Keywords

Introduction

The theoretical foundations of this chapter are based on the work of Mayer, Davis, and Schoorman (1995) and Schoorman, Mayer, and Davis (2007) (see Fig. 2.1). First, we briefly review the competing definitions of trust that have been widely adopted in the past 20 years. Next, we examine the implications of the choice of definition, and finally we will review some of the constructs used to represent trust across a number of disciplines. In doing so we view these constructs from the lens of the Mayer et al. (1995) definition.

Fig. 2.1
figure 1

Model of trust from Mayer, Davis, and Schoorman, (1995)

Full size image

Definitions of Trust

In Mayer et al. (1995) we define trust as the “willingness to be vulnerable” in a relationship. Trust, therefore means the intention to take risks in the relationship. Several elements of this definition are very important to our present analysis. First, trust is clearly defined as a relational construct. This relationship could be with another person, a group, or an organization. Thus, trust could be described with respect to each relationship independently. Second, we clearly articulated the need to differentiate between trust, the antecedents of trust, and the consequences of trust. One antecedent is the trustworthiness of the other party, which is described as one’s judgments of their ability, benevolence, and integrity. The other antecedent in the model is the propensity to trust which is a stable predisposition of the individual (trustor) to trust another party. Thus propensity is a generalized trust for others, with no specific referent person or entity. In this model, trust is conceived as a behavioral intention (willingness to be vulnerable) but no action (behavior) has occurred. This behavioral intention is a deliberate choice. We proposed that trust would lead to risk taking in the context of the relationship. This risk-taking would be the actual behavioral consequence of trusting.

In an article published in the same year, McAllister (1995) also developed a model of trust with a slightly different definition. McAllister (1995) defines trust as “…the extent to which a person is confident in, and willing to act on the basis of, the works, actions and decisions of another” (p. 25). This definition of trust focuses on confident expectations of the behavior of others, but also that the behaviors will be positive towards the trustor, later referred to as “confident positive expectations” (Lewicki, McAllister, & Bies, 1998). Like the Mayer et al. definition, McAllister (1995) defines trust as a willingness to act rather than the action itself. McAllister (1995) also views trust as a relational construct.

Prior to the to the publication of these two papers on trust the dominant definition of trust was that of Rotter (1967) whose definition of trust is that of a generalized willingness to trust people, institutions, or entities. In this generalized definition of trust there is no specific referent, and trust is viewed as a stable quality of the person rather than as a relational construct. In the Mayer et al. (1995) model the Rotter conceptualization of trust is included as an antecedent of trust in the variable called propensity to trust.

Rousseau, Sitkin, Burt, and Camerer (1998) summarized the literature on trust in a special issue on the topic in the Academy of Management Review and noted that the most frequently cited definition was “willingness to be vulnerable.” They also reported that other authors “say the same thing in different words” and cited as an example “confident positive expectations” from the work of Lewicki et al. (1998). They combined the two definitions into a single one: “Trust is a psychological state comprising the intention to accept vulnerability based on the positive expectations of the intentions or behaviors of another” (p. 395). It is our view that this combining of definitions has contributed to the lack of clarity and precision in the choice between these two definitions in the literature. We will return to this point in our discussion of distrust later.

In a more recent review of the trust literature Fulmer and Gelfand (2012) noted that the vast majority of trust research focuses on these two dimensions: “positive expectations of trustworthiness” and “willingness to accept vulnerability.” In summarizing their findings regarding definitions they stated a very clear recommendation: “To provide clarity, we recommend that researchers be very clear about the emphasis their definition places on positive expectations, vulnerability, or both, or neither and the rationale for this emphasis” (p. 1171). We could not agree more with this recommendation.

The relational definitions of trust necessarily require a referent or trustee. Schoorman et al. (2007) noted it is important that trust researchers specify the referent of trust in their models and in the empirical assessment of trust. On the other hand, the definition of trust as a dispositional variable or as generalized trust does not require a specific referent. In the case of dispositional trust, the measure is akin to the concept of trustworthiness and resides in the individual.

In our reading of the trust literature it appeared that almost all trust research draws from one of these three definitions. Both the Mayer et al. (1995) paper and the McAllister (1995) paper were published in the Academy of Management Review as was the special issue paper by Rousseau et al. (1998). We found that in the literatures in Management and specifically in Organizational Behavior and in Psychology, the research on trust generally adopted the view that trust is relational, consistent with these articles. In other fields such as political science, public policy, and law there are some researchers who continue to draw on the definition of trust as a generalized disposition that resides in the individual rather than in the specific relationship, thus relying more on the definition that originated in the work of Rotter (1967).

Differentiating Antecedents and Trust

A notable difference in McAllister’s (1995) conceptualization is that he does not differentiate between trustworthiness (our antecedent) and trust itself. For example, in his paper he states that “…the beliefs of managers about the trustworthiness of peers can be measured along two dimensions, the extent of affect-based trust and the extent of cognitive-based trust” (p. 51). In our model we would represent this as the extent of trustworthiness is measured by perceived benevolence (affect-based) and ability (cognition-based). The conceptual differences are subtle but by adding the word trust to the adjectives it has led to a confounding of the antecedents of trust and trust itself.

