Introduction

Many nonprofit organizations (NPOs) rely to a large extent on volunteers—defined as people (a) performing activities out of free will, (b) without remuneration, (c) in a formal organization, and (d) benefiting others (Cnaan et al. 1996)—to deliver services to society (Farmer and Fedor 2001). NPOs try to keep their volunteers motivated and happy to insure the continuity and quality of these services. To attain this goal, NPOs engage in exchange relationships with volunteers (Farmer and Fedor 1999; Vantilborgh et al. 2011), meaning that they promise to offer certain inducements to volunteers (e.g., recognition) and in return they expect that volunteers make certain contributions to the NPO (e.g., arrive on time). These mutual obligations between the volunteer and the NPO form the psychological contract (Rousseau 1995). Such psychological contracts are important to study because they help explain why exchange relationships can thrive or deteriorate based on volunteers’ perceptions of the degree to which the NPO is living up to its end of the deal (i.e., psychological contract fulfillment). Research suggests that a lack of psychological contract fulfillment has deleterious consequences for paid employees, such as reduced satisfaction and performance (Bal et al. 2008; Zhao et al. 2007). However, a comprehensive empirical examination of volunteers’ responses to psychological contract fulfillment is currently missing in the literature (Nichols 2012). I address this gap in the literature, and build on the exit–voice–loyalty–neglect (EVLN) framework (Farrell 1983) to examine volunteers’ reactions to various degrees of psychological contract fulfillment.

This study makes three main contributions to the literature. First, little is known regarding volunteers reactions to psychological contract fulfillment. Hence, it remains unsure whether volunteers react similarly to psychological contract fulfillment as paid employees. The few studies to date that focused on volunteers suggest that psychological contracts are important to understand volunteers’ attitudes and behaviors (e.g., Starnes 2007) and that volunteers may react differently to psychological contracts than what one would expect from paid employees (Vantilborgh et al. 2012a). It therefore seems imperative to improve our understanding of volunteers’ reactions to psychological contract fulfillment. I relate fulfillment to the EVLN framework, which is ideally suited for this purpose as it captures a range of outcomes that have considerable practical importance. As a result, this study’s findings may inform volunteer managers on the reasons underlying volunteers’ behaviors. Second, I expand upon the previous studies that utilized the EVLN framework in the context of paid employment by distinguishing two types of voice behavior, namely aggressive and passive voice (Hagedoorn et al. 1999). This distinction is important as a high-quality exchange relationship (e.g., high psychological contract fulfillment) is likely to stimulate constructive, problem solving behavior (i.e., passive voice). In contrast, a low-quality exchange relationship (e.g., low psychological contract fulfillment) is likely to elicit destructive, contending behavior (i.e., aggressive voice). Put differently, both types of voice behavior may have distinct relationships with the degree to which the psychological contract is fulfilled. Third, the studies to date that linked psychological contract fulfillment to EVLN responses did not consider mediators (Lemire and Rouillard 2005; Si et al. 2008; Turnley and Feldman 1999). Mediators can help us to understand why fulfillment influences behavior, and hence, can be used to offer advice on how to manage volunteers’ reactions to (a lack of) psychological contract fulfillment. I include two variables that are considered important mediators in the psychological contract literature: violation (e.g., Zhao et al. 2007) and trust (e.g., Montes and Irving 2008). Surprisingly, these mediators have not yet been considered simultaneously in the psychological contract literature. This is nonetheless important, as controlling for the shared variance of both mediators would enable researchers to assess their unique contribution.

Psychological Contract Theory

The psychological contract can be defined as “an individual’s beliefs regarding the terms and conditions of a reciprocal exchange agreement between that focal person and another party” (Rousseau 1989, p. 123). This means that the psychological contract is idiosyncratic: it concerns the mutual promises that an individual perceives rather than the actual promises made between the individual and the organization. For example, employees may believe that their organization is obliged to provide training opportunities to them (even if the organization never promised this explicitly), and in return these employees perceive that they are obliged to work extra hours. The psychological contract garnered a great deal of scholarly attention, especially, in research on paid employees. The majority of this attention has been devoted to psychological contract fulfillment, which provides a compelling way to link the psychological contract to outcomes (Conway and Briner 2009). These studies often use the terms “breach” and “a lack of fulfillment” interchangeably, as they consider breach as the opposite of fulfillment. However, recent studies indicate that breach and fulfillment are distinct dimensions (Lambert et al. 2003; Vantilborgh et al. 2012a). I acknowledge this distinction and focus on fulfillment in this paper, which represents an individual’s general perception of the degree to which the organization is fulfilling its obligations, ranging from low fulfillment (few obligations fulfilled) to high fulfillment (many obligations fulfilled).

