would increase its 5). Perceived risk moderates the relationship be-
trust propensity. tween trust and risk taking in our model. Trust is
In summary, groups and organizations can the willingness to take risk, and the level of
both garner trust from other parties and trust trust is an indication of the amount of risk that
other parties. Our model was designed to under- one is willing to take. Clearly, control systems
stand the major factors that explain trust from are an alternate mechanism for dealing with
not only the individual level but from the group risk in relationships. Recently, several scholars
and organizational perspectives as well. have speculated about the relationship between
trust and control systems in dealing with risk
(McEvily et al., 2003; Sitkin & George, 2005).
The Time Dimension
Our views on this issue are developed further
One of the issues explicit in our theory was in Davis, Schoorman, and Donaldson (1997), in
that time would play an important role in the which we argue that one of the major distinc-
meaningfulness of the variables in the model. tions between agency theory and stewardship
We noted that propensity, as a dispositional theory is the use of trust versus control systems
quality, would be an important factor at the very to manage risk. However, we do not see these
beginning of the relationship. We also noted mechanisms as being mutually exclusive. On
that judgments of ability and integrity would the contrary, when the risk in a situation is
form relatively quickly in the course of the rela- greater than the trust (and, thus, the willingness
tionship and that benevolence judgments would to take risk), a control system can bridge the
take more time. Proposition 3 states that the difference by lowering the perceived risk to a
effect of integrity on trust will be most salient level that can be managed by trust. For exam-
early in the relationship prior to the develop- ple, in an organization that has a culture of
ment of meaningful benevolence data (1995: open book management and transparency in
722); Proposition 4 states that the effect of per- numbers (a control system), the levels of per-
ceived benevolence on trust will increase over ceived risk may be lower. There is a greater
2007 Schoorman, Mayer, and Davis 347

opportunity to empower employees by trusting has been a different story. We defined trust as a
them to manage larger budgets and for employ- willingness to be vulnerable to another party.
ees to trust the supervisor that performance- As such, suitable measurement of the construct
based compensation is fair. necessitates that questions be asked that assess
However, there is an important caveat that the extent to which a trustor is willing to volun-
must be noted. If there is a very strong system of tarily take risks at the hands of the trustee.
controls in an organization, it will inhibit the We developed a short, four-item measure,
development of trust. Not only will there be few with each of the items tapping into how willing
situations where there is any remaining per- the trustor was to be vulnerable to the trustee.
ceived risk but trustworthy actions will be at- We (Schoorman, Mayer, & Davis, 1996a) found
tributed to the existence of the control system that veterinary doctors took bigger risks with
rather than to the trustee (cf. Strickland, 1958). those employees they trusted more. The impact
Thus, a trustees actions that should be inter- of trust went beyond that explained by the abil-
preted as driven by benevolence or by integrity ity, benevolence, and integrity of the trustee.
may be viewed simply as responses to the con- Despite the measures brevity, we found its in-
trol systems. The use of control systems is how ternal consistency strong (Cronbachs alpha
agency theory proposes dealing with risk man- .82).
agement, and this does not foster the develop- We (Davis, Schoorman, Mayer, & Tan, 2000)
ment of trust. used the same measure in a restaurant setting
to measure the trust that employees had in their
general manager and found an alpha of .62. This
The Reciprocity of Trust
measure of trust in the leader significantly pre-
One of the limitations of our model that we dicted subsequent sales, profits, and employee
noted in the conclusion of our 1995 paper was turnover in the restaurants (Davis, Mayer, &
that our conceptualization was unidirectional. Schoorman, 1995; Davis et al., 2000). Based on
We did not explore the reciprocity in trusting these results, we concluded that if trust in the
relationships. This is a particularly salient issue general manager could be developed and sus-
in the area of leader-subordinate relationships, tained, it would be a significant competitive ad-
since the dominant view among leadership the- vantage to the firm, and the framework, includ-
orists is that leader-member exchange (LMX) is ing ability, benevolence, and integrity, merited
mutual and reciprocal (Graen & Uhl-Bien, 1995; further consideration as an approach to build-
Liden, Wayne, & Stillwell, 1993). In an extension ing trust in management.
of our model, Brower, Schoorman, and Tan (2000) Using this same measure, a quasi-experiment
argued that, unlike LMX, trust is not necessarily (Mayer & Davis, 1999) showed that trust in top
mutual and is not reciprocal. One of the impli- management was significantly improved by
cations of this argument is that, in a relation- identifying and replacing an invalid appraisal
ship, A can trust B, but B may not trust A. This is system. While the alpha in this study was lower
completely consistent with the approach to trust (i.e., .59 and .60 in two waves of data), the test-
and trust formation that we presented in our retest reliability was quite strong, at .75 over
model but is inconsistent with the views in the five months and .66 over nine months. Further-
leadership literature. Empirical studies that ex- more, the quasi-experimental results were sig-
amine this reciprocal linkage of how one partys nificant, even though the sizes of the groups
trust affects the other partys trust in return, being compared were modest (i.e., twenty-two
rather than assuming them to be equivalent, are and fifty-seven). In additional analyses we
rare (Serva, Fuller, & Mayer, 2005). We feel that found an average interitem correlation of r .32.
this presents a fruitful area for future research. For each item we calculated a correlation be-
tween it and a composite of the other three
items. These correlations ranged from .23 to .49,
Measurement of Trust As Willingness to Be
with an average of .38.
Vulnerable
These results fall well within Klines (1986)
While the theory of trust described in the 1995 description of a measure of a complex construct
paper has been very influential in the develop- that has maximum validity. Kline noted that
ment of trust research, the measurement of trust such a measure would only have low internal
348 Academy of Management Review April

