The Long‐Term Effectiveness of Mediating Workplace Conflicts Open Access

Empirical research on workplace mediation effectiveness has shown that, in general, mediation has good settlement rates, varying from 60 percent to 80 percent (Kim et al. 1993; Mareschal 2005; Wood and Leon 2005; Swaab and Brett 2007; Poitras and Le Tareau 2009), with some variation depending on, for example, conflict intensity, type of conflict before the mediation (i.e., relationship conflict versus goal conflict) (Swaab and Brett 2007), as well as parties’ collaborative orientation, relationship hostility, and the mediator’s skill base (Mareschal 2005). Study findings also indicate that most participants feel satisfied with the mediation process, its outcomes, and the mediator and have a high level of confidence in the agreement (Poitras and Le Tareau 2009; Bollen, Euwema, and Müller 2010; Bollen, Ittner, and Euwema 2012).

These studies have typically measured effectiveness shortly after the mediation ended (mostly within four weeks) and have therefore presented only short‐term effects (McDermott et al. 2002; Bingham 2003, 2004; Conlon 2005; Bingham 2012; Bollen and Euwema 2013). Are these short‐term effects sustained over the longer term? In other words, what are the effects of workplace mediation over a longer term, one year after the mediation?

This question is relevant because scholars in other fields of mediation such as community mediation and family mediation have argued that short‐term mediation success may not necessarily predict long‐term mediation success (Pruitt et al. 1993; Emery, Sbarra, and Grover 2005) because only after a considerable amount of time can parties fully appreciate the consequences of what they agreed to during the mediation.

Surprisingly, we do not know of any study reporting how parties perceive the long‐term effectiveness of workplace conflict mediation. The current study seeks to explore this.

We first examine whether and how parties’ perceptions of short‐term mediation effectiveness relate to their perceptions of long‐term mediation effectiveness. We then test how hierarchical position and the (dis)continuation of an employment relationship affect parties’ perceptions of long‐term effectiveness in order to illuminate the impact of context on mediation effectiveness.

Previous studies have shown that such mediation context characteristics as hierarchical relations, legal frameworks, organizational culture, and the decision to end the employment contract (or not) can affect or influence the mediation and its effects (Coleman et al. 2014; Coleman, Kugler, and Mazzaro 2016). Three dimensions in particular – regulations, roles, and relations – seem most relevant (what we have called previously the “3‐R model”; Bollen, Euwema, and Munduate 2016). The relationship dimension can refer to the characteristics of the relationship between the disputants, as well as their relationship with the mediator.

In this study, we focus on the disputants’ relationship only, by taking into account participants’ hierarchical position and the (dis)continuing nature of the work relationship. Workplace mediations frequently involve hierarchical conflicts, that is, subordinates in conflict with their supervisor, manager, or employer (Uitslag, Kalter, and de Gruil 2011). Previous studies of this type of mediation have shown that the relative hierarchical positions that parties occupy affect their perceptions of mediation effectiveness in the short term (Bollen, Euwema, and Müller 2010; Bollen, Ittner, and Euwema 2012; Bollen and Euwema 2014; Bollen, Verbeke, and Euwema 2014). Therefore, our second question is to what extent supervisors and subordinates perceive mediation effectiveness differently over the long term.

As mentioned before, we also focus on the long‐term effectiveness of mediations in which parties arrange the termination of their employment relationship, also known as “exit mediations” (Rasmussen and Greenwood 2014; Munduate, Bollen, and Euwema 2016). To our knowledge, no empirical research has been conducted on exit mediations.

Thus, the three questions we ask in this study are (a) What is the relation between short‐term and long‐term mediation effectiveness? (b) To what extent are perceptions of long‐term mediation effectiveness affected by disputants’ hierarchical position? and (c) To what extent do perceptions of long‐term mediation effectiveness differ between participants in exit mediations versus nonexit mediations?

Until now, no empirical studies have measured the long‐term effectiveness of workplace mediation, and consequently we know little about whether reported positive outcomes are sustained. For disputants, this could raise the question of whether they should participate in a process whose long‐term effectiveness is uncertain. For employers, it similarly raises the question of how many resources they should expend promoting and implementing that same process. It is important for organizations, mediation practitioners, and disputants to know whether workplace mediation is an effective and durable means of dispute resolution.

In addition, some research has suggested that indeed long‐term mediation outcomes can be different from short‐term outcomes (Emery, Sbarra, and Grover 2005; Bollen and Euwema 2013). Perceptions of mediation effectiveness can change as time passes and parties’ perspectives on the practical consequences of the mediation outcome shifts (Donnelly and Ebron 2000; Kaiser and Gabler 2014). Parties who feel satisfied and relieved that a conflict seems to be over may feel differently if the agreement turns out to be less advantageous than they had imagined, if implementation difficulties arise, or if external parties (family members, friends, colleagues) express negative opinions of the agreement.

