Straining for shared meaning in organization science: Problems of trust and distrust

We present a problem-centered organizing framework of trust, ln which prominent conceptualizations of trust and distrust from the organizational and allied social sciences are categorized based on the questions they attempt to answer. The frame work we outline here is intended to complement earlier typologies by suggesting alternative strategies for employing the diverse trust literature, identifying questions that could be profitably addressed through interdisciplinary research efforts, and distinguishing disagreements where debate and research would seem to be espe cially worthwhile from those where such activities would appear to be much less useful. inten

A scientific concept has meaning only because scientists mean something by it. The meaning is scientifically valid only if what they intend by it becomes actual: problems are solved and intentions are fulfilled as inquiry continues (Kaplan, 1964: 46).
What is trust? ,This question has been receiving increasing attention by scholars in organizational science and related fields. However, as the volume of research emphasizing constructs labeled "trust" and "distrust" has expanded over the last several decades, the number of uses and meanings ascribed to these terms has grown also. The present conceptual diversity in the literature on trust is reflected in the works of several scholars, who have suggested typological systems intended to organize the vast interdisciplinary research on the subject {e.g., Bromiley & Cummings, 1995;Hosmer, 1995;Lewicki & Bunker, 199Sa,b;Mishra, 1996;Sitkin & Roth, 1993).
For example, Sitkin and Roth {1993) suggest that the work on trust could be collected into four basic categories: (1) trust as an individual attribute, (2) trust as a behavior, (3) trust as a We presented an earlier version of this article at the annual Academy of Management meeting in Boston, 1997. We thank Marta Elvira, Paul Olk, Marianne Lewis, Daniel McAllister, Alladi Venkatesh, Special Issue Editor Sim Sitkin, and the anonymous reviewers, who provided helpful comments on earlier versions of this article. 405 situational feature, and (4) trust as an institutional arrangement. Hosmer {1995) posits that individual expectations, interpersonal relations, economic exchanges, social structures, and ethical principles represent the major approaches to, or contexts of. trust. Following closely on the work of Worchel {1979), Lewicki and Bunker {1995a,b) contend tha~ research on trust could be grouped into three primary categories, each associated with a particular disciplinary perspective: (1) personality theorists' view of trust as an individual difference, (2) sociologists' and economists' notion of trust as an institutional phenomenon, and (3) social psychologists' conceptualization of trust as an expectation of another party in a transaction.
The remarkable diversity in conceptualizations of trust seems to be disconcerting for many scholars {e.g., Barber, 1983;Butler, 1991;Hosmer, 1995, Lewicki & Bunker, 1995aShapiro, 1987;Zucker, 1986). For instance, Shapiro laments that the attention scholars have given has resulted in "a confusing potpourri of definitions applied to a host of units and levels of analysis" {1987: 624). Hosmer agrees, stating that "there appears to be widespread agreement on the importance of trust in human conduct, but unfortunately there also appears to be equally widespread lack of agreement on a suitable definition of the construct" {1995: 380). Lewicki and Bunker add that, despite the level of interest and range of viewpoints in the area of trust, "there has been remarkably little effort to integrate these different perspectives or articulate the key role that trust plays. in critical social processes " (1995b: 135).
The uneasiness associated with the lack of consensus on the meaning of trust (and distrust) is understandable, because disagreement . of this sort has been regarded as a serious impediment to collective efforts at scientific advancement (Pfeffer, 1993). Here, though, we challenge the notion that the use of multiple conceptualizations of trust in organizational science has been a major obstacle to our understanding of organizational phenomena. In fact, we assert that considerable knowledge has accumulated on issues of trust. However, it has accrued to specific problems and related theoretical frameworks, rather than to the topic area of trust as a whole (cf., Lewicki & Bunker, 1995b). Conse-q1,1ently, we contend that a more reasoned discourse on trust-related issues in organizational science would likely consist of research programs acknowledging or attempting to take advantage of the extant conceptual variety, rather than trying to eliminate it altogether.
Accordingly, we propose a problem-centered approach designed to assist organizational scientists to manage more effectively the diversity on the trust topic. The particular framework we present groups prominent conceptualizations of trust and distrust from the organizational and allied social sciences into categories based on the general types of questions they attempt to answer. This method contrasts with earlier classification schemes, in that it is not based primarily on the theoretical differences or disciplinary roots of the works examined. Although the previous systems have been quite useful in helping scholars to organize the oceanic volume of literature focusing on constructs of trust, they have also accentuated the topic area's conceptual diversity without providing an effective means of dealing with it.
The framework we outline here is intended to complement earlier typologies by suggesting to researchers alternative ways of employing the diverse and voluminous trust literature. We have grounded the framework on the idea that differences among theories or definitions of trust require attention only to the extent that the differing conceptualizations themselves attempt to address the same organizational problems.