The norm of reciprocity dictates that individuals who perceive that their organization fulfills its obligations reciprocate with positive attitudes and behaviors, such as increased commitment and performance (Gouldner 1960). When employees perceive that their organization is not fulfilling its obligations, they reciprocate with negative attitudes and behaviors, such as lower satisfaction and withholding effort. Several studies in the context of paid employment demonstrated that low psychological contract fulfillment leads to unfavorable outcomes, such as reduced psychological wellbeing (Conway and Briner 2002a), trust (Grimmer and Oddy 2007; Robinson 1996), commitment (Grimmer and Oddy 2007), and performance (Robinson 1996), and increased negative emotions (Conway and Briner 2002b) and deviant behaviors (Restubog et al. 2007).

While psychological contract fulfillment has been shown to be a key concept to explain paid employees’ attitudes and behaviors, there is a dearth of research on volunteers’ reactions to (a lack of) fulfillment. Among the few studies that examined volunteers’ psychological contracts, the majority focused on the content of the contract (Nichols 2012). These studies examined what volunteers expect from their organization (e.g. Taylor et al. 2006; Vantilborgh et al. 2012b) and how these expectations are formed (e.g., Liao-Troth 2005; Vantilborgh et al. 2013). In comparison, relatively few studies investigated how volunteers respond to perceiving (a lack of) psychological contract fulfillment. These studies suggest that volunteers who perceive that their psychological contract is fulfilled engage in more activities (Farmer and Fedor 1999), volunteer more hours (Starnes 2007), and exert more effort (Vantilborgh et al. 2012a). However, they also provide tentative evidence that volunteers may react differently to psychological contract fulfillment than paid employees. For example, volunteers exerted more effort when NPOs did not keep value-driven promises, whereas one would expect paid employees to exert less effort (Vantilborgh et al. 2012a). These distinct reactions may be due to differences between volunteers and paid. On the one hand, volunteers are not bound by the usual ties of employment (Cnaan and Cascio 1998), meaning that they can easily leave the NPO in case of a lack of fulfillment. On the other hand, the moral commitment experienced by many volunteers (Cnaan and Cascio 1998) may cause them to react less strongly to a lack of fulfillment than paid employees. Given that there is evidence that psychological contracts matter for volunteers (e.g., Farmer and Fedor 1999; Starnes 2007) and that psychological contract research based on paid employees cannot be simply translated to volunteers (e.g., Vantilborgh et al. 2012a), it appears imperative to examine volunteers’ reactions to psychological contract fulfillment on a broad range of outcomes.

Exit, Voice, Loyalty, and Neglect Framework

The EVLN framework provides a typology of general responses to dissatisfaction in exchange relationships (Rusbult et al. 1988). Exit refers to voluntarily withdrawing from the organization by (thinking about) quitting, searching for a new job, or transferring to a different job in the same organization (Rusbult et al. 1988). Thus, it can also be considered as an intention to leave the job or organization, because actually leaving the organization may not be a viable option for everyone when confronted with unpleasant events at work (Naus and Roe 2007). Voice can be defined as “actively and constructively trying to improve conditions through discussing problems with a supervisor or co-workers, taking action to solve problems, suggesting solutions, seeking help from an outside agency like a union, or whistle-blowing” (Rusbult et al. 1988, p. 601). In other words, voice means that people actively and constructively try to improve working conditions (Naus and Roe 2007). People who respond with loyalty optimistically wait for conditions to improve, meaning that they publicly continue to support their organization and practice good citizenship (Rusbult et al. 1988), while hoping that everything will work out in the end (Naus and Roe 2007). People who respond with neglect passively allow conditions to deteriorate by reducing their interest or effort, frequently arriving late or being absent, or reducing the quality of their work (Rusbult et al. 1988).

Hagedoorn et al. (1999) further extended the EVLN framework by proposing that two types of voice responses can be distinguished. On the one hand, people can constructively react to unpleasant events with voice behaviors in an attempt to solve the problem, while considering both parties concerns (considerate voice). On the other hand, people can respond destructively to unpleasant events with the intent of “gaining” from the situation and without considering the organization’s concerns (aggressive voice). I believe that distinguishing considerate from aggressive voice is important, because aggressive voice is believed to arise when an individual perceives that there is misalignment between the statements and actions of the organization, which is typical for a lack of psychological contract fulfillment (Naus and Roe 2007). Consequently, I believe that it is essential to include all five types of responses in an extended EVLN framework when studying psychological contract fulfillment. These five reactions can be positioned in a circumplex (see Fig. 1) along the dimensions constructiveness versus destructiveness and active versus passive (Hagedoorn et al. 1999).

Fig. 1
figure 1

Circumplex of exit, aggressive voice, passive voice, loyalty, and neglect responses (Hagedoorn et al. 1999)

Full size image

Relating Psychological Contract Fulfillment to the Extended EVLN Framework

Exit

When psychological contract fulfillment is low, an individual may wonder whether staying in the organization is mutually beneficial for both parties (Turnley and Feldman 1999). For example, when volunteers perceive that they are not getting the recognition that they expect from the organization, they may decide to leave the organization in favor of other activities where they receive recognition. Previous studies on paid employees indeed show that psychological contract fulfillment is negatively related to exit responses (Si et al. 2008; Turnley and Feldman 1999) and to turnover intentions (Zhao et al. 2007). In the context of volunteering, Starnes (2007) finds no significant relationship between psychological contract fulfillment and turnover intentions. However, this result may be due to low-statistical power. In line with the norm of reciprocity (Gouldner 1960), I therefore hypothesize:

Hypothesis 1

Psychological contract fulfillment relates negatively to exit responses.