consistency reliability. Internal consistency es- than desirable alpha levels of our original mea-
timates suffer because of the brevity of this mea- sure. While we think that a combination of its
sure. As Cortina says, [Alpha] must be inter- conceptual clarity, test-retest reliability, and re-
preted with the number of items in mind (1993: lationship with other variables in the nomolog-
102). If the measure were truly unreliable, we ical net across a number of studies is just as
would not have been able to attain the signifi- important a consideration as the alpha level of a
cant results we have repeatedly found with its four-item measure, we recognize that develop-
use (Nunnally, 1978). The results of these studies ment of the more recent longer measures with
all support the idea that while our four-item higher alpha levels was warranted. The mea-
measure of willingness to be vulnerable has at sures we have developed more recently appear
times had lower than desired levels of Cron- to have overcome concerns with Cronbach al-
bachs alpha, it is a robust, stable, and valid pha levels. We believe that as these measures
measure of the construct as we defined it. become more widely available, more research-
More recent efforts to develop the measure ers will choose to measure trust as defined in
have yielded higher alpha levels. Mayer and our 1995 paper.
Gavin (2005) expanded the measure to ten items, On a related note, work on trust would be
yielding alphas of .82 and .76 for the plant man- facilitated by further development of measures
ager and the top management team, respec- of propensity. Many researchers have found Rot-
tively. Further analysis revealed two factors, ters (1967) twenty-five-item measure too long to
however. One factor consisted of the original include as a variable in studies with many other
measure with an additional item, whereas the variables and are concerned about its multidi-
second factor consisted of five items that gener- mensionality. Our adaptation (Mayer & Davis,
ally cited willingness to engage in specific be- 1999; Schoorman et al., 1996a), while much
haviors that would put a trustor at risk, such as shorter and unidimensional, has not consis-
communicating sensitive information to the tently produced high Cronbachs alphas (e.g., .55
trustee. The five-item general willingness to be and .66 in Mayer & Davis, 1999). We are not
vulnerable scale yielded an alpha of .81 for the aware of any brief, unidimensional published
focal plant manager and .72 for the top manage- measure of propensity that produces consis-
ment team. Both of these were improvements of tently high alpha levels. Development of such a
.06 and .07, respectively, over the four-item mea- measure might enable finding more relation-
sure. While results using the five-item scale ships between propensity and other variables of
were reported in that study, the ten-item scale interest, particularly early in the development of
yielded nearly identical results. a relationship.
In what may be the most promising measure
to date, Schoorman and Ballinger (2006) ex-
panded the original measure to seven items, NEW DIMENSIONS IN THE RESEARCH ON
taking care to maintain the conceptual defini- TRUST
tion but not create redundant items. It has pro-
Affect, Emotion, and the Impact on Trust
duced an alpha level of .84 in a sample of vet-
erinary hospital employees. This measure The basis of our model was to understand how
appears in the Appendix. parties process information about others,
Gillespie (2003) developed and validated a thereby deciding how much risk to take with
measure of trust based on the willingness to be those others. Perceptions of others and percep-
vulnerable definition. This ten-item Behavioral tions of risk inherent in the behaviors being
Trust Inventory has good psychometric proper- considered must be processed in order to come
ties and shows promise for future research to decisions about taking risks. For instance, as
based on this conceptual definition. one evaluates a trustees ability in the domain
The upshot of this discussion is that measur- of interest or considers relevant inputs about a
ing trust as we defined it involves asking ques- trustees integrity, one is thinking. As such, our
tions that measure a trustors willingness to be model represents a cognitive approach to trust.
vulnerable. We are aware from numerous per- More recent work has pointed to the fact that
sonal communications from other researchers trust also involves emotion. Williams (2001) has
that many of them are concerned with the lower pointed out that affective responses influence
2007 Schoorman, Mayer, and Davis 349