The utility of a problem-centered focus derives from several sources. First, it provides a way for researchers to coherently delimit the work on trust and distrust to that which is pertinent to their own specific research issues. Second, it affords scholars a means of distinguishing disagreements where debate and research would seem to be especially worthwhile from those where such activities would appear to be less 'useful. Third, it suggests how two substantively different-and seemingly incommensurateconceptualizations of trust may be employed together, without necessarily compromising the theoretical integrity of either. In general, the framework we introduce in this paper is part of a shift in the question, "What is trust?" to what may be a better question: "Which trust and when?" · We begin the article with a discussion in which we attempt to suggest the magnitude of the incongruities that exist among trust constructs. We contend that while the extant diversity probably cannot be accommodated under a single conceptualization of trust, there does appear to be some coherence within the topic area, which we discuss in terms of a trust theme. Next, we introduce the problem-centered categorization system, considering several illustrative approaches to trust within each problem class, and we specify the advantages of employing this type of approach. We conclude the article with a discussion of how an appreciation for the diversity in the organizational problems related to trust perspectives can benefit future research.

VULNERABILITY: A COMMON THEME OF TRUST RESEARCH
Organizational scientists employing the terms "trust" and "distrust" have drawn from a broad range of social science disciplines in order to gain perspectives, develop models. and identify methods for explaining and predicting an array of organizational phenomena at varying levels and units of analysis. In fact. the extant variety in approaches to trust is largely a function of the diverse theoretical perspectives and research interests of those scholars who have used the term (cf., Lewicki & Bunker, 1995b). A comparison of several prominent models is illustrative of the extent of conceptual diversity on this topic.
Rotter, for instance, defines trust as "a generalized expectancy held by an individual that _ the word, promise, oral or written statement of another individual or group can be relied upon" (1980: 1). He views it as a disposition that would be most predictive in situations where individuals are relatively unfamiliar with one another. In contrast. Gambetta (1988) conceives of trust as a calculated decision to cooperate with specific others, based on information about others' personal qualities and social constraints. Dunn (1988) explicitly advances two distinct trust concepts: on_e based on calculated decisions and another grounded in human emotions. Alternatively, Zucker proposes that trust is "a set of expectations shared by all those involved in an exchange" (1986: 54). and she maintains that these expectations are preconscious in nature in that they are taken for granted as part of the world known in common, until they are violated. Finally, Shapiro (1987) defines trust as an agentprincipal relationship.
In addition to the diversity associated with authors' basic definitions of trust, the various subcategories many scholars have posited have introduced further complexity. For example, McAllister argues for two kinds or bases of trust (1995: 25): one grounded in cognitive judgments of another's competence or reliability (referred to as "cognition-based trust"), and another founded in affective bonds among individuals (referred to as "affect-based trust"). Lewicki and Bunker (1995a,b) distinguish three types of trust: calculus-, knowledge-, and identification-based trust. Sitkin (1995) also identifies three aspects of trust; his are competency-, benevolence-, and value-based trust.
Furthermore, although constructs of trust have been the most common foci for organizational scholars within the topic area, they have increasingly emphasized distrust (e.g., Bies & Tripp, 1996;Fein & Hilton, 1994;Kramer, 1994;Sitkin & Roth, 1993;Sitkin & Stickel. 1996). Moreover, in some recent works centering on distrust. researchers treat it as a distinct concept from that of trust. For example, Sitkin and Roth posit that trust is a "belief in a person's competence to perform a specific task under specific circumstances," whereas distrust is a "belief that a person's values or motives will lead them to a · pproach all situations in an unacceptable manner" (1993: 373). Defining trust and distrust as independent constructs is at odds with the perspectives of many other scholars, who view trust and distrust as polar opposites (e.g., Rotter, 1967).
These examples help to demonstrate that meaningful and substantive differences exist among various conceptualizations of trust and distrust in organizational science and the allied social sciences. Furthermore, they suggest that the threads connecting many of the different trust-type constructs may be very thin, or even nonexistent. Yet. an extensive consideration of the trust literature does seem to indicate a coherent theme. When the terms "trust" and "distrust" have been evoked in the social sciences, they almost always have been associated with the idea of actor vulnerability.
For instance, Mishra (1996) draws on Moor-. man, Zaltman, and Deshpande (1992) to make the point that, in the absence of vulnerability, the ~oncept of trust is not necessary, since outcomes are not of consequence to trustors. Kee and Knox (1970) suggest that the study of trust involves situations where at least one party has something meaningful at stake and is cognizant of the potential for betrayal and harm from the other. Likewise, Gambetta claims that '.'for trust to be relevant. there must be the possibility of exit, betrayal, defection" (1988: 217). Similarly, Coleman proposes that trust situations are those "in which the risk one takes depends on the performance of another actor" (1990: 91). and Granovetter (1985) insists that the very nature of trust provides the ' opportunity for trustee malfeasance. Moreover, many organizational scientists who have attempted to understand how economic transactions come to be organized (e.g . . Bradach & Eccles, 1989;Bromiley & Cummings, 1995;Chiles & McMacklin, 1996;Cummings & Bromiley, 1996;Granovetter, 1985;Nooteboom, 1996;Nooteboom, Berger, & Noorderhaven, 1997;Ring, 1996;Ring & Van de Ven, 1992;Zaheer & Venkatraman, 1995) view trust as a mechanism that mitigates against the risk of opportunistic behavior among those engaged in various types of economic transactions.
In some instances scholars have embedded the notion of vulnerability directly within their trust definitions (e.g ., Mayer, Davis, & Schoorman, 1995;Mishra, 1996;Sabel, 1993;Zand, 1972). For example, Mayer et al. propose that trust is "the willingness of a party to be vulnerable to the actions of another party based on the expectation that the other will perform a particular action important to the truster, irrespective of the ability to monitor or control that other party" (1995: 712). Sabel employs the idea of vulnerability somewhat differently: "Trust is the mutual confidence that no party to an exchange will exploit the other's vulnerability" (1993: 1133).