Aggressive and Considerate Voice

Turnley and Feldman (1999) reported that a lack of psychological contract fulfillment was positively related to voice behavior by paid employees. However, they did not distinguish between aggressive and considerate voice behavior. A lack of psychological contract fulfillment has been shown to create cynical attitudes in paid employees, as it leads people to believe that their organization lacks integrity (Johnson and O’Leary-Kelly 2003). In some cases, people may use a lack of fulfillment to their advantage, and use it as leverage to bargain for other inducements. For example, a volunteer who perceives that the organization failed to fulfill the promise of providing reimbursements for expenses may negotiate to get other inducements instead. In addition, it has been demonstrated that low psychological contract fulfillment elicits feelings of revenge and a need to retaliate in paid employees (Bordia et al. 2008). In line with the norm of reciprocity (Gouldner 1960), I therefore propose that volunteers will generally respond to low levels of psychological contract fulfillment with negative attitudes and behaviors, and to high levels of psychological contract fulfillment with positive attitudes and behaviors. As aggressive voice is a destructive response (Hagedoorn et al. 1999), I expect that it will be negatively related to psychological contract fulfillment. In contrast, as considerate voice is a constructive response (Hagedoorn et al. 1999), I expect that it will be positively related to psychological contract fulfillment.

Hypothesis 2

Psychological contract fulfillment relates negatively to aggressive voice responses.

Hypothesis 3

Psychological contract fulfillment relates positively to considerate voice responses.

Loyalty

When the psychological contract is fulfilled, people are likely to become more loyal due to the norm of reciprocity (Gouldner 1960). In other words, perceiving that the organization is fulfilling its obligations strengthens the bond between the individual and the organization (Sturges et al. 2005). Studies indeed show that paid employees who perceive high psychological contract fulfillment report increased levels of organizational commitment (Sturges et al. 2005). Likewise, Robinson and Rousseau (1994) show that paid employees respond to low fulfillment by reducing their loyalty. Within the EVLN framework, both Turnley and Feldman (1999) and Si et al. (2008) found support for a positive relationship between psychological contract fulfillment and loyalty responses. In line with these findings and with the norm of reciprocity (Gouldner 1960), I hypothesize that

Hypothesis 4

Psychological contract fulfillment relates positively to loyalty responses.

Neglect

Social exchange theory states that people who experience low psychological contract fulfillment may try to adjust the balance in the exchange by exerting less effort (Blau 1964; Zhao et al. 2007). For example, a volunteer may decide that there is no reason to work hard if she or he perceives that the organization is not providing the promised level of autonomy. In line with this, Turnley and Feldman (1999) and Si et al. (2008) demonstrated that psychological contract fulfillment was negatively related to paid employees’ neglect responses, meaning that people who perceived a lack of fulfillment exerted less effort, were more absent, and delivered lower quality of work. Support for this relationship can also be found in volunteering studies. For example, Starnes (2007) showed that psychological contract fulfillment was positively related to the amount of time spent volunteering, while Vantilborgh et al. (2012a) found that psychological contract fulfillment related positively to the effort exerted by volunteers. Based on these findings, I hypothesize that

Hypothesis 5

Psychological contract fulfillment relates negatively to neglect responses.

Violation and Trust as Mediators

While Turnley and Feldman (1999) and Si et al. (2008) examined how psychological contract fulfillment related to EVLN responses of paid employees, they did not consider the possibility that these relationships were mediated by other variables. Nonetheless, two mediators have been repeatedly used in psychological contract research, albeit separately: violation and trust. The former—violation—can be defined as “the emotional and affective state that may, under certain conditions, follow from the belief that one’s organization has failed to adequately maintain the psychological contract” (Morrison and Robinson 1997, p. 230). When an individual perceives low psychological contract fulfillment, she or he may develop feelings of anger, frustration, disappointment, and distress. The experience of such emotions can then lead to changes in attitudes and behaviors (Morrison and Robinson 1997), as they trigger specific action tendencies (Frijda et al. 1989). The idea that violation—constituting a mix of emotions—mediates the relationship between psychological contract fulfillment and outcomes is in line with affective events theory (Weiss and Cropanzano 1996). This theory states that events at work elicit certain emotions. These emotions, in turn, prompt certain attitudes and behaviors because they influence the content and the process of people’s thinking (Zhao et al. 2007). Several studies on paid employees have empirically demonstrated that violation mediates the relationship between psychological contract fulfillment and outcomes, such as deviant behaviors at work, job satisfaction, organizational commitment, intentions to quit, and organizational citizenship behavior (e.g., Bordia et al. 2008; Suazo 2009; Zhao et al. 2007).