In 2007, Colquitt, Scott, and Lepine conducted a meta-analysis on trust research and confirmed that ability, benevolence, and integrity did indeed act as antecedents of trust and that each of these variables had independent effects on trust. They also found that trust was a predictor of risk-taking behaviors as indicated in the Mayer et al. model. Perhaps the most important confirmation in their meta-analysis was the value of treating trustworthiness, and the factors that make up trustworthiness as antecedents of trust and distinct from trust itself. Although the “willingness to be vulnerable” definition is now widely accepted in these fields, there are a number of research streams that continue to confound trustworthiness and trust. Following the lead of the McAllister model which referred to affect-based trust and cognition-based trust, there has been proliferation of research that identified a new antecedent (condition or process) and links it to trust and proposes this as a unique form of trust. The most recognized of these are calculative-trust (Williamson, 1993), relational trust (Rousseau, Sitkin, Burt, & Camerer, 1998), identity-based trust (Coleman, 1990), integrity-based trust (Kim, Ferrin, Cooper, & Dirks, 2004), and knowledge-based trust (Lin, 2011). A quick search of the literature revealed over two dozen such adjective-trust combinations that are considered a unique form of trust. This approach also lead to the development of a unique measure of trust for each new antecedent-trust combination and prevented the aggregation of research findings across studies (McEvily & Tortoriello, 2011). We believe the best way to advance the literature on trust is to treat the antecedents of trust as separate variables from trust; this would not preclude those who wish to identify new antecedents of trust to be added to the model.

Distrust

There is a growing body of research that is interested in the concept of distrust (e.g., Lewicki et al., 1998; McKnight & Choudhury, 2006). Schoorman et al. (2007) addressed this issue quite extensively so we will only summarize the arguments here and refer to this work. Most researchers view distrust as the opposite end of the trust scale (McKnight & Chervany, 2001). Our view is that if one adopts the definition of trust as “willingness to be vulnerable,” the lowest level of the variable is a complete unwillingness to be vulnerable. This would be the complete absence of trust or a zero point on the trust scale. We do not need a separate construct for the low end of the trust scale. On the other hand if one adopted the definition of “confident positive expectations” it is possible to argue that one could have negative expectations, and this would be distrust. However, this would still be on the same scale as trust, just the opposite end.

Trust by Other Names

Trust is often used interchangeably with other words that are intended to mean trust or imply that trust is a factor. We will review some of these concepts and briefly discuss how they are not the same thing as trust.

Cooperation

Most people believe there is a close, if not one-to-one relationship between cooperation and trust, and cooperation between individuals or organizations must be an indicator of trust in the relationship (e.g., Ferrin, Bligh, & Kohles, 2007). It is very likely that in a relationship where there is trust there will be a great deal of cooperation. However, because there is cooperation it should not be assumed that there is any risk, or a willingness to be vulnerable. People and organizations cooperate if there is some mutual gain from the cooperation, but this does not require risk or vulnerability. For example, in a business relationship between a supplier and a customer there could be a high level of cooperation even though there is no trust. Wal-Mart as a customer exerts a great deal of pressure on supplies regarding price and delivery often squeezing the suppliers’ margins to where they can barely survive. The suppliers will continue to cooperate with Wal-Mart, but there is considerable evidence that there is no trust (Fishman, 2003).

Predictability

We often hear the word trust used to mean predictability (e.g., McKnight & Chervany, 2002). If an event occurs with a high probability we refer to it in terms of trust. You may say you trust that your train will be on time, or we may report we trust it will get warmer in the spring. These events do not require a relationship, nor do they involve any risk or vulnerability. Perhaps they do represent confident expectations, but it still fails the relationship test.

Confidence

In some disciplines that are more quantitatively oriented, like Operations Management, there is a great interest in reducing the variance in processes and procedures. In statistics we talk about confidence intervals that are a function of the variance in a distribution. Sometimes trust is used as a surrogate for having confidence in a process or a result. However, it is not clear that there is a risk involved in expressions of confidence or a vulnerability by choice. This does not fit with our definition of trust.