Several studies in the context of court‐connected, community, and family mediation have found an ambiguous relationship between short‐ and long‐term mediation effects (Van Slyck, Stern, and Newland 1992; Pruitt et al. 1993; Emery, Matthews, and Kitzmann 1994; Donnelly and Ebron 2000; Kaiser and Gabler 2014). A study by Peter Kaiser and Andrej Marc Gabler (2014) on court‐connected mediation showed a positive relationship between short‐term and long‐term satisfaction with the mediation experience. A study on community mediation by Dean Pruitt and his colleagues (1993) showed a positive relationship between short‐term satisfaction with the mediation and the absence of new problems in the long term. In addition, follow‐up studies in family mediation research showed both negative (Center for Families, Children and Courts 1993; Kelly 2004) and positive relationships (Van Slyck, Stern, and Newland 1992; Center for Families, Children and Courts 2000; Donnelly and Ebron 2000; Kelly 2004) between short‐term and long‐term mediation effectiveness as perceived by clients.

In previous studies, data were collected from three months (Van Slyck, Stern, and Newland 1992) to three years after the mediation (Donnelly and Ebron 2000). In our current study, we chose an intermediate time lag of one year (De Cuyper et al. 2012; Kaiser and Gabler 2014), allowing enough time for the results of the agreement to be felt, while reducing the likelihood that conflict parties would forget about the mediation or that parties would become unreachable (i.e., move away) (Pruitt et al. 1993).

Whether the mediation produced an agreement or not is a typical measure of short‐term effectiveness (Hollett et al. 2002; Kelly 2004). In contrast, long‐term measures can include both “hard” and “soft” indicators. Hard indicators (which are still subject to interpretation) can include whether there has been compliance with the agreement, and “soft” indicators include disputants’ perceptions, such as their satisfaction with the mediation process, outcome, and/or mediator, and whether they reconciled with each other. The extent of reconciliation between parties, in particular – whether they can interact positively and with trust and respect toward each other – may only be perceptible over time (Lederach 1999; Moore 2014). In our study, we measured perceptions of reconciliation; satisfaction with the mediation process, outcome, and mediator; and the extent to which parties perceive that the other conflict party complied with the mediation agreement (Pruitt et al. 1993). The presence of such perceptions would suggest that the mediation was successful because it achieved the goal of producing long‐lasting agreements honored by both parties (see McCorkle and Reese 2015).

Most of the studies of other types of mediation have shown that parties perceive mediation effective in the long term (Van Slyck, Stern, and Newland 1992; Pruitt et al. 1993; Donnelly and Ebron 2000; Kaiser and Gabler 2014), although some others were more ambiguous (Emery, Matthews, and Kitzmann 1994). In sum, studies show that disputants perceive mediation effective in the short term (Poitras and Le Tareau 2009; Bollen, Euwema, and Müller 2010; Bollen, Ittner, and Euwema 2012) and often also in the long term (Van Slyck, Stern, and Newland 1992; Pruitt et al. 1993; Donnelly and Ebron 2000; Kaiser and Gabler 2014).

We assume that perceptions of long‐term mediation effectiveness can be explained by its results in the short term. Consequently, our first hypothesis is as follows:

Conflicts between supervisor and subordinate are described as hierarchical conflicts. In hierarchical conflicts, supervisors are typically more powerful than their subordinates (Euwema 1992; McKenzie 2015). Many studies show that hierarchy and power have a profound impact on disputants’ feelings, cognitions, and behaviors (Van de Vliert, Euwema, and Huismens 1995; Keltner, Gruenfeld, and Anderson 2003; Galinsky, Rucker, and Magee 2016; Guinote 2017). Hierarchy continues to play a role in workplace mediation (Wiseman and Poitras 2002) and affects how parties perceive short‐term mediation effectiveness (Bollen, Euwema, and Müller 2010; Bollen, Ittner, and Euwema 2012; Bollen, Verbeke, and Euwema 2014). In general, it takes more to satisfy subordinates than supervisors immediately after the mediation. Supervisors typically feel more satisfied with the mediation and perceive the process as more effective than do subordinates.

In addition, the results of these studies have shown that the conditions for an effective mediation are different for supervisors and subordinates. For example, for subordinates, having the mediator acknowledge their anger can enhance their perceptions of mediation effectiveness (Bollen and Euwema 2014), but this is not the case for supervisors. Similarly, subordinates’ perceptions of procedural justice enhance their perceptions of mediation effectiveness, whereas experiencing uncertainty about the mediation is likely to diminish their sense that the process is effective, but these perceptions do not have the same effect on supervisors (Bollen, Euwema, and Müller 2010; Bollen, Ittner, and Euwema 2012).