how people evaluate their level of trust in an- ative responses. Dating back to the 1950s (e.g.,
other party. Similarly, Jones and George (1998) Kruglanski, 1970; Strickland, 1958), some work
have argued that emotions and moods provide has been done that begins to lay the ground-
people with information on how they are expe- work for how attributions are made when trust is
riencing trust. Dunn and Schweitzer (2005) found broken. We suggest that which trustworthiness
that emotional states even unrelated to the factor is damaged and how it was damaged
trustee or the situation have an effect on trust. influence not only how repairable the damage is
Weber, Malhotra, and Murnighan (2005) have likely to be but how effective various repair
shown that emotional attachments can cause a strategies are likely to be. Repair attempts
trustor to take a sudden risk not warranted by might be looked at either as mediating pro-
the available evidence. cesses within the feedback arrow in our model
Proponents of the strictly cognitive approach or as influencing the trustworthiness factors di-
to decision making about trust would argue that rectly (e.g., receiving information about trust-
while emotions may create a temporary irra- worthiness factors outside the process of having
tionality about the data on ability, benevo- taken a risk that went awry). Further theoretical
lence, and integrity, after a period of time the work in this area to understand the conditions
perception would return to a rational perspec- under which various repair strategies are effec-
tive. Nonetheless, it appears to be clear that tive would help in developing prescriptive road-
emotions do influence the perception of the an- maps for repairing broken trust.
tecedents of trust and, therefore, the trust in re- Forgiveness is another evolving area that
lationships. It is also likely that this emotion holds promise for understanding trust repair af-
does dissipate over time after a violation of ter a violation. An insightful paper by Aquino,
trust. What is not clear is whether it ever com- Grover, Goldman, and Folger (2003) provides an
pletely dissipates and returns to a nonemotional important separation between resolving nega-
evaluation. Alternatively, while emotions are tive emotions (i.e., forgiveness itself) and behav-
being experienced, they may lead the trustor to iors that follow that restore the relationship.
update prior perceptions of the trustworthiness More work in this area is needed to determine
dimensions and trust such that even after the such basic issues as the conditions under which
emotions dissipate, the effect on the cognitive forgiveness enhances trust repair after a viola-
evaluations remains. We think the role of emo- tion, what conditions increase the likelihood
tions is a very interesting area of research and that an offended party will forgive the violator,
will add a new dimension to the model. and the role that forgiveness plays in the trust
repair process.
Violation and Trust Repair
Some exemplary work has been done to un-
The Concept of Distrust
derstand trust violation and repair (e.g., Lewicki
& Bunker, 1996; Morrison & Robinson, 1997; Rob- There has been considerable discussion in the
inson, 1996), but this topic warrants more re- literature about the concept of distrust, as well
search. As part of their contributions, these au- as the relationship between trust and distrust.
thors highlight the idea that violation of trust is Lewicki, McAllister, and Bies (1998) framed the
likely to be an emotional event for the trustor. At debate in the organizational literature, arguing
the time of this writing, a special issue of the that trust and distrust are separate dimensions
Academy of Management Review (edited by and not the opposite ends of a single continuum.
Dirks, Lewicki, and Zaheer) focused on trust re- They noted that two-factor models of satisfac-
pair is in preparation. This is a very appropriate tion and dissatisfaction had been proposed be-
topic and promises to add valuable insight into fore (Herzberg, Mausner, & Snyderman, 1967),
the process by which trust development can and that the contemporary evidence from stud-
move forward after trust has been damaged. ies of positive and negative affectivity supports
We believe that in order to repair trust, it is this view. Their main reason for suggesting such
critical to first understand how it was damaged an approach to the trust-distrust distinction was
in the first place, since different means of dam- because relationships are multifaceted or mul-
aging trust are likely to require different repar- tiplex (1998: 442), and we need a model that
350 Academy of Management Review April