Besides violation, trust has also been found to mediate relationships between a lack of psychological contract fulfillment and outcomes (Lo and Aryee 2003; Montes and Irving 2008; Robinson 1996). Trust can be defined as “one’s expectations, assumptions, or beliefs about the likelihood that another’s future actions will be beneficial, favorable, or at least not detrimental to one’s interests” (Robinson 1996, p. 576). When people perceive low psychological contract fulfillment, they may conclude that their organization cannot be trusted to fulfill other obligations that are part of the contract and that their organization does not care about them (Montes and Irving 2008). This lack of trust is then believed to have adverse effects on the individual’s attitudes and behaviors (Robinson 1996). Studies on paid employees indeed demonstrate that psychological contract fulfillment relates positively to trust, and that trust relates to several outcomes such as job satisfaction, civic virtue behavior, withdrawal behavior, turnover intentions, and employee contributions (Lo and Aryee 2003; Montes and Irving 2008; Robinson 1996). Moreover, trust is a key variable in the volunteering literature, as volunteers possess higher levels of trust than nonvolunteers (Bekkers 2012). Consequently, it makes sense to include trust in studies on volunteers’ reactions to psychological contract fulfillment. Zhao et al. (2007) report, based on meta-analytic findings, that trust mediated the relationships between psychological contract fulfillment and outcomes for paid employees, and that the effect sizes of these mediated relationships via trust were larger than those of the mediated relationships via violation. However, none of the primary studies included in their meta-analysis jointly examined violation and trust as mediators. This is nonetheless important, as both trust and violation have strong affective components (Zhao et al. 2007) and may therefore share variance. Hence, including both mediators in the same study is necessary to assess the unique contributions of violation and trust, after controlling for each other.

In sum, no mediators have yet been examined for volunteers’ reactions to psychological contract fulfillment, meaning that we do not know why volunteers react in a certain way to fulfillment. Theory and prior studies on paid employees suggest that violation and trust may explain why volunteers display less exit, aggressive voice, and neglect behavior, and more constructive voice, and loyalty behavior when perceiving high psychological contract fulfillment. I therefore hypothesize:

Hypothesis 6

Violation mediates the relationships between psychological contract fulfillment and exit, (aggressive and considerate) voice, loyalty, and neglect responses.

Hypothesis 7

Trust mediates the relationships between psychological contract fulfillment and exit, (aggressive and considerate) voice, loyalty, and neglect responses.

Method

Sample and Procedure

Because the experience of psychological contract fulfillment can be influenced by organizational characteristics and policies (Conway and Briner 2009), I decided to sample volunteers from a diverse set of organizations in various Belgian nonprofit sectors. This purposive sampling procedure aimed to increase the variance in psychological contract fulfillment. I contacted 428 NPOs from various sectors, of which 102 NPOs agreed to participate (response rate: 23.83 %). Participating NPOs were asked to distribute an email containing a link to an online survey among their active volunteers (henceforth called “volunteers survey”). In addition, NPOs were asked to send a second email containing a link to a different online survey to a board member of the NPO (henceforth called “board survey”). While the volunteers survey contained measures of the focal variables of this study, the board survey pertained to characteristics of the NPO (e.g., number of active volunteers, sector) that were later on used to describe the sample. Listwise deletion was employed, meaning that only those respondents (a) who completed the entire volunteers survey and (b) of whom the NPO board member also completed the board survey were retained. This resulted in a final sample of 215 respondents from 43 organizationsFootnote 1. The average number of respondents per organization was 5 (minimum = 1, maximum = 27).

The NPOs in the final sample were active in the domains of wellbeing (34.88 %), health (18.60 %), social and cultural minorities (11.63 %), nature and environment (9.30 %), youth-movements (6.98 %), socio-cultural events (6.98 %), sports and leisure activities (6.98 %), third-world aid (2.33 %), and mobility (2.33 %). The majority of the sampled organizations employed less than 10 paid employees (55.81 %), followed by 10–24 (13.95 %), 50–99 (13.95 %), 100–199 (4.65 %), 200–499 (4.65 %), and more than 500 paid employees (4.65 %). Most NPOs had between 25 and 49 active volunteers (32.56 %), followed by 50–99 (18.60 %), less than 10 (16.28 %), 10–24 (13.95 %), 200–499 (6.98 %), more than 500 (6.98 %), and 100–199 active volunteers (4.65 %). Most NPOS were solely active in Belgium (90.7 %), while some were international organizations (9.3 %). The sample contained slightly more female (60.38 %) than male volunteers, whereas the average age of the respondents was 48.06 years (SD = 16.82, minimum = 17, maximum = 80). The average tenure of respondents in their current organization was 7.87 years (SD = 7.18, minimum = 1 month, maximum = 44 years) and they, on average, volunteered for 5.51 h per week (SD = 6.49, minimum = 15 min, maximum = 40 h). The majority of the respondents had maximally attained a secondary school degree (39.44 %), followed by master degrees or higher (29.11 %), bachelor degrees (28.17 %), primary school degrees (1.88 %), and no degree (1.41 %). With the exception of gender (χ 2(41) = 70.92, p < .01), no statistically significant differences could be observed between sampled NPOs in terms of volunteers’ education (χ 2(164) = 191.54, ns.), age (F(1,163) = .23, ns.), tenure (F(1,205) = .60, ns.), or hours volunteered (F(1,202) = .52, ns.).