The above examples are far from exhaustive. In fact. considered as a whole, the research on the topic of trust appears to be premised on the general idea that actors (e.g., individuals, groups, organizations) become, in some ways, vulnerable to one another as they interact in social situations, relationships, or systems. Trust has been viewed as a suitable label for various constructs employed to understand different phenomena connected· with the issue of actor vulnerability.
Despite this common ground, the likelihood of devising a universal conceptualization of trust and distrust that is useful for organizational science researchers appears lo:w, for at least two reasons. First, the idea of vulnerability itself has been incorporated into projects in various ways. Some scholars view trust as a willingness to be vulnerable (Mayer et al., 1995;Mishra, 1996). Shapiro (1987), however, views trust as actual vulnerability when she equates it with the agency relationship. Other researchers conceive of trust as a rational decision regarding the extent to which others are likely to cause harm (e.g., Coleman, 1990;Gambetta, 1988). From this perspective, willingness to be vulnerable or actual exposure to potential harm (e.g., when one cooperates with another) may subsequently follow from the trust decision. In contrast, Rotter (1980) attempts to explain how the dispositions of different individuals lead them to react differently in a particular type of vulnerable circumstance (e.g., interactions with unfamiliar others}, and Zucker's (1986) notion of trust as a preconscious expectation suggests that vulnerability is only salient to trustors after a trustee has caused them harm.
Second, tenets of theory building indicate that a universal conceptualization of trust and distrust may have difficulty attaining a sufficient level of theoretical and empirical viability for research purposes. Dubin (1978) emphasizes that theory building must acknowledge the constraints of human cognitive ability. On the one hand, given the extreme conceptual diversity discussed above, a construct that incorporates the relevant complexities presented by the var-ious extant perspectives is likely to be cognitively overwhelming. On the other hand, a conceptualization that omits these complexities in favor of a more abstract definition is likely to be too obscure to suggest a specific array of concrete data to be collected in empirical investigations (cf., Osigweh, 1989). In other words, attempts to "stretch" trust to cover the extant range of usages seem to be at an extremely high risk of producing constructions that are either too elaborate for theoretical purposes or relatively meaningless in the realm of empirical observation.
Still, although the theme of vulnerability may not necessarily result in a universally accepted or particularly useful conceptualization of trust, one can use it to help establish the current boundaries of the topic area, and it also suggests an explanation for the increasing use of the term in organizational science research. As organizational arrangements have become more complex, actors' vulnerabilities to one another have become broader and deeper, and problems revolving around how actors cope with these new conditions seem to have emerged as central concerns.

PROBLEMS OF TRUST AND DISTRUST
We present our basic organizing framework in Table 1. in which we classify works on trust according to their general problem focus: interactions among unfamiliar actors, interactions among familiar actors within ongoing relationships, and organization of economic transactions. These categories emerged from our analysis of the trust literature and are based on logical distinctions among the fundamental problems addressed by trust researchers. The first classification in the table contains work that centers on social interactions among actors who have not gathered information about-or established affective bonds with-one another. The next consists of research focused on interactions among those who have meaningful information about-or established affective bonds with-one another. The final grouping concerns work addressing the general question of how economic transactions come to be governed structurally.
One question we confronted as we developed this framework was how finely tuned a catego-

TABLE 1 Problem Foci for Trust and Distrust Research
Our bias· was to create as few categories as possible, both for the sake of clarity and because it allowed us to juxtapose more works of authors from different disciplinary backgrounds who. have addressed similar issues. Thus. this framework seemed to reflect the best level of abstraction of researchers' problem foci. Employing these groupings, we discovered insights that would not have been apparent using a system with narrower problem classes. Even so, this classification system is intended to clarify and spur problem-focused research-not to confine it.
Although the system we propose is meant to elucidate the general kinds of organizational problems to which various conceptualizations of trust and distrust thus far have been applied and proven useful, we acknowledge that it is not the only problem-focused framework possible. Moreover, we recognize that other problem formulations may suggest different insights and alternative deployments of the trust literature. This is not inconsistent with our main objective, which is to argue that knowledge of organizational phenomena where trust-type constructs are concerned is gained more effectively by attending to researchers' common problems, rather than by attempting to force those working with widely disparate research perspectives and objectives to adopt a common definition or theory of trust.
The notion that scientific constructs, and the theories they constitute, have utility to the extent that they contribute to .our understanding of the social phenomena they may be applied to has been a central tenet of social science research (e.g., Dubin, 1978;Hempel 1965;Kaplan, 1964;Kerlinger, 1986). The value of the problemcentered scholarship that we espouse derives from a re-emphasis of this fundamental idea in conjunction with the acknowledgment that trust researchers have developed different trust constructs to address disparate sets of questions. Its practical utility results from the strategies it presents to scholars for effectively and coherently employing the enormous volume of trust literature in their efforts to address the specific issues in which they are interested. We will highlight some of these strategies throughout the remainder of the article.