allows for both trust and distrust to exist in the that you dont need different constructs to ac-
same relationship. count for this.
In our model we chose to take the opposite McKnight and Chervany (2001) produced an
(and more traditional) view that trust and dis- excellent summary of the literature on defini-
trust are the opposite ends of the same contin- tions of trust and distrust and the models that
uum. This is consistent with dictionary defini- describe each of the constructs. They reviewed
tionsfor example, Websters defines distrust the literature on trust and distrust and devel-
as the lack or absence of trust and Random oped separate conceptual models (antecedent
House as to have no trust in. In sociology, and contextual variables) for each construct.
Ross, Mirowski, and Pribesh define mistrust as The resulting models are identical for both trust
the absence of faith in other people (2001: 568). and distrust, which suggests to us that perhaps
Luhmann argues that distrust is a functional we do not need both models. In fact, these au-
equivalent of trust (1979: 71). Our definition of thors conclude that most trust theorists agree
trustwillingness to take risk (i.e., be vulnera- that trust and distrust are separate constructs
ble) in a relationshipmeans that at the lowest that are opposites of each other (2001: 42). We
level of trust, one would take no risks at all. We would simply add that if they are opposites of
felt that the complete lack of trust and distrust each other, there is little added value to treating
are the same thing. them as separate constructs.
In our model of trust, however, we argued that A review of empirical work on the conceptual-
ability is an important antecedent of trust, along ization of distrust in the literature produced sur-
with benevolence and integrity. This was a de- prising results. Some who argue that trust and
distrust are different dimensions only study one
viation from an emerging view that trust was
of the constructs at a time, making it difficult to
more affective. This point is further clarified in
develop data on the differences. It is particularly
Schoorman, Mayer, and Davis (1996b). The im-
interesting that some researchers who study the
portant implication of the addition of ability to
concept of distrust (e.g. McAllister, Pang, Tan, &
the antecedents of trust is that it creates a
Ruan, 2006) have used our measure of trust as a
framework of trust that is domain specific. This
willingness to be vulnerable (Mayer & Davis,
is noted in our original article (1995: 717) and
1999; Schoorman et al., 1996b) and reverse-
follows from the work of Zand (1972). Trust being
scored it to represent their measure of distrust.
domain specific allows for the multifaceted and
In sum, we can find no credible evidence that a
multiplex relationships about which Lewicki et concept of distrust that is conceptually different
al. (1998) raised concerns. For example, it may be from trust is theoretically or empirically viable.
appropriate to trust a colleague to do a good job
collaborating on a research project but to not
trust him/her to do a good job teaching your International and Cross-Cultural Implications
class in your absence. The difference in the level for Trust
of trust within the same relationship is a func- Over the same time frame in which interest in
tion of the different abilities across different do- trust accelerated, there was a significant in-
mains. The skills required to present and inter- crease in interest in studying cross-national and
act effectively in class differ from those cross-cultural differences. It is therefore not sur-
necessary to do research. prising that much of the explosion of interest in
Lewicki et al. (1998) produced a chart with a trust research has come from around the globe.
high trust and high distrust condition in which Over 20 percent of the 1,100 studies listed in
one would presumably trust, but verify. We Google Scholar that cite our paper were written
feel this is not a reasonable argument within in a language other than English. The World
domain. If you trust a partner, you do not need to Economic Forum is made up of the worlds lead-
verify. Doing so would be the clearest indication ing politicians and business leaders. It meets
that you do not trust. We do agree that you might annually in Davos, Switzerland to discuss a
trust your colleague to produce a literature re- wide range of international issues. It has been
view but may need to verify his/her ability to monitoring public trust levels since 2003 through
deliver in the classroom by reviewing his/her a biannual global public opinion poll conducted
lecture notes and presentation. Our model says by GlobeScan Incorporated. The latest findings
2007 Schoorman, Mayer, and Davis 351