Legitimacy

In political science and public policy scholars often discuss the legitimacy of institutions, of governments, and of organizations (e.g., Jackson, Bradford, Stanko, & Hohl, 2013; Thomassen & Schmitt, 1999). In a review of the literature on legitimacy, Tyler (2006) defines it as the “belief that authorities, institutions and social arrangement are appropriate, proper and just” (p. 376). The measurement of legitimacy frequently relied on the concept of confidence in the referent (Gibson, Caldeira, & Spence, 2003). Procedural justice has also been used as an indicator of legitimacy (e.g., Sunshine & Tyler, 2003; Tyler, 1994). If an organization is considered high in legitimacy could we also say that we trust that organization? The key issue here is the notion of whether there is a relationship involved, and whether there is vulnerability by choice. For example, there has been considerable interest in the legitimacy of the Supreme Court and how it has changed through a number of controversial events over the past two decades. While it makes sense to talk about legitimacy, this would not be an issue of trust, because there is, at best, a very superficial relationship, and even if there is a vulnerability there clearly is no choice. Most of us cannot do anything about it. This analysis is different if we are talking about the legitimacy of an organization that is more proximal and there would be an opportunity to do something about it. For example, the legitimacy of the Mayo Clinic may be very salient if you were faced with a decision about cancer surgery. The legitimacy of a particular school may be relevant if you were considering enrolling. In these cases, you do have a choice to make yourself vulnerable and therefore trust the institution. In this context legitimacy can be viewed as an antecedent of initial trust for the organization. Initial trust is not based on any experience with the trust referent but on institutional cues (McKnight, Cummings, & Chervany, 1998). Initial trust can be differentiated from a propensity measure because there is a very specific referent. In our model, legitimacy could be seen as an antecedent to the trustor’s propensity or initial trust (Fig. 2.2).

Fig. 2.2
figure 2

How legitimacy might fit in the trust model

Full size image

Swift Trust

Meyerson, Weick, and Kramer (1996) have described a new construct that they called “swift trust.” They characterize swift trust as that which occurs in a temporary group that is assembled for a specific task in a short term. The presumption is that the group are initially strangers and know that they will be together only for this task. The task is usually difficult and intense, there is a clear goal for the group and a substantial reward associated with accomplishing the goal. The members are assigned to the group and have no choice about joining. Meyerson et al. (1996) describe the group as the “organizational analog of a one-night-stand” (p. 167), suggesting the very short-term duration of the relationship. What should be clear is that these conditions do not fit any of the criteria for this being trust. What we do see in these groups is a very high level of cooperation, which is driven by a very attractive reward system.

Faith

The most common use of the work trust in the United States is the phrase “In God we Trust,” that appears on all U.S. money. It is interesting that most people do not have a good explanation for why it is on the money but they accept the sentiment as a reasonable one. The actual history of the adoption of the phrase as the official motto of the United States and its appearance on paper money in 1956 is quite interesting but probably not relevant to this chapter. If we accept our definition of trust as vulnerability by choice, that trust is a cognitive and considered choice, it becomes curious that one would apply it to a relationship with God. If you believe in God, you do not have a choice about trusting. And if you do not believe in God then trust is not an issue. Clearly, one’s relationship with God is best described as blind trust, and well captured by the concept of Faith.

Other Implications of Our Conceptualization of Trust

In this section we explore some of the broader implications of our definition and conceptualization of trust. In addition to the definitional issue there are several theoretical issues that evolve from our choices of variables and conceptual relationships.

Trust Is Not Reciprocal

In our model trust is a function of the trustor’s judgment of the trustworthiness of the trustee. In the context of any relationship each party makes judgments of the trustworthiness of the other. We believe that these judgments are independent and that it is not the case that trust in one direction is always reciprocated. Trust researchers also referred to this as trust asymmetry. Clearly, in any relationship the opportunity to achieve mutual trust (or trust symmetry) is a valued goal but it is a state that requires effort on both sides and should not be assumed. Too often it is not until trust is broken that it becomes apparent to the parties that trust was not reciprocal. One type of relationship where this is often true in organizations is in the relationship between supervisors and subordinates. It is generally the case that supervisors trust subordinates more than subordinates trust supervisors. This reality comes as a great surprise to supervisors, usually the first time they participate in a 360 review process. The reasons for this discrepancy are also fairly clear. There is an information asymmetry in this relationship because the supervisor has access to most data about subordinates but subordinates usually to not have the same access to information about their supervisor. Absent this information it is more difficult for subordinates to make judgments about trustworthiness. This is an important theoretical issue especially as the leadership literature, particularly the research on Leader-Member Exchange (LMX), conceptualizes the leadership relationship as mutual and reciprocal (Graen & Uhl-Bien, 1995).

Trust Is Domain Specific

In describing the antecedents of trust Mayer et al. (1995) theorize that ability is domain specific. Although benevolence and integrity are constant across domains, ability would be assessed uniquely for each domain of activity. This means that because of variance in ability across domains one would trust an individual in one domain but not in another. This would seem to be reasonable in that you would likely trust your best engineer to manage the plant in our absence but not necessarily trust him or her to run a press conference. Similarly, we may have a colleague who is a great research collaborator who you would not trust to teach a class for you. Although this would seem to be quite rational as one thinks about varying abilities across domains, most people are startled to have to confront the reality that there is no person they truly trust completely.

Different Weighting of Trustworthiness Factors

One of the benefits of a model that differentiates trustworthiness and trust itself is that it allows us to examine how the trustworthiness factors vary in different contexts. While research has consistently shown ability, benevolence, and integrity to be independent predictors of trust (e.g., Colquitt, Scott, & Lepine, 2007), the model allows us to speculate about how the beta weights of these factors in predicting trust would change in systematic ways.