Before elaborating on the trust problems contained in the table, we need to note that several prominent works familiar to readers do not appear there. We have omitted some of these because they address a problem not usually considered within the purview of organizational science: tlie problem of social order-broadly considered (e.g., Barber, 1983;Durkheim, 1933;Garfinkel, 1963;Lewis & Weigert, 1985;Parsons, 1967;Simmel, 1964). We excluded others because they center solely on conceptualizations, measures, definitions, or critiques, without linking their discussions to specific organizational problems (e.g., Cook & Wall, 1980;Cummings, Harnett, & Stevens, 1971;Hosmer, 1995;Johnson-George & Swap, 1982;Williamson, 1993). Our intention was to limit the works in the table to those that address relatively specific problems and that appear to have direct implications for the organizational researcher.
In addition, while we have made every attempt to be comprehensive in citing organizationally focused work, the lengthy list must necessarily remain incomplete in allied areas, such as person perception, negotiations, law, and so forth. However, we hope we have provided at least enough of a sample of these works to assist interested readers in further exploration of these fields' approaches. The table, then, represents our attempt to provide a problem-focused context for research and debate on the topic of trust in organizational science.

Interactions Among Unfamiliar Actors
One major problem addressed by trust researchers pertains to interaction among unfamiliar actors-that is, actors whp have little information about, or have not established affective bonds with, one another. Several different disciplinary perspectives are represented here. We highlight three of the more prominent and distinct perspectives pertinent to understanding interactions among relative strangers in this section, in order to demonstrate the utility of the framework. Previous typologies typically would have considered them separately.
The first perspective, involving dispositional theories of trust, assumes that factors exist within individuals that predispose them to trust or distrust others, whom they do not know. The central issues for these types of models revolve around the questions of how individuals develop their propensities to trust and how these predilections affect their thoughts and actions (e.g., Hardin, 1993;Rotter, 1967Rotter, , 1971Rotter, , 1980. Rotter's (1967Rotter's ( , 1971Rotter's ( , 1980 research on trust is the most representative of this category and has been among the most widely recognized and acknowledged work in the organizational studies literature on the topic. He posits that trust is a fairly stable belief based on individuals' extrapolations from their early-life experiences. Further, he suggests that the strength of trust's impact on behavior is a function of the situational novelty with which people are confronted. According to Rotter, as situations become in-creasingly unfamiliar to individuals, the influence that their trusting dispositions have over their behavior grows. Put another way, as people become more acquainted with specific others, their personal knowledge of those others becomes the primary driver of their thoughts and actions.
Another example is provided by Hardin (1993), who recently argued that those who develop a distrusting predisposition tend to avoid cooperative activities (because they expect to be exploited in such ventures}, so are apt to. have fewer positive interactional experiences that can function to adjust somewhat initial distrust levels. By their own actions, these people contribute to perpetuating their distrusting predilections. In organfaations those who distrust may be expected to seek roles that have limited dependencies on others or to resist job changes that cause them to be more reliant on coworkers. A second major approach to understanding trust among unfamiliar actors can be called "bee havioral decision theory." Like dispositional models, many behavioral decision theories address questions associated with the interaction of unfamiliar others (although many also have implications for ongoing relationships and, thus, we include them in the following problem category as well). In contrast to the dispositional approaches, however, these frameworks tend to focus on immediate situational factors in the context of game settings and posit that "trusting" is a function of relatively rational decisionmaking processes (e.g., Axelrod, 1984;Deutsch, 1958Deutsch, , 1960Deutsch & Krauss, 1962;Loomis, 1959;Matthews, Kordoriski, & Shimoff, 1983), rather than personality characteristics. Typically, they define or operationalize trust and distrust in terms of cooperative and competitive behavior, respectively, and they usually attempt to ascertain how changes in the game affect these behaviors. Although behavioral decision theories have been used to investigate relationships of varying lengths, much of the research pertains to situations where partners are strangers. According to Good (1988), some of the situations that have been linked to trusting behaviors include those where the long-term interests of the participants were stressed initially (e.g., Pruitt & Kimmel. 1976), where only small initial rewards were at stake (Deutsch & Krauss, 1962), where there was no potential for threat (Deutsch & Krauss, 1962). and where there was great paten-tial for successful communication (e.g .. Wichman, 1970).
Institutional frameworks represent a third prominent approach. Like behavioral decision theories, these models emphasize the causal role of situational factors in fostering trust among strangers. However, they are typically concerned with the effects of organizational and institutional structures and processes. Shapiro (1987), for instance, concentrates on the problem of how certain conditions necessary for economic exchange are maintained. She defines trust as "a social relationship in which principals-for whatever reason or state of mindinvest resources, authority, or responsibility in another on their behalf for some uncertain future return" (1987: 626). These principal-agent relationships allow individuals, groups, or organizations to bridge the extreme social and physical distances occurring in a complex industrialized society so that they may obtain the benefits of more extensive trade with strangers. Since principals regularly find themselves in situations where they cannot specify, scrutinize, evaluate, or constrain the performance of those on whom they depend, certain social mechanisms (e.g .. procedural norms, selection criteria, risk-spreading devices-all of which Shapiro calls "guardians of trust") function to maintain trust. In the aggregate, "the guardians of trust offer a mix of normative prescriptions, socialization practices, institutional arrangements, structural constraints, and networking strategies designed to maintain the integrity of agency relationships" (Shapiro, 1987: 635). Zucker (1986) also emphasizes economic exchange and attempts to explain the emergence of specific social structures and processes in the United States' economy. She describes how certain institutional arrangements (e.g., rational bureaucratic organizations, the professions, regulations, and laws) have developed for the purpose of producing the trust required to support complex economic systems. For example, rational bureaucratic organizations generate common expectations (i.e., trust) through written rules and formal hierarchy in order to support transactions within and between organizations. Thus, the bureaucratic organizational form is an example of a trust-producing mechanism for situations where the scale and scope of economic activity overwhelm interpersonal trust relations. Pearce, Bigley, and Branyiczki (in press) tested several of these ideas, finding that, indeed, relatively more bureaucratic organizational practices are associated with greater participant trust.