from the poll show that trust in a range of insti- laborative, being-oriented, feminine cultures
tutions has dropped significantly since January tend to put more emphasis on the benevolence
2004 to levels not seen since the months follow- variable. While these are broad generalizations
ing the September 11 terrorist attacks in 2001 of relationships between culture and trust, they
(www.weforum.org). GlobeScan reports that the illustrate the potential value of future research
overall trust level for global companies is the to develop these links more carefully.
lowest since the tracking began. While we are
not sure what definition of trust was used in this
poll, these results present obvious reasons for
concern. Context-Specific Models of Trust
The recent GLOBE project by House and col- It was our intention in developing the model
leagues (House, Hanges, Javidan, Dorfman, & to be as parsimonious as possible and to de-
Gupta, 2004; House, Javidan, Hanges, & Dorf- velop a model that would be generalizable to
man, 2002) measured the cultural orientation of the broadest number of contexts. In order to
sixty-two societies around the world based achieve this, we neglected many specific con-
largely on the cultural dimensions identified by text variables that would be relevant to a more
Hofstede (1980). These scholars used nine di- restricted trust domain. We think it would be
mensions of culture, as well as twenty-one lead- appropriate to specify contextual variables for
ership dimensions. All of this research has led to the model that are unique to studying trust
the inevitable question of how trust is different within a particular context. For example, much
across cultures (Den Hartog, 2004; Wasti, Tan, of the research on trust in organizations has
Brower, & Onder, in press). focused on the relationship between supervisors
We believe that one of the ways in which and subordinates. In this context, the hierarchi-
culture affects trust is through the propensity
cal power difference and the asymmetry of in-
variable. We have proposed that the anteced-
formation that exist between the two individu-
ents of propensity include personality, experi-
als in the trusting relationship have some
ences, and culture. There is evidence in the cul-
important implications for how trust might de-
ture literature that initial trust of strangers
velop. If the supervisor has more access to infor-
varies across cultures. One of the dimensions of
mation about the subordinate and can initiate
culture that is most relevant to this issue is the
opportunities to gather information about abil-
task versus relationship orientation of a culture.
ity, benevolence, and integrity, and if these op-
Task-oriented cultures seem to have a higher
initial trust of strangers and therefore a higher portunities are not available to the subordinate,
propensity, while relationship-oriented cultures we would expect that the supervisors trust in
need time to develop a relationship prior to the subordinate would develop more quickly
working on the task. The cultural variable of than vice versa.
uncertainty avoidance is well-established as a Additionally, since risk taking in the relation-
predictor of predispositions to take risk or be ship is caused by an interaction between trust
risk averse (Sully de Luque & Javidan, 2004). If and risk, ones perceptions of risk in the action
trust is the willingness to take risk in a relation- being contemplated (which our model separates
ship, how does uncertainty avoidance as a dis- from the trustee in question) will affect risk-
positional quality affect the development of taking actions. Ceteris paribus, the party who
trust? We think there is considerable work that has more power in the relationship will likely
needs to be done in fine-tuning what we know perceive by virtue of that powerless risk
about the influence of culture on the propensity and, thus, will engage in more risk-taking ac-
to trust. tions. This would give the appearance that this
Culture can also affect the perception of abil- partys trust is higher still. In the context of trust
ity, benevolence, and integrity and the impor- between peers, there is likely to be a different
tance given to each of these variables in the set of variables that predict the development
model. More action-oriented, competitive, per- and use of trust. We expect that studies in par-
formance-oriented cultureswhat Hofstede has ticular contexts will develop additional vari-
called masculine culturestend to place a ables that help better explain the antecedents
higher value on the ability variable. More col- and consequences of trust.
352 Academy of Management Review April