Individual Differences

Our research has shown that there are individual differences in the weighting of these variables across individuals. For some individuals it may be that ability plays a bigger role in deciding to trust someone and for others it may be benevolence. One implication of this is that it is useful for individuals to recognize their own predispositions in how they determine who they will trust.

Cross-Cultural Differences

One of the most interesting aspects of applying our model of trust across cultures is that it is clear that cultures tend to influence the weighting of the factors of trustworthiness. U.S. culture in general tends to place a higher weight on ability, while Chinese culture places a higher weight on benevolence. These patterns have been observed by many researchers doing cross-cultural research, yet we do not have a systematic accounting of how cultures differ in weighting these factors. We are currently working on a 3-year research project to study how the weighting is different in different cultures, and we hope we will soon have more specific evidence on this topic.

Difference in Role and Perspective

We have also observed that there is a different weighting of the factors based on the role that one is in. For example, in a U.S. audience when asked to consider the trustworthiness of one’s subordinates, ability is generally the most important factor. However, when the same individual is asked to consider supervisors, integrity is much more important. We are in the process of examining the weighting for other roles and perspectives.

Understanding Trust Repair

As the trust model indicates, once you take a risk in a relationship you experience the outcomes. These can be positive if the other party delivers on their commitment, or negative if the trust is betrayed. In both cases there is a feedback loop that updates the judgment of ability, benevolence, and integrity. If the result is positively perceived trustworthiness increases and trust increases as a result. However, if there is a betrayal, the factors are adjusted downward very precipitously and trust is broken. There is a growing body of research that examines the process of trust repair (e.g., Gillespie & Dietz, 2009; Tomlinson & Mayer, 2009). There is one aspect of trust repair that is relevant to our discussion of factors of trustworthiness. When a betrayal occurs and the individual reevaluates ability, benevolence, and integrity, they generally make an attribution as to which factor they had misjudged that led to the betrayal. What we have found is that if they decide that they had misjudged ability, recovery of the relationship is likely. For example, if you left a subordinate in charge of the plant and when you returned you found that the work had not been done on schedule which caused serious problems for the company and for you, how would you react? If you decided that you had misjudged the subordinate’s technical skills necessary to complete the work, in other words misjudged their ability, you will likely recover the relationship with this subordinate. If, on the other hand, you decided that they did have all the skill but just did not seem to care enough to make sure they got the work done, in other words you had misjudged their benevolence, the probability of rebuilding the trust is very low. Thus, the model gives us the flexibility to apply it to trust repair and make predictions about the success of the effort.

Institutional Trust

There is some discussion in the literature about extrapolating models of trust to the institutional level. While there is considerable evidence to support the application of our model of trust to the macro level (e.g., Fulmer & Gelfand, 2012), one of the ways to think about the question of what does it mean for an individual to trust an institution is through the weighting of the factors of trustworthiness. In thinking about the ability, benevolence, and integrity of an institution, it is likely that we assess the ability and integrity factors but do not have any data on benevolence. When the referent is an institution it is hard to gather data on the benevolence in the relationship. Thus it may well be that the relative beta weights on the factors of trustworthiness explain the uniqueness of the trusting relationship.

Trust Is Giving Up Control

An important implication of our model is related to the tradeoffs between trust and control as risk management mechanisms. One way to illustrate this trade-off is seen in our “bucket model” of this relationship (Fig. 2.3). In bucket A, perceived risk exceeds trust and therefore the difference will be bridged with control systems which reduce the risk to a level that is acceptable. For example, a senior manager is away from work for a week attending a training seminar. The manager left a subordinate in charge of the workplace, responsible for making sure work proceeds normally. How often does the manager feel the need to call back to work to make sure everything is going as expected? If the trust in the subordinate exceeds the perceived risk, the manager may not call back all week. If there is a small gap, the manager may call back every 2 days. If the gap is large, the manager is often calling back at every opportunity. The act of making the phone call is a monitoring behavior that is a control system. As shown in bucket B, trust does not get built in an organization that relies exclusively on control systems.

Fig. 2.3
figure 3

Trust versus control systems as risk management mechanisms

Full size image

This is best illustrated with what happens with most empowerment programs in companies today. The term empowerment has always seemed to put the burden of change on the workers in the organization who are sent off to “empowerment training.” When they return to the organization, they return to the same environment with the same senior manager who has not learned to let go and give up control. Imagine if we had called it depowerment instead. We would have put the burden of change where it belongs and sent the senior manager to “depowerment training” and maybe we would have had more success.