Both Shapiro (1987) and Zucker (1986) focus on understanding which organizational and institutional arrangements produce trust among strangers. Although works such as these typically have originated in sociologists' interest in explaining why certain modern organizational forms arise, such researchers as Xin and Pearce (1996) more recently have extended this work to questions of how individuals working in societies without stable modernist institutions sustain their organizations through aggressive use of personal trust relationships.
The three perspectives on trust (dispositional, behavioral decision, and institutional) we discuss above offer potential insights into understanding organizational situations involving the interaction of unfamiliar actors. By focusing on different levels and units of analysis, they each are suggestive of a different locus of explanation for trust among strangers: trustors' personalities, relatively more proximal situational factors (e.g., "rules of the game"}, and relatively more distal contextual elements (i.e:, organizational or societal structures and processes).
Particular applications-either individually or in combination-of the trust models con-· tained in this class may depend, to a large extent, on the more specific components of the problem under investigation. For example, both institutional and behavioral decision theory approaches tend to emphasize different components of actors' situations in their explanations of behavior, whereas dispositional models focus on stable psychological elements to predict individuals' actions in novel circumstances. Therefore, a determination of the situational strength (cf., Mischel, 1973) facing the actors being studied may help investigators select the framework(s) that would be most useful. If the objective is to explain initial levels of cooperation among strangers trying to form new associations in the absence of influential contextual factors, then dispositional models may be the most useful theories to use when addressing the issue.
In addition to assisting researchers select the work that may have the most potential for answering their specific questions, problemfocused scholarship can help them ascertain where theoretical debate and empirical testing would be beneficial. As one example, Rotter (1967Rotter ( , 1971Rotter ( . 1980 and Hardin (1993) appear to make competing claims regarding the conditions under and the extent to which dispositions to trust are malleable. Rotter's work indicates that the trusting (or distrusting) personality characteristic is quite fixed, once it has been established in one's relatively early life. Hardin, however, argues that the predilection to (dis)trust is, at least partially, a function of individuals' own behavior. In particular, since distrustors tend to avoid cooperative situations, they do not provide themselves with enough opportunities to modify their basic inclinations to view others generally as untrustworthy. So, for instance, if the specific research issue involves whether distrusting dispositions can be modified through organizational interventions (e.g., compelling individuals to work with unfamiliar · others on a series of project teams}, a test of these two theories seems warranted.
Grouping approaches to trust according to their problem focus can help distinguish useful frameworks and identify areas where debate and testing are worthwhile. The possibilities that this approach suggests are not readily apparent through the use of disciplinary-based topological systems alone.

Interactions Among Familiar Actors
As we stated previously, this problem category consists of research focused on interactions among actors who have accumulated meaningful knowledge about. or established affective bonds with, one another. This classification comprises the largest grouping of organization and organization-related trust arid distrust work, and it includes studies of relationships at various levels of analysis. Some of these inves-. tigations focus on interpersonal relationships (e.g., Mayer et al., 1995;McAllister, 1995); others highlight relations among different groups, organizations, or classes (e.g., Dodgson, 1993;Fox, 1974;Lorenz, 1992Lorenz, , 1993Sabel, 1993). In addition, although most of this research has emphasized trust constructs, recent work increasingly has been accentuating distrust conceptualizations (e.g., Bies & Tripp, 1996;Fein & Hilton, 1994;Kramer, 1994;Sitkin & Roth, 1993;Sitkin & Stickel, 1996). A consideration of all of the various relationship levels would be beyond the scope of this article. Instead, we focus primarily on interpersonal interactions in this section as a way of showing the usefulness of the approach. Researchers whose works have been grouped here typically have conceptualized trust as a state of mind. However, although there appears to be substantial consensus on this point, there is much less agreement when it comes to the · more precise formulations of trust that have emerged over the last several years. One. of the most contested issues along these lines relates to whether trust is exclusively t. he product of individuals' calculative decision making processes or is emotion based.
The conceptualizations set forth by Gambetta (1988) and Coleman (1990) are fairly representative of the rational choice viewpoint. Gambetta contends, for instance, that some agent A trusts some other agent B, when A calculates that the probability of B performing an action that is beneficial, or at least not detrimental, to A is high enough for A to consider engaging in some form of cooperation with B. Similarly, Coleman argues that situations of trust can be viewed as a subset of those involving risk: "The elements confronting the potential truster are nothing more or less than the considerations a rational actor applies in deciding whether to place a bet" (Coleman, 1990: 99).