SUMMARY AND CONCLUSIONS relative importance of ability, benevolence, and

integrity across cultures. Finally, we note that
In this paper we have taken the opportunity to
while we deliberately focused on a model that
reflect on our thinking as we developed the 1995
was maximally generalizable and parsimoni-
piece, An Integrative Model of Organizational
ous, it is appropriate to now examine context-
Trust, and the knowledge context in which it
specific variables that might add to the model of
occurred. This paper has given us an opportu-
trust. One such area that our research has taken
nity to revisit some of the decisions that we
us into is the specific relationship between su-
made and to evaluate them in light of the re-
pervisor and subordinate in the workplace. Is-
search that followed. A part of our charge in
sues of power and information asymmetry make
writing this paper was to indicate how our own this relationship and the trust it produces some-
research in this area has progressed and how what unique.
we view the growing body of literature on the While a great deal of research has occurred in
topic. the area of trust over the past decade, the new
The levels-of-analysis issue continues to be research only suggests that there is a lot more to
an interesting topic of discussion, and more ex- be done and many very promising avenues to
plicit extensions to group and organizational pursue. We hope that our comments in this pa-
levels are warranted. We note that time issues per can be a catalyst for some of this research.
addressed in our paper have not received the
attention that we expected. Research attempts
to study trust in very short laboratory simula-
tions have yielded mixed but not unexpected APPENDIX
results. By including a consideration of time, Trust Items from Schoorman and Ballinger
studies of trust should lead to more predictable (2006)
results. The interplay of trust, risk, and control
systems continues to be a much debated topic. My supervisor keeps my interests in mind when
We have attempted to clarify our thinking on making decisions.
I would be willing to let my supervisor have complete
these dynamics. We see trust and control sys- control over my future in this company.
tems as alternate and sometimes compatible If my supervisor asked why a problem occurred, I
means for managing risk. We have extended our would speak freely even if I were partly to blame.
thinking about the reciprocity of trust to explic- I feel comfortable being creative because my
supervisor understands that sometimes creative
itly recognize the notion that, unlike relational
solutions do not work.
leadership constructs (e.g., LMX), trust is not mu- It is important for me to have a good way to keep an
tual and not necessarily reciprocal. The mea- eye on my supervisor.
sures of trust that we had developed based on Increasing my vulnerability to criticism by my
our definition of trust have produced mixed re- supervisor would be a mistake.
If I had my way, I wouldnt let my supervisor have
sults in terms of internal consistency reliability any influence over decisions that are important to me.
estimates. Here we have explored some of the
developments and advances in the measure-
ment area and have noted the need for contin-
Response Scale
ued improvement.
We also have reviewed some of the interest- 1 2 3 4 5
ing new directions in the research on trust.
Prominent among these is the inclusion of the strongly somewhat neither somewhat strongly
disagree disagree agree agree agree
role of affect and emotion, trust violations, and nor
repair. We believe these constructs will add new disagree
dimensions to the model of trust and provide for
valuable research in the future. Another area
seeing rapid growth in interest is the role that REFERENCES
international and cross-cultural dimensions
Aquino, K., Grover, S. L., Goldman, B., & Folger, R. 2003. When
play in the model of trust. We see the greatest push doesnt come to shove: Interpersonal forgiveness
opportunities in the development of the concept in workplace relationships. Journal of Management In-
of propensity across cultures, as well as for the quiry, 12: 209 216.
2007 Schoorman, Mayer, and Davis 353