Trust Across Disciplines

Accounting and Financial Institutions

Invoking the disciplines of Accounting and Finance in a chapter on trust invariably reminds people of the accounting scandals of the 1990s at companies like Enron, Tyco, WorldCom, and others where senior leaders took advantage of the systems that were in place and betrayed the trust that had been placed in them by their customers. The public was outraged by the scandals and demanded action by the U.S. Congress. The financial community expressed great concern and sought ways to rebuild trust. In 2002, the Federal Government passed legislation known as Sarbanes-Oxley Act, which oversees, regulates, and disciplines firms with fines and other criminal penalties. Corporations now have to report more to the government and have stricter regulations with harsher penalties if they are not followed. The Public Company Accounting Oversight Board (PCAOB) was founded by Congress to manage audits within a public company as well as the Financial Accounting Standards Board (FASB), home of the Private Company Council (PCC). Together, their goal is to protect the people or customers as well as the investors and employees of the company.

These actions served as a remedy for situations where trust needed to be rebuilt. The interesting question for researchers in trust is, was it rebuilt? Based on our model of trust we would argue that what they did was build in control systems to reduce the level of risk, thereby reducing the need for trust. Researchers have struggled to develop a clear definition of trust because of the trade-offs between trust building and risk-reduction that comes with the implementation of control systems.

In order to address this, some authors view trust as a confidence in the reliability of both individuals and systems; to have trust in a system, there must be a face of an individual to help alleviate uncertainty in the user (Bachmann, 2001; Busco, Riccaboni, & Scapens, 2006). The personal connection invokes feelings of trust in the user and gives him or her a relational interaction with the organization (Lewis & Weigert, 1985; Seal & Vincent-Jones, 1997). Other authors argue that trust and control are completely separate; you either trust or control, not both (Das & Teng, 1998; Leifer & Mills, 1996; Madhok, 1995; Ring & Van de Ven, 1994). More congruent with our trust model is the belief that trust and control are supplementary with each other in that they both contribute to managing the level of risk (Das & Teng, 1998; Inkpen & Currall, 1997).

Operations Research

The discipline of Operations Research has long been focused on developing efficiencies in process largely by eliminating wasteful steps in the process and standardizing operating procedures. Among the concepts that are common in this discipline are “lean” manufacturing, process mapping, road maps, and production gates. They emphasize the importance of training in project management so as to take advantage of these efficiencies. These practices have also led to the establishment of international standards related to manufacturing that are referred to as ISO standards. Organizations are required to obtain certification in ISO standards in order to do business with or supply to many companies that subscribe to this philosophy. From a risk management perspective it is interesting to note that, like the accounting field, operations research has focused on developing control systems to reduce the need to build trust.

There are researchers in the operations area who have theorized about the role of trust in managing relationships, particularly between suppliers and customers or what the discipline would call the supply chain. Bradach and Eccles (1989) reported that interfirm trust mitigates the risk that a partner firm will act opportunistically in alliances. Gulati (1995) built on this finding in a study of how equity in alliances affects trust between firms. Brinkhoff, Ozer, and Sargut (in press) hypothesized that if there is a pre-existing relationship between project partners that includes trust, the project is more likely to be a success. They cited the definition of trust proposed by Rousseau et al. (1998); however their measurement of trust did not reflect any of the definitions of trust we have reviewed here. One impediment for studying trust in the operations research area (and in many other related fields) is that they relied heavily on archival data for testing their hypotheses. This means that even when they do consider trust as a variable in their models they are usually inferring trust from archival data that is usually not attitudinal data. In a very interesting study of trust in this field that illustrates these concerns, Ozer, Zheng, and Ren (2014) used data from a simulation where they inferred trustworthiness from the behavior of the trustee. Trustworthiness was defined as how much a retailer distorted his or her forecast in a report to the supplier. “A more trustworthy retailer tends to distort her forecast to a lesser extent…” (p. 9). It would be clear to trust researchers who accept our model of trust that the trustworthiness of a trustee is a function of the perception of the trustor rather than the actual behavior of the trustee. This is a case where the common empirical methods of the fields make it difficult to study perceptual variables.

Trust, Risk, and Control Systems

In both the Accounting and the Operations Management disciplines the issue of how to manage trust is addressed in a similar way. Schoorman et al. (2007) discussed how trust and control systems were alternate means of managing risk in organizations. Davis, Schoorman, and Donaldson (1997) examined the widely adopted management philosophy of Agency theory where principals (owners) are concerned about the opportunistic behavior of agents (the managers they hire to run the business). This view is based on the premise that individuals act to maximize their own utility and will readily take advantage of others to do this. The solution to this principal-agent problem in this framework is to establish strict control systems that would include both incentive structures and monitoring and reporting requirements for the agents. Davis et al. (1997) described stewardship theory as an alternate way to manage the risk in the situation. Stewardship theory is based on the alternate assumption about human nature that rejects the notion that individuals are always self-serving in their behavior. This approach suggests that when there is a relationship between the principal and the agent that is based on trust that the agent will act to maximize the outcomes of the collective rather than the individual. As our discussion above illustrates, the literatures in Accounting and Operations Research are based on Agency theory and therefore are focused on control systems to manage risk. A closer examination of the trust model from Mayer et al. (1995) indicates the decision to take a risk and be vulnerable is a function of both the trust and the perceived risk (see Fig. 2.4). What the model clearly argues is that the decision to engage in a risk-taking behavior is not based only on the level of trust but an interaction of trust and the perceived risk in the situation. In an earlier section we describe our “bucket model” of trust and control systems (see Fig. 2.3) that illustrates this trade-off. Consistent with this theoretical argument, in a recent meta-analysis Breuer, Hüffmeier, and Hertel (2014) have shown that trust matters more to team outcomes under conditions of high perceived risk compared to conditions of low perceived risk but that the influence of trust is reduced when control systems such as process documentation are applied.