Measures

Unless mentioned otherwise, all items were translated to Dutch using back-translation. Some items were reworded to fit with the volunteering context (e.g. “employer” was changed into “organization”). Table 1 in the Appendix provides an overview of all items.

Table 1 Descriptive statistics and inter-correlations of focal variables (internal reliability scores of scales are shown between parentheses)
Full size table

I assessed perceptions of psychological contract fulfillment with the two-item fulfillment scale in Rousseau’s (2000) Psychological Contract Inventory. Respondents rated these items on a five-point Likert scale, ranging from “Not at all” (1) to “To a great extent” (5). This scale attained a good internal reliability score (r = .83). I measured violation with Robinson and Morrison’s (2000) scale, which consists of four items. These items were rated on a five-point Likert scale, ranging from “Strongly disagree” (1) to “Strongly agree” (5). The violation scale also attained a good internal reliability score (α = .95). Trust in the organization was measured with a scale from Robinson (1996), containing seven items that are rated on a five-point Likert scale, ranging from “Strongly disagree” (1) to “Strongly agree” (5). The internal reliability score of the trust scale was good (α = .84). I measured exit, aggressive voice, considerate voice, loyalty, and neglect with the Dutch version (Hagedoorn 1998) of the EVLN questionnaire by Hagedoorn et al. (1999). Respondents were asked to indicate how likely they were to respond to a problematic event at their voluntary work in a certain way, on a seven-point Likert scale ranging from “Definitely not” (1) to “Definitely yes” (7). The exit (6 items—α = .82), aggressive voice (7 items—α = .78), considerate voice (11 items—α = .90), and neglect (5 items—α = .75) scales all attained good internal reliability scores. However, the loyalty scale’s internal reliability score fell below recommended values (5 items—α = .50) and could not be improved by removing items. This scale also attained the lowest internal reliability score (α = .69) in Hagedoorn et al. (1999) study. I therefore did not include the loyalty scale in subsequent analyses. Finally, I included age, tenure, and gender as control variables based on prior research (e.g., Bal et al. 2008).

Analyses

I analyzed the data in Mplus version 7 using structural equation modeling to take measurement error into account. As the data has a nested structure—individuals nested within organizations—I used a stratification technique (type = complex) to correct standard errors for the nonindependence of the data (Satorra and Muthen 1995). I commenced by estimating a measurement model (i.e., confirmatory factor analysis) in which the items from the scales loaded on seven latent variables (i.e., psychological contract fulfillment, violation, trust, exit, considerate voice, aggressive voice, and neglect). This model was than compared against alternative measurement models. Next, I estimated paths between the latent variables (i.e., the structural model). Model fit was evaluated on a number of criteria (Schermelleh-Engel et al. 2003): the χ 2 value and its associated p value (.01 ≤ p ≤ .05: acceptable fit; .05 < p ≤ 1.00: good fit), the Comparative Fit Index (.95 ≤ CFI < .97: acceptable fit; .97 ≤ CFI ≤ 1.00: good fit), the Tucker Lewis Indicator (.95 ≤ TLI < .97: acceptable fit; .97 ≤ TLI ≤ 1.00: good fit), and the root mean square error of approximation (.05 < RMSEA ≤ .08: acceptable fit; 0 ≤ RMSEA ≤ .05: good fit). Finally, indirect effects were estimated based on the products-of-coefficients approach (Preacher et al. 2010).

Results

Descriptive Statistics

Table 1 shows the correlations between the focal variables. Psychological contract fulfillment correlates negatively with violation and exit, and positively with trust and considerate voice. Violation is negatively correlated with trust and considerate voice, and positively correlated with exit, aggressive voice, and neglect. Trust is negatively correlated with exit, aggressive voice, and neglect, while it is positively correlated with considerate voice.

Measurement Models

Table 2 shows the fit indices of the estimated measurement models. I commenced by estimating a measurement model based on the literature (Model A) with seven latent variables: psychological contract fulfillment (2 indicators), violation (4 indicators), trust (7 indicators), exit (6 indicators), considerate voice (11 indicators), aggressive voice (7 indicators), and neglect (5 indicators). This model offered a good fit to the data. Next, I estimated a couple of alternative measurement models. First, I estimated a model where trust and violation indicators loaded on one latent variable (Model B). While Model B still offered a fair fit to the data, the χ 2 difference test clearly shows that it fitted worse to the data than Model A. Model C was similar to Model A, except that the aggressive voice and the considerate voice indicators loaded on one latent variable. Model C does not fit adequately to the data in terms of CFI and TLI, although the RMSEA value indicates a good fit, and has a significantly worse fit to the data than Model A. Finally, in Model D, all indicators were allowed to load on a single latent variable. Model D did not fit adequately to the data and offered a significantly worse fit than Model A. Based on comparing these measurement models, I chose to continue with Model A to estimate the structural models. Table 1 in the appendix provides an overview of all items and their factor loadings, based on Model A.