Brower, H. H., Schoorman, F. D., & Tan, H. H. 2000. A model of Jones, G. R., & George, J. M. 1998. The experience and evolu-
relational leadership: The integration of trust and lead- tion of trust: Implications for cooperation and teamwork.
er-member exchange. Leadership Quarterly, 11: 227250. Academy of Management Review, 23: 531546.
Cook, J., & Wall, T. 1980. New work attitude measures of trust, Kline, P. 1986. A handbook of test construction: Introduction
organizational commitment and personal need nonful- to psychometric design. London: Methuen.
fillment. Journal of Occupational Psychology, 53: 39 52.
Kramer, R. M., & Tyler, T. R. (Eds.). 1996. Trust in organiza-
Cortina, J. 1993. What is coefficient alpha? Journal of Applied tions: Frontiers of theory and research. Thousand Oaks,
Psychology, 78: 98 104. CA: Sage.
Currall, S. C., & Inkpen, A. C. 2002. A multilevel approach to Kruglanski, A. W. 1970. Attributing trustworthiness in super-
trust in joint ventures. Journal of International Business visor-worker relations. Journal of Experimental Psychol-
Studies, 33: 479 495. ogy, 6: 214 232.
Cyert, R. M., & March, J. G. 1963. A behavioral theory of the Lewicki, R. J., & Bunker, B. B. 1996. Developing and maintain-
firm. Englewood Cliffs, NJ: Prentice-Hall. ing trust in work relationships. In R. M. Kramer & T. R.
Davis, J. H., Mayer, R. C., & Schoorman, F. D. 1995. The trusted Tyler (Eds.), Trust in organizations: Frontiers of theory
general manager and firm performance: A strategic ad- and research: 114 139. Thousand Oaks, CA: Sage.
vantage. Paper presented at the Annual International Lewicki, R. J., McAllister, D. J., & Bies, R. J. 1998. Trust and
Strategic Management Society Conference, Mexico City. distrust: New relationships and realities. Academy of
Davis, J. H., Schoorman, F. D., & Donaldson, L. 1997. Toward Management Review, 23: 438 458.
a stewardship theory of management. Academy of Man- Liden, R. C., Wayne, S. J., & Stillwell, D. 1993. A longitudinal
agement Review, 22: 20 47. study on the early development of leader-member ex-
Davis, J. H., Schoorman, F. D., Mayer, R. C., & Tan, H. H. 2000. changes. Journal of Applied Psychology, 78: 662 674.
The trusted general manager and business unit perfor- Luhmann, N. 1979. Trust and power. New York: Wiley.
mance: Empirical evidence of a competitive advantage.
Mayer, R. C., & Davis, J. H. 1999. The effect of the performance
Strategic Management Journal, 21: 563576.
appraisal system on trust for management: A field qua-
Den Hartog, D. N. 2004. Assertiveness. In R. J. House, P. J. si-experiment. Journal of Applied Psychology, 84: 123
Hanges, M. Javidan, P. Dorfman, & V. Gupta (Eds.), Cul- 136.
ture, leadership and organizations: The GLOBE study of
Mayer, R. C., Davis, J. H., & Schoorman, F. D. 1995. An inte-
62 societies: 395 431. Thousand Oaks, CA: Sage.
grative model of organizational trust. Academy of Man-
Driscoll, J. W. 1978. Trust and participation in organizational agement Review, 20: 709 734.
decision making as predictors of satisfaction. Academy
Mayer, R. C., & Gavin, M. B. 2005. Trust in management and
of Management Journal, 21: 44 56.
performance: Who minds the shop while the employees
Dunn, J. R., & Schweitzer, M. E. 2005. Feeling and believing: watch the boss? Academy of Management Journal, 48:
The influence of emotion on trust. Journal of Applied 874 888.
Psychology, 88: 736 748.
McAllister, D. J. 1995. Affect- and cognition-based trust as
Gillespie, N. 2003. Measuring trust in work relationships: The foundations for interpersonal cooperation in organiza-
Behavioral Trust Inventory. Paper presented at the an- tions. Academy of Management Journal, 38: 24 59.
nual meeting of the Academy of Management, Seattle.
McAllister, D. J., Pang, K., Tan, H. H., & Ruan, Y. 2006. Social
Graen, G. B., & Uhl-Bien, M. 1995. Relationship-based ap- dynamics of paranoia and distrust in teams. Paper pre-
proach to leadership: Development of leader-member sented at the annual meeting of the Academy of Man-
exchange (LMX) theory of leadership over 25 years: Ap- agement, Atlanta.
plying a multi-level multi-domain perspective. Leader-
McEvily, B., Perrone, V., & Zaheer, A. 2003. Trust as an orga-
ship Quarterly, 6: 219 247.
nizing principle. Organization Science, 14: 91103.
Herzberg, F., Mausner, B., & Snyderman, B. B. 1967. The mo-
McKnight, D. H., & Chervany, N. L. 2001. Trust and distrust
tivation to work (2nd ed.). New York: Wiley.
definitions: One bite at a time. In R. Falcone, M. Singh, &
Hofstede, G. 1980. Cultures consequences. Beverly Hills, CA: Y.-H. Tan (Eds.), Trust in cyber-societies: 2754. Berlin &
Sage. Heidelberg: Springer-Verlag.
Hosmer, L. T. 1995. Trust: The connecting link between orga- Morrison, E. W., & Robinson, S. L. 1997. When employees feel
nizational theory and philosophical ethics. Academy of betrayed: A model of how psychological contract viola-
Management Review, 20: 379 403. tion develops. Academy of Management Review, 22:
House, R., Javidan, M., Hanges, P., & Dorfman, P. 2002. Un- 226 256.
derstanding cultures and implicit leadership theories Mossholder, K. W., & Bedeian, A. G. 1983. Cross-level infer-
across the globe: An introduction to project GLOBE. Jour- ence and organizational research: Perspectives on inter-
nal of World Business, 37: 310. pretation and application. Academy of Management Re-
House, R. J., Hanges, P. J., Javidan, M., Dorfman, P., & Gupta, view, 8: 547558.
V. 2004. Culture, leadership and organizations: The Nunnally, J. C. 1978. Psychometric theory. New York:
GLOBE study of 62 societies. Thousand Oaks, CA: Sage. McGraw-Hill.
354 Academy of Management Review April