Fig. 2.4
figure 4

Relationship between trust and perceived risk

Full size image

Online Marketing and Retailing

The Internet has become a central part of private and business lives and has changed the way people communicate, work, and live. One important business field influenced by the Internet is the field of e-commerce and online retailing. The market share of online retailing is growing dramatically: The percentage of e-commerce in total U.S. retail sales increased from 1.6 % in 2002 (44.527 millions of dollars) to 5.4 % in 2012 (226.878 millions of dollars; U.S. Census Bureau, 2014). Trust has been identified as a key issue in the use of online vendors (e.g., Chen & Dibb, 2010; Gefen, 2002). For example, trust is helpful for dealing with the risk of sharing personal information with the web-based vendors (McKnight, Choudhury, & Kacmar, 2002), or for doing financial transactions online (e.g., Pavlou & Gefen, 2004). Moreover, practitioners and researchers discussed the critical role of trust for attaining and retaining customers in a market which is characterized by a large number of competing online retailers for a similar product (Reichheld & Schefter, 2000).

Definitions of Online-Trust

There are a number of different and inconsistent definitions of trust in the online retailing literature. A review of the literature suggests that there are two major reasons for this ambiguity. First there is a lack of specificity in identifying the referents of trust and second, the research often fails to differentiate between the antecedents of trust, specifically trustworthiness and initial trust, and trust itself (Mayer, Davis, & Schoorman, 1995; McEvily & Tortoriello, 2011; McKnight et al., 1998; Meyerson, Weick, & Kramer, 1996).

A special problem in defining online trust is this issue of whether the referent should be the online retailer, the Website, or the technology. Who is it that one is supposed to trust? Even the term online retail is not distinct enough since some retailers are only online, and some actually represent a bricks-and-mortar store with an online presence (McKnight & Chervany, 2002).

Can a website be the trust referent? Corritore, Kracher, and Wiedenbeck (2003) argue it can and suggest the trust in the website encompasses the underlying Internet technology, the interactive user experience, and the people behind the website. They define online trust as “an attitude of confident expectation in an online situation of risk that one’s vulnerabilities will not be exploited” (p. 740). Thus, their definition of trust is a description of a general attitude not an intention to make oneself vulnerable in a specific relationship. Does a person who trusts online retailing because he or she has many years of experience buying online and is willing to take the risk associated with doing so, trust the technology that they are familiar with or the retailer that is behind the sales?

If one accepts the definition of trust as a willingness to be vulnerable in a relationship the trust referent has to be more specific. It has to be clear whether the object of trust is the underlying Internet technology or a person behind the website since the willingness to be vulnerable depends on the specific relationship with this trust referent (Mayer et al., 1995). A person might trust in the Internet technology because he or she has many years of experience with online shopping in general and is willing to rely on the Internet technology transferring his or her order information steadily. The same person might hesitate to buy at a specific online store because he or she does not believe that the people behind the website are trustworthy. Some scholars argue that one can trust in technology and that technology can be regarded as the trust referent (Li, Hess, & Valacich, 2008; McKnight, 2005). Here, trust in technology is defined as the willingness of the trustor to behaviorally depend on a piece of software to do a task (McKnight, 2005). For example, a regular offline customer of a grocery store might hesitate to buy groceries at this store online since he or she is not willing to share his or her personal information on an Internet connection. In this case, the buyer trusts the brand but not the technology which suggests that technology can be a trust referent. Clearly there is a need for more research on this issue of whether a technology can be a trust referent absent the association with the store behind it.

Most of the research on trust in e-commerce specified the online retailer as the trust referent (e.g., Gefen, 2002; Gefen, Karahanna, & Straub, 2003). For example, Gefen and colleagues (2003) argued that an e-vendor is more than its interface since it has to be regarded as a business entity with whom the customer is economically interacting. Jarvenpaa, Tractinsky, and Saarinen (1999) define trust in the Internet store as “a consumer’s willingness to rely on the seller and take action in circumstances where such action makes the consumer vulnerable to the seller”(p. 4). In this definition they make clear that the e-vendor is a human being and that there is a relationship between the buyer and the seller. McKnight et al. (2002) define trust in an online retailer as “the intention to engage in trust-related behaviors with a specific Web vendor” (p.336). These definitions differentiate three relevant risks in the context of e-commerce: The willingness to provide the retailer personal information, the willingness to engage in a purchase transaction and the willingness to act on a retailer’s advice.