Table 2 Fit indices of estimated measurement and structural models
Full size table

Structural Models

I commenced by estimating a direct effects model. This model included indicators and latent variables of psychological contract fulfillment, and exit, aggressive voice, considerate voice, and neglect responses. The indicators and latent variables of trust and violation were not yet included in this model. Paths from psychological contract fulfillment to the EVLN responses were freely estimated. As can be seen in Table 2, the fit indices of the direct effects model indicated that this model fitted well to the data, with the exception of the TLI which fell just below the recommended value. Looking at the paths from fulfillment to the EVLN responses, I find significant negative relationships between fulfillment and exit (ß = −.19, p < .01), aggressive voice (ß = −.17, p < .01) and neglect (ß = −.19, p < .001) responses. Hence, I can confirm hypotheses 1, 2, and 5, respectively. In line with hypothesis 3, I find a significant positive relationship between fulfillment and considerate voice responses (ß = .29, p < .001). I could not test hypothesis 4 as loyalty was not included in the structural model, due to the problems associated with this scale in the measurement model.

Next, I estimated a full mediation model, meaning that the direct paths from psychological contract fulfillment to exit, aggressive voice, considerate voice, and neglect responses were fixed to zero but that indirect paths from fulfillment to these outcomes via trust and violation were freely estimated. The fit indices of the full mediation model indicated that this model offered a good fit to the data. Likewise, the fit indices of the partial mediation model—in which the paths from psychological contract fulfillment to exit, aggressive voice, considerate voice and neglect responses were freely estimated—indicated a good fit to the data. The statistically nonsignificant χ 2 difference test implies that both models fit the data equally well. In case of equivalent models, it is recommended to proceed with the most parsimonious model (Byrne 2011). I therefore proceed with the full mediation model; a choice that is further supported by the fact that the direct paths from psychological contract fulfillment to exit (ß = −.01, ns.), aggressive voice (ß = .21, ns.), considerate voice (ß = .03, ns.), and neglect (ß = .06, ns.) were all statistically nonsignificant. Figure 2 shows the estimated path coefficients between the focal variables in the full mediation model, while Table 3 provides an overview of all estimated paths in the structural part of the full mediation model. As can be seen in this figure, psychological contract fulfillment was negatively related to violation and positively related to trust. In turn, violation is negatively related to considerate voice responses and positively related to aggressive voice responses. Violation is also positively related to exit responses, albeit only marginally significant. Trust relates positively to considerate voice and negatively to neglect responses. It is also noteworthy that violation and trust are negatively related to each other.

Fig. 2
figure 2

Estimates between focal variables (unstardized estimates)

Full size image
Table 3 Estimated paths in the full mediation model
Full size table

Turning to the indirect effects of psychological contract fulfillment on exit, aggressive voice, passive voice, and neglect responses, I find that there is an indirect effect of fulfillment on considerate voice responses (ß = .14, p < .05) and on aggressive voice responses (ß = −.16, p < .05) via violation. In other words, when volunteers perceive a low degree of psychological contract fulfillment, they experience more violation and, in turn, display less considerate voice and more aggressive voice responses. Moreover, I find an indirect effect of fulfillment on considerate voice (ß = .13, p < .05) and on neglect (ß = −.10, p < .05) responses, via trust. Put differently, volunteers who perceive a low degree of psychological contract fulfillment report lower levels of trust in the organization, which in turn relates to less considerate voice responses and more neglect responses. The indirect effect of psychological contract fulfillment on exit responses, via violation, was marginally significant (ß = −.17, p < .10). The remaining indirect effects—via violation on neglect responses (ß = −.09, ns.); via trust on exit responses (ß = −.01, ns.); and via trust on aggressive voice responses (ß = −.06, ns.)—were not statistically significant. In sum, these findings partially support hypotheses 6 and 7.

Discussion

The aim of the current paper was to examine volunteers’ reactions to various degrees of psychological contract fulfillment in terms of exit, (aggressive and considerate) voice, loyalty, and neglect responses. I hence addressed a major gap in the volunteering literature and expand on studies in the context of paid employment by distinguishing aggressive from considerate voice responses. Moreover, I investigated whether these relationships between psychological contract fulfillment and EVLN responses are mediated by trust and violation. While both mediators have been shown to be important in studies on paid employees, their joint influence has not been tested to date in a context of volunteering. I hence unravel why volunteers display certain behaviors as a reaction to psychological contract fulfillment.