Robinson, S. L. 1996. Trust and breach of the psychological Serva, M. A., Fuller, M. A., & Mayer, R. C. 2005. The reciprocal
contract. Administrative Science Quarterly, 41: 574 599. nature of trust: A longitudinal study of interacting
Ross, C. E., Mirowsky, J., & Pribesh, S. 2001. Powerlessness teams. Journal of Organizational Behavior, 26: 625 648.
and the amplification of threat: Neighborhood disadvan- Simon, H. 1957. Administrative behavior. New York: Macmil-
tage, disorder and mistrust. American Sociological Re- lan.
view, 66: 568 591.
Sitkin, S. B., & George, E. 2005. Managerial trust-building
Rotter, J. B. 1967. A new scale for the measurement of inter- through the use of legitimating formal and informal
personal trust. Journal of Personality, 35: 651 665. control mechanisms. International Sociology, 20: 307
Rousseau, D. M. 1985. Issues of level in organizational re- 338.
search: Multi-level and cross-level perspectives. Re- Strickland, L. H. 1958. Surveillance and trust. Journal of Per-
search in Organizational Behavior, 7: 137. sonality, 24: 200 215.
Rousseau, D. M., Sitkin, S. B., Burt, R. S., & Camerer, C. 1998. Sully de Luque, M., & Javidan, M. 2004. Uncertainty avoid-
Not so different after all: A cross-discipline view of trust. ance. In R. J. House, P. J. Hanges, M. Javidan, P. Dorfman,
Academy of Management Review, 23: 393 404. & V. Gupta (Eds.), Culture, leadership and organizations:
Schoorman, F. D. 2002. Discussant at symposium titled Inte- The GLOBE study of 62 societies: 603 644. Thousand
grating trust perspectives: Foundations for a revised in- Oaks, CA: Sage.
tegrative model of organizational trust. Annual meeting
Wasti, S. A., Tan, H. H., Brower, H. H., & Onder, C. In press.
of the Academy of Management, Denver.
Cross-cultural measurement of supervisor trustworthi-
Schoorman, F. D., & Ballinger, G. A. 2006. Leadership, trust ness: An assessment of measurement invariance across
and client service in veterinary hospitals. Working pa- three cultures. Leadership Quarterly.
per, Purdue University, West Lafayette, IN.
Weber, J. M., Malhotra, D., & Murnighan, J. K. 2005. Normal
Schoorman, F. D., Mayer, R. C., & Davis, J. H. 1996a. Empow- acts of irrational trust: Motivated attributions and the
erment in veterinary clinics: The role of trust in delega- trust development process. Research in Organizational
tion. Presented in a symposium on trust at the 11th Behavior, 26: 75101.
Annual Conference, Society for Industrial and Organi-
zational Psychology (SIOP), San Diego. Weiner, B. 1986. An attributional model of motivation and
emotion. New York: Springer-Verlag.
Schoorman, F. D., Mayer, R. C., & Davis, J. H. 1996b. Organi-
zational trust: Philosophical perspectives and concep- Williams, M. 2001. In whom we trust: Group membership as
tual definitions. Academy of Management Review, 21: an affective context for trust development. Academy of
337340. Management Review, 26: 377396.
Scott, C. L., III. 1980. Interpersonal trust: A comparison of Zaheer, A., McEvily, B., & Perrone, V. 1998. Does trust matter?
attitudinal and situational factors. Human Relations, 33: Exploring the effects of interorganizational and inter-
805 812. personal trust on performance. Organization Science, 9:
141159.
Scott, D. 1980. The causal relationship between trust and the
assessed value of management by objectives. Journal of Zand, D. E. 1972. Trust and managerial problem solving.
Management, 6: 157175. Administrative Science Quarterly, 17: 229 239.

F. David Schoorman ([email protected]) is a professor of management in the Kran-

nert Graduate School of Management at Purdue University. He earned his Ph.D. from
Carnegie Mellon University. He has published research in the areas of trust, leader-
ship, decision making, and organizational effectiveness.

Roger C. Mayer ([email protected]) is professor of management at The University

of Akron. He earned his Ph.D. from Purdue University. His research includes employee
motivation, decision making, organizational effectiveness, and social capital.

James H. Davis ([email protected]) is an associate professor, the Siegfried Director of

Entrepreneurship, and the John F. OShaughnessy Professor of Family Enterprise at
the University of Notre Dame. He earned his Ph.D. from the University of Iowa. His
research interests include trust, corporate governance, family business, stewardship
theory, and social capital.