Other researchers, however, have argued that trust is a product of peoples' emotions, at least to some extent (e.g., Holmes, 1991;Johnson-George & Swap, 1982;Lewicki & Bunker, 1995a,b;Mayer et al., 1995;McAllister, 1995;Rempel, Holmes, & Zanna, 1985;Ring, 1996). Furthermore, empirical work seem!;, to support the distinction between · these rational and emotional bases of trust. For example (and as we mentioned above), research conducted by McAllister· (1995}, with a sample of managers and professionals, distinguishes between cognition-and affect-based trust. He also suggests that cognition-based trust is an antecedent of affect-based trust. Holmes (1991) presents a somewhat different kind of relationship between calculative and emotional processes in his trust theory. In his developmental framework of trust in close relationships, he posits that existing states of trust affect people's perceptions and evaluations of their partners and function to inhibit the calculative aspects of trust decisions. Trust forestalls the monitoring and evaluating of a partner's behavior. In fact, Holmes suggests, "A tru,sting relationship can probably be recognized by the absence of an active appraisal process in the normal course of events" (1991: 82). Therefore, although there is spme agreement that the psychological nature of trust in ongoing relationships involves a decision based on information about the other and the situation, there is considerable debate over whether-and in what form-emotional processes are involved.
We contend that the best approach to these types of disputes is, again, to attend to researchers' specific organizational problems. For example, decision-based conceptualizations of trust may be quite useful for problems involving interactions taking place within contexts that present actors with salient and harsh penalties for untrustworthy behavior. One illustration is Coleman's (1990) investigation of the problem of how close communities allow diamond dealers in London and Ne'?" York to engage in transactions worth considerable sums of money solely on the basis of verbal agreements. He concludes that the communities in which these transactions take place support interactions of this sort because they effectively disseminate information regarding a trustee's reputation to all those on whom he or she must rely for business in the future. This particular problem may call for a purely rational choice perspective on trust. In other circumstances, where personal relationships are expected to be particularly strong and situational factors relatively weak, one of the trust theories that involves emotional elements may be most predictive.
In some instances, however, conceptualizations of trust or distrust may be competing to answer the same research questions. Here, direct empirical testing may be called for to determine which is preferred. Mayer et al. (1995) andMcAllister (1995), for instance, represent a case where a head-to-head test would seem to be appropriate. Both works provide carefully argued, literature-grounded trust models attempting to account for approximately the same types of organizational phenomena. However, they are contradictory on key points. One of the most fundamental discrepancies between the two pertains to the dimensionality of the trust construct. Mayer et al. view trust in unidimensional. terms, conceptualizing it as a willingness to be vulnerable based on reasoned judgment. As mentioned previously, McAllister distinguishes among two kinds of trust (cognition and affect based), which seem to have different antecedents, and they appear to differentially affect certain work-related behaviors.
Given that these two theories seem to have been intended to address the same sorts of problems, empirical attention to theoretical discrepancies appears warranted. Tests pitting these two trust theories against each other to predict important work behaviors, such as employee citizenship or coworker cooperation, would help researchers determine which best advances our understanding of organizational behavior.
Similar logic applies to the trust-distrust issue emerging in organizational science. Although many researchers seem to regard these terms as labels for antithetical or mutually exclusive concepts, some scholars have used the terms to identify more independent constructs (e.g., Mishra, 1996;Sitkin & Roth, 1993;Sitkin & Stickel,. 1996). Here again, the differences between these two types of theories are problematic only to the extent that they attempt to address the same problems. The questions asked by Sitkin and Roth (1993), for instance, center on why organizations' legalistic responses to trust-type problems often have unintended, negative consequences. Debate and empirical testing pitting their perspective against distrust-as-the-opposite-of-trust formulations are required, to the degree that the latter can be effectively applied to Sitkin and Roth's specific concerns.
Finally, grouping organizational trust research according to problems highlights certain areas begging for integrative research. One such example is the lengthy list of antecedents employed by various researchers investigating ongoing relationships with constructs labeled "trust" or "distrust." These antecedents include the other's reputation for trustworthiness (Jarillo, 1988), symbolic exchange (Haas & Deseran, 1981), transformational leadership (Podsakoff, MacKenzie, & Bommer;, openness (Butler & Cantrell, 1984), and organizational development programs (Kegan & Rubenstein, 1973). For some purposes it may be useful to include many, of these in a single project. Must certain combinations of them be present in minimal amounts before an ongoing relationship can be sustained? Are some substitutes for one another? This seems to be a valuable area for future research.

Organization of Economic Transactions
Research in this problem class centers on the general question of how economic transactions come to be governed structurally. Scholars in the transaction cost economics tradition originally addressed this issue (cf., Coase 1937;Commons, 1934;Williamson 1975Williamson , 1985. More recently, scientists from other organizational research perspectives have become increasingly interested in problems of transaction governance (e.g., Bradach & Eccles, 1989;Bromiley & Cummings, 1995;Granovetter, 1985;Ring & Van de Ven, 1992).
From the basic transaction cost economics viewpoint, the transaction-that is, the transference of a good or service across a technologically separable interface (Williamson, 1985: 1)-is regarded as the fundamental unit of analysis. Researchers typically analyze how transactions of different types become aligned with various governance structures (e.g., market, hierarchy, or hybrid) tn ways that supposedly minimize their costs. When researchers consider intra-and interorganizational relationships, they view them within the context of transaction costs and transaction governance.