Regarding the direct relationships between psychological contract fulfillment and EVLN responses, this study’s findings offered support for hypotheses 1, 2, 3, and 5. This means that volunteers who perceived low psychological contract fulfillment reported increased exit, aggressive voice, and neglect responses and decreased considerate voice responses. These findings indicate that a low degree of psychological contract fulfillment may lead volunteers to call into question whether the exchange agreement with their organization is worth continuing (Turnley and Feldman 1999). Volunteers may conclude that they are better of pursuing other (voluntary) activities and therefore experience a higher intention to leave the organization. However, volunteers may also decide to stay in the organization but to adjust the balance in the exchange agreement by exerting less effort and neglecting their tasks. For example, they may react by arriving late or by reducing the quality of their work (Rusbult et al. 1988). These findings of negative relationships between psychological contract fulfillment and exit and neglect responses are in line with prior studies in the context of paid employment (Si et al. 2008; Turnley and Feldman 1999). However, this study’s finding on the relationship with aggressive voice is novel as this type of voice response has not yet been included in psychological contract research. I show that volunteers who perceive a lack of psychological contract fulfillment respond with aggressive voice. This may be because they wish to use the situation as leverage to negotiate a better deal or because they wish to retaliate and take vengeance (Bordia et al. 2008). I also found that psychological contract fulfillment was positively related to considerate voice. Volunteers who perceive high levels of fulfillment respond with increased considerate voice. It is likely that volunteers engage in considerate voice behavior, which is a constructive response, as long as their exchange agreement with the organization is good. However, when their exchange agreement with the organization turns sour, this voice behavior becomes less constructive and thus more destructive. This finding shows the importance of including both aggressive and considerate voice responses in psychological contract studies. Finally, I could not test the direct relationship between psychological contract fulfillment and loyalty, as the loyalty scale had inferior psychometric properties. I return to this issue in the limitations section of the paper.

Turning to the role of violation and trust in the relationships between psychological contract fulfillment and EVLN-responses, the analysis clearly supports that violation and trust act as mediators. First, the results indicate that violation mediated the relationships between psychological contract fulfillment on the one hand and considerate voice, aggressive voice, and—albeit only marginally significant—exit responses on the other. In other words, volunteers who perceived low psychological contract fulfillment experienced a mix of negative emotions, such as anger, frustration, and disappointment (Morrison and Robinson 1997). These negative emotions (i.e., violation) were, in turn, related to an increase in cynical and aggressive voice responses as well as an increase in intention to leave the organization and a decrease in the use of constructive debate to solve the situation. Hence, these findings are in line with the extant literature on reactions of paid employees to psychological contract fulfillment and the mediating role of violation (e.g., Bordia et al. 2008; Suazo 2009; Zhao et al. 2007). In particular, these findings show that emotions are important for both paid employees and volunteers to explain reactions to psychological contract fulfillment. Second, this study’s findings illustrate that trust also acted as a mediator. In particular, trust mediated the relationships between psychological contract fulfillment on the one hand and considerate voice and neglect responses on the other. Put differently, when volunteers perceived that their organization did not fulfill its promises, they lost trust in the organization (Robinson 1996). This loss of trust was associated with a decrease in considerate voice responses and an increase in neglect responses. As such, these findings supplement the previous studies on paid employees that demonstrated the mediating role of trust in reactions to psychological contract fulfillment (e.g., Lo and Aryee 2003; Montes and Irving 2008; Robinson 1996).

It is noteworthy that trust and violation mediated different relationships. In particular, a sense of violation was primarily related to active and destructive responses, i.e., aggressive voice and exit, while trust was primarily related to passive and destructive responses, i.e., neglect. This distinction in response patterns might be tied to a “hot/cool system” of reactions (Metcalfe and Mischel 1999). On the one hand, violation, which entails a mix of emotions (Morrison and Robinson 1997), can be linked to a “hot” response system. Such a system is characterized by active, fast responses (Metcalfe and Mischel 1999). On the other hand, trust can be linked to a “cool” response system, as it is more cognitive. Such cool systems are characterized by more complex, slow responses (Metcalfe and Mischel 1999). As such, violation and trust may serve different functions in the process of psychological contract fulfillment. In some cases, a lack of fulfillment may elicit immediate affective responses, i.e., violation, which gives rise to active behaviors to address the situation. In other cases, a lack of fulfillment may elicit more cognitive responses, i.e., a loss of trust, which elicits passive behaviors. Moreover, whereas the “hot” response system is likely to be triggered immediately following a perceived lack of fulfillment and also to diminish quite rapidly, the “cool” response system is likely to develop more gradually following a perceived lack of fulfillment but may also last longer (Metcalfe and Mischel 1999). For example, a volunteer may immediately respond with feelings of anger when she or he notices that the psychological contract is not fulfilled and, in turn, develop intentions to leave the organization. After some time, this feeling of anger may diminish in the volunteer and be replaced by the feeling that the organization can no longer be trusted. Consequently, this person may exert less effort while volunteering. The idea that violation and trust are distinct but linked processes receives some tentative support from the negative correlation between both variables in this study.

