Trust in Management Information Systems (MIS)

Perceived Trustworthiness of MIS

Following the general structure of the model of organizational trust (Mayer et al., 1995), we consider potential users of an MIS as trustors (i. e., the entity that makes her-/himself vulnerable) and the MIS as trustee (i. e., the entity that can potentially harm the trustor). Thus, the extent to which users perceive an MIS as trustworthy should be one main predictor of their experienced level of trust in the MIS. However, we deliberately distinguish trustworthiness perceptions and trust because perceived trustworthiness depicts the appraisal of an MIS’s characteristics, whereas trust describes an emergent state of an individual willing to depend and to make oneself vulnerable (Gefen et al., 2008; Mayer et al., 1995). While both constructs are positively related, the relationship is not deterministic and can be qualified by moderating factors such as specific experiences with the MIS over time.

Related distinctions between perceived trustworthiness and trust in technology have been made in the field of e-commerce: Komiak and Benbasat (2004) offered a distinction between cognitive trust, which can be seen as similar to perceived trustworthiness, and emotional trust, described as, “the trustor’s feeling toward the behavior of relying on the trustee” (Komiak & Benbasat, 2006, p. 944). In an experimental study, the authors found that both cognitive trust and emotional trust had positive effects on intentions to adopt a recommendation agent. Moreover, emotional trust partially mediated the effect of cognitive trust on the intention to adopt the recommendation agent (Komiak & Benbasat, 2006). These results are in line with our proposed distinction between perceived trustworthiness and trust, with the assumed partial mediation process between perceived trustworthiness, trust, and behavioral consequences. However, our model considers trust not exclusively as an affective state but as an experienced state that includes both affective and cognitive facets (for an initial operationalization, see the trust items used in Thielsch et al., 2018).

In general, we assume that users are more willing to rely on an MIS and to make themselves vulnerable to an MIS when they perceive the MIS to predictably and adequately fulfil their needs to master a situation. Consequently, users’ perceptions of the trustworthiness of an MIS should affect their experienced trust in the MIS. We further specify four different facets of trustworthiness perceptions that are grounded both in more general theoretical accounts (i. e., McKnight et al. 2011) as well as more recent empirical research (Thielsch et al., 2018). McKnight and colleagues (2011) considered three factors to be relevant trustworthiness perceptions of technologies: the extent to which a piece of technology has features that allow users to fulfil their intended tasks (functionality), the extent to which the technology operates in a continual and accurate manner (reliability), and the extent to which the technology provides help for the user (helpfulness). These factors have been widely adopted in information systems research in order to measure trust and its effects on the intention to explore, the intention to trust, or the intention to continue to use a specific information system (e. g., Lankton et al., 2015; Thatcher et al., 2011). Complementing and further extending these factors, Thielsch and colleagues (2018) applied an explorative approach to examine antecedents and consequences of trust in information systems at work using the critical incident interview technique (Flanagan, 1954). Their results suggest that users assess the trustworthiness of an MIS not only by considering features of the MIS itself, but also by assessing the credibility of the information that the MIS provides. Thielsch and colleagues (2018) successfully validated the explorative findings in a second study using a quantitative approach. Users may assess the trustworthiness of an MIS before their first use, for instance, based on conversations with colleagues, supervisors, or the technical support staff, or from their first impression of the user interface. Together, we propose:

Proposition 1: Individuals’ trust in an MIS is predicted by their perceived trustworthiness of the MIS (i. e., subjective perceptions of functionality, reliability, helpfulness, and credibility of provided information).

Disposition to Trust Technology

As in trust processes more generally, we assume trust in a specific MIS also to be affected by the trustor’s stable dispositions. Mayer and colleagues (1995) defined dispositional factors as the “general willingness to trust others” (p. 715). More specifically in the context of technologies as trustees, McKnight and colleagues (2011) defined a person’s propensity to trust technology as “the‍[ir] general tendency to be willing to depend on technology across a broad spectrum of situations and technologies” (McKnight et al., 2011, p. 7). The authors operationalized this concept through two factors: first, a person’s tendency to assume that technologies have favorable attributes (faith in general technology), and second, a person’s assumption that they can rely on technology in general (trusting stance in general technology). However, in later research, a person’s propensity to trust was operationalized exclusively through their trusting stance (Lankton et al., 2015). Initial empirical results indicated that such disposition to trust technology significantly affected trusting intentions and intentions to continue using a specific technology (Lankton et al., 2015).