In early works on this problem (cf., Williamson, 1975), scholars identified trust as a phenom-. enon that could affect certain kinds of governance costs. Recently, a number of scholars have attempted to formulate trust constructs and specify their functions relative to the basic tenets of transaction cost economics (e.g., Bradach & Eccles, 1989;Bromiley & Cummings, 1995;Chiles & McMacklin, 1996;Cummings & Bromiley, 1996;Granovetter, 1985;Nooteboom, 1996;Nooteboom et al.. 1997;Ring, 1996;Ring & Van de Ven, 1992;Zaheer & Verikatraman, 1995). For example, Bromiley and Cummings (199:5;Cum-mings~ Bromiley, 1996) have developed a conceptualization and measure of trust and tried to demonstrate how it may affect transaction governance. Along similar lines, Nooteboom et al. (1997) argue that their trust construct interacts with governance structures to influence the perception of risk for agents of firms in alliances. Ring and Van de Van (1992) focus on how their notion of trust affects the structure of interfirm relationships.
Some theorists employ their trust concepts as a basis for critiquing the assumptions and theories of transaction cost economics itself. Bra-dach and Eccles (1989), for instance, regard trust as an alternative control mechanism, along with price and authority, and they contend that trust, price, and authority often coexist as governance devices within organizations. Another example is Granovetter (1985), who criticizes Williamson's (1975) theory for not considering that economic activity is typically embedded in networks of social relations, which involve trust.
Researchers whose works we group into this category have used several different types of trust constructs to address a variety of specific transaction governance concerns. For example, Bradach and Eccles (1989) conceive of trust as a calculated decision by one party regarding the likelihood that another will cause harm. Ring and Van de Ven, however, in their work (Ring, 1996;Ring & Van de Ven, 1992), conceptualize trust as being grounded in affect-based or relational factors, such as goodwill and benevolence. Their intention is to explain transaction governance between firms engaged in collaborative activities.
Consistent with our central argument, we contend that the discrepancies between these approaches require attention to the extent that they are applicable to the same particular problem sets. Since Bradach and Eccles (1989) specifically indicate that their view of trust is appropriate for understanding both intra-and interfirm transaction governance, we believe theoretical debate and empirical testing are especially warranted on problems emphasizing transactions. between firms. Testing may involve specific questions concerning, for example, which conceptualization best explains the governance structure and stability of transactions between organizations engaged in multinational joint ventures.
Applying trust constructs to the problem of transaction governance is a relatively recent occurrence in the organization studies literature. Although their central concern has been the organization of transactions, many of the researchers using a transaction cost economics perspective have attempted to demonstrate how the organization of transactions within an economic system is dependent, at least partially, on the quality of interpersonal, intergroup, or interorganizational relationships. Consequently, the research in this category is somewhat less selfcontained than that of the two other problem areas. For this reason these theorists seem to be at particular risk of applying conceptualizations of trust originally developed to address problems quite different from their own, without adequate care in specifying the logical connections between the problems.

PROBLEM-CENTERED ORGANIZATION SCIENCE RESEARCH
Social science researchers striving to address divergent sets of problems from different perspectives have defined trust and distrust in a variety of ways and at various levels of abstraction. A consideration of the works of previous scholars, who attempted to organize the trust literature according to theory types (i.e., Bromiley & Cummings, 1995;Hosmer, 1995;Lewicki & Bunker, 1995a,b;Mishra, 1996;Sitkin & Roth, 1993), reveals that the conceptual diversity among trust constructs is considerable. ·We have examined several prominent works to illustrate· the degree of the diversity. Rotter (1980), for instance, views trust as a personality characteristic, whereas Gambetta (1988) suggests it is a rational decision. Alternatively, Zucker (1986) conceives of it as a preconscious expectation. In her work Shapiro (1987) equates the principalagent relation to trust. These are not trivial differences.
Although the theme of actor vulnerability may represent the common ground of trust research, the degree ·of diversity in the literature nevertheless seems to preclude the possibility of a useful universal definition. We have argued that efforts to incorporate existing trust perspectives under one conceptualization are likely to result in concepts that are either unreasonably complex or inordinately abstract for organizational science research purposes. In addition, attempts to force disparate approaches together may result in misapplications of previous approaches.
For instance, Rotter is explicit in his contention that dispositional trust has predictive power only in novel situations-that is, where individuals have little information about the object(s) of their trust. Mayer et al. (1995) cite Rotter's research and incorporate a dispositional moderator in their model of one person's trust in another at work. However, their model neglects the situational ambiguity condition discussed by Rotter. The personality factor included in the Mayer et al. model is expected to function independently of the truster's familiarity with the trustee. Although such a dispositional element ultimately may prove useful, it is not Rotter's dispositional trust.
In view of what may be unresolvable differences among trust constructs, and in line with Kaplan's (1964) quote at the beginning of this article, we have presented an approach intended to assist researchers in managing the conceptual diversity. Our framework has several specific benefits. First, it provides a basis for organizational ·scholars to delimit the enormous volume of trust research to that which is pertinent to their specific research questions. Scholars should not be compelled to discuss definitions and theories focused on problems far removed from their own, simply because those also use the word "trust." The general categories of organizational problems that we propose represent an initial set of decision rules for this task. The more specific questions employed in research may further circumscribe the body of relevant works.