In sum, this study illustrates that volunteers who perceive low psychological contract fulfillment report increased exit, aggressive voice, and neglect responses and decreased considerate voice responses. This is likely due to low fulfillment triggering either a “hot” emotional response system via violation and/or a “cool” cognitive response system via trust.

Limitations

A number of limitations need to be considered when interpreting the findings of this study. First, the cross-sectional nature of the study means that I cannot draw causal inferences. However, longitudinal studies support a temporal ordering in which fulfillment influences violation and trust (Robinson and Morrison 2000; Robinson 1996). Nonetheless, I believe that experimental studies are warranted to fully untangle the causal effects of psychological contract fulfillment. Second, all the data in this study are self-reported, which increases the chance that common method variance may have biased the estimates in the models (Podsakoff et al. 2003). However, the one-factor model that was estimated during the confirmatory factor analyses (i.e., Model D) offered a bad fit to the data. This lack of fit suggests that bias due to an unmeasured latent common method variable may not pose a significant problem (Podsakoff et al. 2003). Even so, future studies could benefit from using third-party ratings (e.g., supervisor rating of volunteer’s performance) or from using objective data (e.g., actual volunteer turnover rates). Third, the loyalty scale did not perform well in terms of internal reliability and was therefore excluded from subsequent analyses. It is possible that the items of this scale do not adapt well to a context of volunteering. However, it should be noted that the internal reliability score of the loyalty scale was also relatively low in (Hagedoorn et al. 1999) study. Given that loyalty appears to be a conceptually interesting variable in volunteering studies, I urge scholars to assess the validity and reliability of the EVLN measure in the volunteering context and to develop a revised scale if necessary. Finally, the study’s sample included volunteers from several organizations in various sectors. I chose this sampling strategy to increase the variance in the focal variables. For example, by only including volunteers from a single organization, one risks to obtain limited variance in reported psychological contract fulfillment because they are all exposed to the same organizational policies (Suazo et al. 2009). Moreover, by using a stratification technique, I corrected the standard errors for the nested structure in the data. Nevertheless, the study’s sample may not be representative due to self-selection bias. For example, nonprofit organizations undergoing major changes may have opted out of the study because they feared that respondents would paint a negative picture of the organization. It is therefore advisable to replicate this study’s findings in other samples.

Implications

Based on this study’s findings, I highlight two main implications for future research. First, the findings illustrate the importance of considering violation and trust concomitantly as mediators. Both variables may play a role in distinct processes: whereas violation can be linked to a “hot” response system, trust can be linked to a “cool” response system. However, given that both variables are negatively correlated—which is not surprising given that trust also has an affective component (Zhao et al. 2007)—studies that omit one of both variables may erroneously conclude that the included variable plays a significant role in both the “hot and cool” response systems. I therefore recommend future studies that consider mediators in the relationships between psychological contract fulfillment and outcomes to include both violation and trust. Moreover, research attention should be directed towards disentangling the unique role of both mediators. In particular, since “hot” and “cool” response systems are theorized to develop fast and slow, respectively (Metcalfe and Mischel 1999), future studies should examine how feelings of violations and trust develop and influence each other over time. Second, trust is commonly examined in a volunteering context, as volunteers have higher baseline levels of trust compared to nonvolunteers (Bekkers 2012). Consequently, trust may play a more important role in reactions to psychological contract fulfillment in a volunteering context compared to a paid employment context. I therefore recommend future studies on volunteers’ experiences of psychological contract fulfillment to pay special attention to trust. In particular, attention could be paid to how trust is repaired following a lack of fulfillment and whether a loss of trust has consequences for engaging in other volunteering activities.

Turning to the practical implications of this study’s findings, NPO managers should be aware that low psychological contract fulfillment might lead to destructive complaints of volunteers, volunteers neglecting their work, or volunteers showing an increased desire to leave the organization. Consequently, attempts should be made to fulfill volunteers’ psychological contracts. This requires that volunteers, local coordinators, and central staff members communicate openly on what they can realistically expect from each other. However, it is unlikely that an organization can fulfill the psychological contract of all its volunteers. In such cases, the mediators identified in this study provide some insights on how NPO managers can respond to volunteers experiencing a lack of fulfillment. If volunteers show a decrease in considerate voice behavior and an increase in aggressive voice and/or exit behavior, NPO managers should be cognizant that such a change in behavior can be due to an emotional response to low psychological contract fulfillment. In such cases, steps should be taken to reduce the negative emotions experienced by volunteers—for example by reappraising the situation—and to increase the experience of positive emotions. However, if volunteers show a decrease in considerate voice behavior and an increase in neglect behavior, NPO managers should check whether this behavior is due to a loss of trust. If this is the case, steps should be taken to repair trust. Lewicki and Wiethoff (2000) explain that this can be achieved by (1) addressing the behaviors that led to a loss of trust, (2) having each party responsible for the loss of trust apologize and provide an explanation, (3) renegotiating what all parties can expect from each other, (4) establishing evaluation procedures that are agreed upon by all parties, and (5) helping parties to establish alternative ways to meet needs.