In line with the general structure of trust-related processes outlined by Mayer and colleagues (1995), we argue that users’ disposition to trust technology affects the extent of their experienced trust in MIS. Users who generally tend to trust different technologies in various situations should experience higher degrees of trust in an MIS, whereas users who are generally unwilling to depend on technology should experience lower trust in an MIS. Thus, we propose:

Proposition 2: Individuals’ trust in an MIS is predicted by their disposition to generally trust technology.

Notably, trust in automation research (e. g., Lee & See, 2004) as well as the organizational trust model (Mayer et al., 1995) considered trust as an attitude. However, given that attitudes are rather stable (Fishbein & Ajzen, 1975), such a conceptualization of trust makes it difficult to represent the dynamic and reactive aspects of trust emergence and development over time. Therefore, we consider trust-related attitudes as part of individuals’ disposition to trust technology in general (see Figure 1), complementing other more stable dispositional influences on trust such as personality factors (e. g., extraversion, conscientiousness, and agreeableness; Furumo, de Pillis, & Green, 2008). By contrast, we consider trust as an emerging state, experienced by the trusting individual, and constantly influenced by the various factors considered. For instance, users’ trust in an MIS is likely to decrease as soon as an MIS works unreliably (e. g., after a defective software update), and trust should gradually increase again after programming errors have been fixed.

Trust in MIS, Behavioral Intentions, and Actual Use of MIS

Having established the concept of trust in MIS and its predictors, we now turn to the consequences of trust, differentiating trust from behavioral intentions and the actual use of an MIS. In online trust research, trust has widely been defined as the willingness to depend on someone or something (e. g., Gefen et al., 2008). Moreover, scholars in general have integrated behavioral intentions into their conceptualization of trust or referred to trust as a behavioral intention (e. g., Gefen et al., 2008; Mayer et al., 1995; McKnight et al., 2011). However, we argue that it is important to distinguish trust from behavioral intentions for two reasons. First, trust in general, and trust in MIS more specifically, becomes relevant when the trustor’s interests are at risk (Mayer et al., 1995), whereas behavioral intentions do not require potential vulnerability as a necessary precondition but precede any kind of deliberate behavior (Fishbein & Ajzen, 1975). Second, trust in MIS and behavioral intentions to use an MIS do not share the same antecedents. As argued earlier, trust in MIS is assumed to be based on trustworthiness perceptions of the MIS and on the general disposition to trust technology, whereas behavioral intentions are also affected by beliefs toward the behavior in question and social norms (e. g., Venkatesh & Davis, 2000). According to the theory of reasoned action, the intention to perform a certain behavior can be considered as the subjective likelihood to conduct a certain behavior (Fishbein & Ajzen, 1975). For MIS use, we assume that trust in MIS is only one out of several factors that positively affect one’s intention to use an MIS. Nevertheless, if users are willing to depend on and be vulnerable to an MIS, they should have a greater intention to use it.

Proposition 3: Individuals’ trust in an MIS partly predicts their behavioral intentions to use the MIS.

Moreover, we assume that intentions to use an MIS enhance the likelihood that users actually use the MIS. Please note that this prediction is not trivial because behavioral intentions are not the only predictors of actual usage behavior (e. g., Bhattacherjee & Sanford, 2009). Other factors that affect technology use at work are, for instance, developed habits and routines, prescribed work processes, or legal regulations. Therefore, we assume that behavioral intentions only partly predict actual use of MIS:

Proposition 4: Individuals’ behavioral intentions to use an MIS partly predict their actual use of the MIS.

Perceived Risks and Contextual Constraints as Moderating Conditions

Mayer and colleagues (1995) considered risk in two different ways. First, they integrated risk in the trust definition because they described the “willingness to make oneself vulnerable” also as the willingness to take risks. Second, the authors considered perceived risk to be relevant for the manifestation of trust, that is, for the translation into actual risk-taking behavior. Applied to trust in MIS, the former includes risks inherent in the relation between a user and an MIS, whereas the latter refers to contextual conditions and comprises the trustor’s belief in the likelihood of gains or losses due to these perceived contextual risks (see Mayer et al., 1995). For instance, even when users trust an MIS (because the MIS is perceived as trustworthy and dispositional trust factors are high), they should not intend to rely on the MIS if context conditions are too risky (e. g., when potential mistakes are too costly, or when the management disapproves of using technologies). Consequently, we contend that perceived risks due to context conditions represent a crucial moderator for the effect of trust on behavioral intentions to use MIS. Moreover, we argue that perceived risks are particularly relevant for the intention formation process when users deliberately consider gains and losses of MIS usage.