Second, our framework helps to distinguish debates that would appear to be beneficial from those that would seem to be less functional. We have suggested that the models presented by McAllister (1995) and Mayer et al. (1995) appear to address very similar problems. Therefore, theoretical debate and empirical testing designed to establish which is the most predictive framework would seem justified. However, the problems addressed by McAllister (1995) and Zucker (1986), for example, appear to have problem foci that are so dissimilar that efforts to identify the best theory between these two would not be warranted (and, indeed, would be irrelevant to problem-focused organizational research).
In other words, discussions in the area of trust that are not anchored in problems are likely to lead to unproductive debates. For example, Williamson (1993), an economist, attempts to marshal a careful and comprehensive critique of the use of trust in organizational science. He focuses on the works of sociologists, such as Coleman (1990) and Gamb~tta (1988), as those most representative of the literature on the topic. Williamson argues that if trust reduces simply to a decision under risk (as explicitly stated by Coleman, for example}, then trust as a scientific construct should be abandoned in favor of the more suitable conceptual machinery found in transaction cost economics. Williamson writes, "[I]t is redundant at best and can be misleading to use the term 'trust' to describe commercial exchange for which costeffective safeguards have been devised in support of more efficient exchange" (1993: 463). Citing Dunn's (1988) work, Williamson does acknowledge a type of trust that is based on people's emotions (which he terms "personal trust" or "nearly noncalculative trust"), only to dismiss it as irrelevant to commercial exchange relationships. He argues that because personal trust involves switching out of a rational decision-making mode, it "is warranted only for very special personal relations that would be seriously degraded if a calculative orientation were 'permitted.' Commercial relations do not qualify" (Williamson, 1993: 486). Williamson's position-that personal trust does not, or should not, operate in economic realms-sharply contrasts with that of other organizational scholars, who consider the emotion-laden aspects of interpersonal relationships (e.g., Lewicki & Bunker, 1995a,b;Mayer et al.,· 1995;McAllister, 1995) to figure prominently in organizational behavior.
Thus, Williamson has staked out a provocative position-the conventional economist's critique of sociology as a discipline, as well as a dismissal of the role of emotions in organizational relationships-one that flies in the face of the work of many organizational behavior scholars. Such strong statements may serve as an invitation to counterargument by at least some of the many whose understanding of trust Williamson (1993) has dismissed. However, the lack of a specific problem focus in Williamson's critique undermines much of the potential for productive debate. In the absence of a concrete problem, disputants have no effective means for evaluating the usefulness of various constructs against one another. In other words, since definitions or theories of trust have developed to address different kinds of organizational problems, general criticisms of this sort typically are not benefi~ cial.
A third benefit of problem-centered scholarship on the topic of trust is that it invites consideration of alternative perspectives rooted in different disciplinary traditions. This is evident when employing our framework in conjunction with any of the previous approaches mentioned earlier. For example, particular works grouped into three of the four categories proposed by Sitkin and Roth (1993)-that is, trust as an indi-vidual attribute, trust as a behavior, and trust as an institutional arrangement-can be brought to bear on our problems of interactions among unfamiliar actors. The same point can be made when using our approach together with the works of other scholars, such as Bromiley and Cummings (1995), Hosmer (1995), Bunker (1995a,b), or Mishra (1996). In this regard, our approach encourages a sort of comprehensiveness in the use of the literature, but one that centers on specific issues or questions.
Finally, the problem-centered framework we propose calls attention to the distinctiveness of trust constructs and suggests that, since various conceptualizations of trust may answer different sorts of questions, it may be more profitable to use them in conjunction with one another, rather than to invariably and forcibly integrate them. For example, one research problem of interest may involve understanding the dynamic development of cooperative behaviors of organizational members during their first year of employment. For this problem it may be useful to consider using both Rotter's (1980) and Gambetta's (1988) theory. Since Rotter designed his model to explain behavior in situations where actors interact with unfamiliar others, it may be most predictive during the initial weeks of employment. Gambetta's model could be the basis for hypotheses about cooperation in later months, since he intended it to explain actors' behaviors in situations where they have some familiarity with one another and with situational constraints. Certainly, when joining different theories of trust, one must identify their boundary conditions. This may be accomplished through logical bridging statements, which are propositions that develop theoretical relationships between existing perspectives (Cappelli & Shere, 1991;House, Rousseau, & Thomas-Hunt, 1995).
Again, we are not arguing that our framework is the only problem-centered one possible. Certainly, other such typologies may provide additional insights. Still, the categories that we. identify reflect the general kinds of problems that have been addressed by those employing trust constructs, and they should provide a reasonable starting point for most investigations. Furthermore, we are not contending that scholars can never benefit from considering work outside their specific problem focus. However, we advocate that very special attention should be given to the rationale justifying the use of such material.
In conclusion, whether because of its nonscientific origins or its wide-ranging appeal as a social science construct label, the term "trust" has taken on an array of diverse meanings in organizational science. Since the area seems to be fundamentally fragmented in its problems and approaches, straining for a shared meaning of trust in the organizational sciences is likely to be a relatively unproductive endeavor. We have attempted to demonstrate that the most relevant criterion for assessing the viability of a conceptualization is the extent to which it is useful in addressing particular organizational problems or issues. We hope that this framework has helped to alert readers to the substantial intellectual differences in conceptual uses of trust in organizational science and that it will encourage them to worry less about attacks on their definitions or the lengths of their reference lists and more about meaningful organizational problems.