Trust and Risk in Online Retailing

As with the other disciplines discussed in this paper the relationship between trust and risk are critical to resolving the “trust” problem in online retailing. Many researchers have noted that the risk associated with online purchases is perceived to be both multifaceted and objectively higher than traditional commerce. Thus trust should play a more critical role in the purchase behavior of online customers (Corbitt, Thanasankit, & Yi, 2003; Wang & Emurian, 2005). Featherman and Pavlou (2003) define perceived risk in e-commerce as “the potential for loss in the pursuit of a desired outcome of using an e-service” (p. 454). Researchers have identified three sources of perceived risk in online retailing: product risk, financial risk, and privacy risk. Product risk refers to the uncertainty about the product quality (Ba & Pavlou, 2002) due to the fact that the product cannot be experienced physically before the purchase. Financial risk encompasses the risk of losing money in financial transactions associated with buying a product (e.g., Chen & Dubinsky, 2003). Privacy risk concerns the potential loss of control over personal information and the lack of confidentiality in the handling of personal data (e.g., Featherman & Pavlou, 2003; Liebermann & Stashevsky, 2002). These three sources of risks can be represented as antecedents of our perceived risk variable in the model when studying trust in the context of online retail (Fig. 2.5). Taken together, online retailing constitutes a context which could lead to a perception of a highly risky situation. Hence, risk-taking behavior of the customer such as purchasing online could be facilitated either by enhancing trust or by reducing the risk.

Fig. 2.5
figure 5

Antecedents of perceived risk

Full size image

Research has provided some guidance on how to solve the trust problem in online retailing. Fuller, Serva, and Benamati (2007) demonstrated in an experiment that reputation information influenced the perceived trustworthiness of an online retailer. In addition, feedback mechanisms (Ba & Pavlou, 2002), certifications from third parties (Shneiderman, 2000), web design features such as facial photos (Karimov, Brengman, Van Hove, & Van, 2011) and e-assurance structures (Bahmanziari, Odom, & Ugrin, 2009) have been proposed as means to enhance the perceived trustworthiness of a web site.

An interesting example of managing the trust problem comes from eBay. A senior manager in eBay, in a presentation to a group of executives, reported that they had found a way to solve the trust problem at eBay. The change that he described was to implement their “Buyer Protection Plan,” according to which any customer who was not satisfied with an eBay transaction and reported this would be given an immediate refund even before eBay investigated the transaction. At eBay, they believed they had changed the level of trust. But as our model would clearly indicate what they did was eliminate financial risk so that trust was no longer required.

In fact, referring to our conceptualization of trust, we would argue most of the strategies discussed and applied to enhance online trust are actually approaches to eliminate risk. For example, concerning reputation information, Fuller, Serva, and Benamati (2007) argued by supplying reputation information the uncertainty in the mind of the consumer is reduced. Regarding our three proposed types of risk antecedents in online retailing we can identify three related mechanisms to reduce the perceived risk: In order to deal with product risk online retailers establish return policies (e.g., Shneiderman, 2000). The perception of financial risk for the customer is reduced by mechanisms for buyer protections such as money back guarantees offered by eBay (eBay, 2014) or money-transfer-systems provided by a third party such as PayPal. Finally, in order to deal with privacy risk companies use privacy policies and publish them on their websites (e.g., Udo, 2001). Although these risk-reducing strategies are appropriate to facilitate risk-taking behavior and well established in the field of online retailing, it is noteworthy to keep in mind that they do not enhance trust but, in contrast, make trust less relevant in the context of online retailing.

Implications for enhancing the customer’s trust instead of reducing his risk can be derived directly from the definitions of trust in online retailing discussed above. First, depending on the object of trust the online retailer should either put effort in enhancing the user’s trust in the specific online retailer or in the Internet technology in general. Therefore, it would be interesting to investigate whether trust in the web vendor or trust in the technology is more critical for risk-taking behaviors such as purchasing a product online. Second, the antecedents of trust such as the factors of perceived trustworthiness (e.g., ability, benevolence, integrity) are relevant starting points in order to enhance the trust in the online retailer (Mayer et al., 1995; McKnight et al., 2002). Thus, a consumer perceives the online retailer as trustworthy when the retailer is willing and able to act in the consumer’s interest, when he is honest in transactions and capable of delivering as promised (McKnight & Chervany, 2002). The direct implication is that online retailers should ensure operational excellence and communicate their optimization efforts to the customer via their website. Finally, online retailers could provide information about the quality of their services and products on their websites; they can publish real customer ratings and reviews or can implement free consumer support via live-chat, e-mail, or service hotlines in order to enhance the customers’ perception of their ability, integrity, and benevolence. As with any relationship, the customer will update their assessments of the ability, benevolence, and integrity of the online retailer over time and with more successful transactions actually build trust in the retailer.