Trust and related risks have been examined in e-commerce studies on private usage of (mainly commercial) online services, in which websites or recommendation agents were the trustees. Results have indicated that e-commerce customers perceived risks particularly with regard to theft of payment details and personal information (e. g., Kim, Ferrin, & Rao, 2008; Liebermann & Stashevsky, 2002). In work-related settings, different risks are to be considered when workers seek to make a living and to fulfil personal needs (e. g., growth, achievement, self-esteem) and social needs (e. g., relatedness, affiliation; see Haslam, Powell, & Turner, 2000). Prominent examples for work-related risks are potential losses regarding social status or career progress.

Social status can be defined as, “respect, admiration, and voluntary deference individuals are afforded by others” (Anderson, Hildreth, & Howland, 2015, p. 1) and represents an important part of workers’ self-esteem. If this social status is threatened when an MIS is used, for instance, when supervisors are overtly critical when errors occur, the intention to rely on an MIS should decrease even if the MIS is perceived to be trustworthy.

Risks regarding career progress relate to negative repercussions for users’ careers. For instance, if users are held accountable for decisions made based on an MIS, the intention to rely on the MIS should decrease even if the MIS itself is trusted. In such situations, users are more likely to apply workarounds, for example, by running extensive double-checks with source data. Together, the effects of experienced trust in an MIS on the intention to use the MIS is assumed to be qualified by contextual conditions that increase or decrease the potential risks related to MIS use. Thus, we postulate:

Proposition 5: The relation between users’ trust in an MIS and their intention to use the MIS is moderated by their perception of related risks when using the MIS: Higher perceived risks (potential losses) reduce the effect of users’ trust on users’ intentions to use the MIS.

In addition to moderating effects of users’ risk perceptions on the relation between trust in an MIS and deliberate intentions to use the MIS, we also consider moderating conditions of the relation between deliberate intentions to use an MIS and the actual usage (i. e., behavior) of the MIS (see Figure 1). In general, the intention–behavior link can be influenced by numerous factors, such as type and properties of intentions and behavior, personality factors, or action control and habits (e. g., Sheeran, 2002).

Applied to our model of trust in MIS, trust-based intentions to use an MIS are only one out of many factors that determine the actual usage of an MIS. Other influencing factors do not directly relate to experienced trust of the user but reflect contextual conditions such as users’ control and autonomy at work, supervisory instructions, or social norms and work routines. These factors not only directly affect the likelihood of MIS use, but they also moderate the strength of the relation between users’ trust-based intentions to use an MIS and actual usage behavior. For instance, perceived behavioral control can affect usage behavior both directly and indirectly through trust-based intentions (Ajzen, 2002). More generally, we assume that the relation between trust-based intentions to use an MIS and actual MIS use is qualified by environmental factors that simplify or impede the voluntary control of the individual user. Thus, we propose:

Proposition 6: The relation between users’ intentions to use an MIS and the actual use of the MIS is moderated by contextual factors: Factors facilitating users’ control increase the relation between users’ intention to use an MIS and actual usage of the MIS.

MIS Use and Experiences With the MIS

In our proposed model of trust in MIS, actual MIS use is the main outcome variable, where actual MIS use reflects the secure dependence or reliance on the MIS instead of trying to control the MIS (McKnight, 2005). That is, actual MIS use refers to the application of an MIS in organizational work processes without individual users conducting unnecessary workarounds or consulting additional information sources. However, actual MIS use is not only considered as an outcome but also as a precursor for following trustworthiness assessments. When using an MIS, individuals gain experience with the MIS, which enables further evaluations and re-evaluation of the trustworthiness of the MIS. Thus, MIS use can change perceived trustworthiness in repeated feedback circles:

Proposition 7: The evaluation of an MIS during usage affects the perceived trustworthiness of an MIS in subsequent usages.

Moreover, as proposed by Schoorman and colleagues (2007), dispositional factors with respect to trust should be particularly relevant in the initial phases of trust building, when trustors have only few personal experiences with the trustee. In our model of trust in MIS, we assume that users rely on more general attitudes and dispositions when having only few or no personal experiences with an MIS. Such general attitudes and dispositions might also influence users’ initial impression of an MIS’s interface, or information about the MIS from others. By contrast, systematic assessments of the trustworthiness of an MIS should become more relevant over time when users gain more personal experience with the MIS. Together, we assume that experiences with an MIS moderate the relative impact of MIS trustworthiness perceptions and dispositions to trust technology:

Proposition 8a: The effect of users’ trustworthiness perceptions on trust in an MIS is moderated by users’ experience with the MIS: More experience with the MIS increases the effect of perceived trustworthiness on trust.

Proposition 8b: The effect of users’ disposition to trust in technology on trust in an MIS is moderated by users’ experience with the MIS: More experience with the MIS decreases the effect of users’ disposition to trust technology.

Research Implications

The presented model provides interesting guidelines for future research. The proposed distinctions between perceived trustworthiness, trust in MIS, and behavioral intentions to use MIS need to be examined empirically, together with considered moderating factors. Moreover, measuring trust in MIS reliably and distinctly from related constructs is still an unresolved topic. Even though scales are provided for measuring trustworthiness factors (e. g., McKnight et al., 2011), emotional trust (Komiak & Benbasat, 2006) and behavioral intentions (e. g., Venkatesh, Morris, Davis, & Davis, 2003), more research is needed to examine whether these scales truly distinguish between the related concepts. To the best of our knowledge, no extensively tested scale is available that applies to the integral psychological state of trust in MIS as it is proposed in the current work. The need for appropriate scales and measurement models is emphasized by measurement model misspecifications that have previously been identified in information system trust research (Söllner & Leimeister, 2013). The moderating effect of perceived risks and contextual factors represents another starting point for future research. Further research is also needed to assess what exactly constitutes perceived risks and contextual factors.

For reasons of parsimony, we focused on the main relationships between the concepts in our proposed model. Nevertheless, additional relationships are conceivable and might be addressed in further investigations. For instance, users’ disposition to trust might moderate experiences with an MIS. Users with low disposition to trust in technology may focus on negative information about the MIS, leading to a more biased perception of MIS’s trustworthiness.

Moreover, differentiating trust from behavioral intentions allows for the analysis of specific trust effects that go beyond mere MIS use. For instance, recent empirical data have shown that trustful usage of MIS at work was accompanied by significantly higher levels of well-being, performance, and post-usage satisfaction as compared with distrustful usage of MIS (Thielsch et al., 2018). These results suggest that trust is not only relevant for actual MIS use but can also contribute to less strenuous working conditions. Given the constantly increasing volumes of data in modern organizations, more research is needed on health determinants that are related to MIS use and to trusting in MIS. As an initial example, a recent experimental study showed that MIS use not only improved the well-being of its users and enhanced individual performance, but it also liberated mental resources for unrelated additional tasks (Hertel at al., 2019). More research is needed in order to clarify the relevance of trust in the related MIS in terms of enhancing performance and well-being.

Additionally, further investigations are needed on the nature of trust. In our proposed model, we understand trust as an experienced state that constantly reacts to trustworthiness perceptions and changes in the course of actual MIS use. With this conceptualization, we draw on the call for more temporalism in applied psychology (Roe, 2008). Still, little research has been conducted on the dynamic properties of trust. Longitudinal studies are desirable, including studies that examine trust in MIS at various points, for example, during a working day. Indeed, the proposed impact of actual MIS use and the related change in perceived trustworthiness and consequently in trust would entail an analysis of trust in various stages of use.

In general, the boundary conditions of the presented model need to be examined. Even though we determined boundary conditions by focusing on vocational management settings, broader contextual factors, such as culture (Leidner & Kayworth, 2006), merit further examination. While the presented model focuses on trust, distrust may involve additional influencing factors or different processes (e. g., Seckler, Heinz, Forde, Tuch, & Opwis, 2015, Thielsch et al., 2018) providing another starting point for further investigation.

Implications for Practitioners

The proposed model entails three major practical implications to enhance MIS use and, thus, to reduce overload through growing amounts of organizational data: enabling trust, reducing perceived risks, and diminishing contextual constraints. First, in order to enable trust, an MIS has to be (perceived as) reliable, functional, and helpful and to provide credible information. Thielsch and colleagues (2018) offer practical implications for enhancing trust in information systems at work, as well as for avoiding distrust. Their implications comprise system quality, information quality, service quality, context, and persons involved. In addition, experiences with the MIS and realistic information about the MIS’s favorable attributes can lower the potentially biased impact of dispositional trust.

Second, reducing the perceived risks of the MIS increases the probability that trust will result in intentions to use the MIS. Therefore, contextual conditions of MIS use such as leadership behavior or organizational policies should be reconsidered and adapted if necessary. A context in which potential personal gains exceed potential personal losses should lower perceived risks and therein make actual MIS use more probable.

Third, in order to strengthen the effects of trust-based intentions to use the MIS on actual MIS use, organizations might consider contextual factors that impede MIS use during the MIS’s implementation, such as to assure easy access and use of the MIS (e. g., Venkatesh, 2000) or to train and support users. In addition, MIS use might be positively influenced by increasing users’ control during the implementation process, through involving them in the acquisition, configuration, and potential changes of MIS relevant to their workplace. Moreover, organizations might review and redefine decision-making practices and support from top management (e. g., McAfee & Brynjolfsson, 2012).