Politics and Government

There is a great deal of interest in the assessment of trust in politics and government. The Edelman Barometer of Trust is a report that is presented to those who attend the annual World Economic Forum in Davos, Switzerland each year (Edelman, 2014). This report presents the results of a survey of how much residents of each country trust their government and trust the business community. There is always a great interest in the relative scores for government and business and how they have changed over the year. In the U.S. the Pew Research Center measures the trust in government and several other similar indicators on an ongoing basis (Pew Research Center, 2014). The National Election Studies (NES) measure the trust in the federal government (2010). These measures were cited and used extensively in politics and public policy, and in the case of the Edelman report in international business. In each of these cases trust was measured with a single item that asks respondents how much they trust the government (or businesses). The single item used by NES is “how much of the time do you think you can trust the government in Washington to do what is right” (NES). Behavioral scientists would take considerable issue with the construction of the item from a methodological point of view in addition to the fact that it is a single item. The use of the word trust in the item allows every respondent to interpret the question according to their own definition of trust.

Although researchers have attempted to develop a common definition of trust there is a wide variation in the definitions of trust that are used in this field. Some examples of this diversity are confidence that citizens have in authorities to create fair policies and to serve the general interest of the public (Citrin & Muste, 1999; Miller, 1974), the willingness to rely on others (Doney, Cannon, & Mullen, 1998; Moorman, Deshpande, & Zaltman, 1993), and expectations that others will act in the trustor’s interests (Das & Teng, 1998; Elangovan & Shapiro, 1998; Hagen & Choe, 1998; Mechanic, 1998; Rousseau et al., 1998).

Several researchers noted trust judgments are made on the basis of ability and integrity (Keel, 2007; Sztompka, 1999) which are the trustworthiness factors in the trust model. There is also variation in the conceptualization of trust as a relational constructs or a dispositional construct (e.g., Uslaner, 2002).

From a theoretical point of view it is interesting to speculate whether it is meaningful to talk about trust in an institution that one has very little connection with at a personal level. When someone is asked if they trust the government, is it a different question than if they were asked if they liked the government, if they voted for the government, or if they had the same party affiliation as the government? The data from the Pew Research suggests that the answers to these questions are highly correlated. Also, the distal nature of the Federal government and the U.S. Supreme Court raise the question of whether there could be any behavioral consequences to trusting or not trusting.

The concept of legitimacy is well established as a means of evaluating and assessing these institutions. Legitimacy has been defined in terms of procedural justice, confidence, and competence. Perhaps we should focus on the legitimacy of these institutions and not confound it with the notion of trust.

Discussion and Conclusions

Preparing this chapter has pushed us to examine how trust is defined and used in many fields. It has forced us to examine our model of trust and test the generalizability of the model to other contexts. We have learned more about our own approach to thinking about trust as we adapted the model to new contexts. As we indicated at the outset of this chapter, our perspective is adopted from the lens of the Mayer et al. (1995) model and our comments on other fields clearly reflect this perspective.

An important set of issues for trust researchers in all fields is the critical importance of specifying a clear definition of trust and being consistent with that definition in the measurement of trust. We have reviewed what are generally considered the main definitions of trust, specifically, “willingness to be vulnerable,” “confident, positive expectations” and a dispositional measure. We believe it is important for researchers to be clear about what definition they will adopt. We could not have articulated this more clearly than Fulmer and Gelfand (2012): “To provide clarity, we recommend that researchers be very clear about the emphasis their definition places on positive expectations, vulnerability, both or neither and the rationale for this emphasis in relation to a particular referent and a particular level of analysis” (p. 1171). We see the choice of a definition of trust to be a two-stage decision tree with the first choice the one of choosing between a dispositional definition or a relational definition. If the choice is relational, then the next choice would be the choice above between the two relational definitions. Perhaps we can now get past the customary introductory paragraph in trust research that laments the great confusion about an appropriate definition of trust.

The next important issue for trust researchers to determine is if they will differentiate the antecedents of trust from trust itself. We have made our case for doing this as have many of the reviews of the trust literature (e.g. Colquitt et al., 2007; Fulmer & Gelfand, 2012). This approach gives us one clear definition of trust that applies to all contexts and the opportunity to build a nomological network of antecedents and consequences around it that are more specific to the level of analysis and context. We believe that being able to treat the factors of trustworthiness as separate constructs has given us much more flexibility in adapting our model to different situations. Our discussion of the different weighting of ability, benevolence, and integrity indicates the validity and explanatory power that come with this approach.

We have discussed the complementary roles of trust and control systems in managing risk, suggesting that these are not mutually exclusive processes. Understanding how to manage these processes can provide useful insight to managers at work. It has also been a revelation to examine what many disciplines refer to as a process of increasing trust through the lens of our model and recognize that what they are changing is the level of risk associated with a behavioral action and not trust at all. For a practical point of view they are, in fact, solving their trust problem, but from a scientific perspective it is important to note that they are not changing trust.