Studies of hierarchy and power have shown pervasive effects that endure even beyond the particular social context in which they were initially experienced (Galinsky, Rus, and Lammers 2011). These findings suggest that the effects of hierarchical positioning on perceptions of the conflict and of the mediation process could still be felt after a longer period of time. Thus, our second hypothesis is as follows:

Another possible indicator of mediation success is parties’ long‐term compliance with the agreement (Pruitt et al. 1993; Long 2003; Herrman, Hollett, and Gale 2006; McCorkle and Reese 2015). Parties’ relative hierarchical position in the conflict may also have some effect on how they perceive their counterpart’s level of compliance with the agreement. We assume that compared to supervisors, subordinates will perceive that the supervisor is less compliant with the mediation than vice versa because supervisors generally have more freedom to behave as they want when compared to their subordinates (Keltner, Gruenfeld, and Anderson 2003; Galinsky, Rus, and Lammers 2011; Galinsky, Rucker, and Magee 2016) and may thus feel freer to comply or not to comply with the agreement. At the same time, it seems more likely that supervisors would correct subordinates if they believe the subordinate is not living up to the agreement. Subordinates in contrast feel more constrained in their behavior (Keltner, Gruenfeld, and Anderson 2003) and as such may feel less capable of speaking out to make sure that the supervisor behaves according to the mediation agreement. In addition, the psychological experience of power increases general optimism (Anderson and Galinsky 2006; Galinsky, Rus, and Lammers 2011), which might also encourage supervisors, who are relatively more powerful, to perceive greater compliance with the agreement (Bollen, Verbeke, and Euwema 2014). Therefore, the last part of our second hypothesis is:

Broadly speaking, a hierarchical conflict in the workplace can be practically resolved by mediation in three ways: (a) both parties remain employed by the organization, and agreements are made to avoid future conflicts; (b) parties remain employed by the organization, however, one or both parties transfer to another department or different job within the organization; (c) one of the parties leaves the organization (Uitslag, Kalter, and de Gruil 2011). If the last option occurs, then the mediation becomes an “exit mediation” (Latreille and Saundry 2014; Munduate, Bollen, and Euwema 2016). It is more common for a subordinate to leave the organization because of a conflict, rather than the supervisor (Uitslag, Kalter, and de Gruil 2011).

Although exit mediations occur commonly in many countries (McAndrew, Morton, and Geare 2004; Walker and Hamilton 2009; Rasmussen and Greenwood 2014), these are arrangements that differ significantly from place to place. The “regulations” dimension of the 3‐R model is relevant in this regard (Bollen, Euwema, and Munduate 2016). In the United States, for example, exit mediations will usually only take place if a discharged employee files a formal complaint and both parties agrees to mediate or if the parties activate some kind of an internal grievance system in place at the workplace (Menkel‐Meadow 2014). Thus, often by the time parties seek mediation (most often the employee), their employment relationship is already over and they use the process to come to a suitable settlement (Rasmussen and Greenwood 2014). In South Africa, for example, there are two systems: a statutory system that requires all disputes dealing with alleged unfair dismissal to be referred for “conciliation” (a quick and robust form of mediation), and a voluntary system of mediation. Although most disputes going to mediation or conciliation in South Africa involve employees who have already been dismissed, sometimes an employee in either system will refer a matter to conciliation or mediation when an ongoing employment relationship remains (Jordaan and de Wulf 2016).

In the Netherlands, where we conducted this research, exit workplace mediations are rather common. Human resources professionals and occupational physicians play a large role in the choice to mediate. Workplace mediation got a boost after the introduction of the Improved Gatekeeper Act (2002) and the 2004 extension of the 1996 Wage Payment during Sickness Act. The Improved Gatekeeper Act facilitates an effective return to work for employees following medical leaves that take longer than six weeks and in which workplace conflict plays a role as assessed by the occupational physician. In these cases, workplace mediation is often advised as a means to solve the problem. Generally, Dutch companies are obliged to collaborate with an occupational physician to ensure a healthy workplace. The Wage Payment during Sickness Act states that when a worker becomes sick, the employer is obliged to continue paying 70–100 percent of the salary for up to two years, during which the worker is protected by law against layoff (De Jong, Everhardt, and Schrijvershof 2011; Hoefsmit, de Rijk, and Houkes 2013; Vossen and van Gestel 2015.) These laws create urgency for employers and employees to resolve their conflict in a constructive way and raise the question of whether a continuation of the employment relationship is appropriate or not (Organization for Economic Cooperation and Development 2015).

In the Netherlands, the decision to terminate the employment can be made before mediation begins (by one or both parties) or during the mediation process itself. In the latter case, employer and employee may intend to maintain their relationship, but conclude during the mediation process that terminating the employment contract is the best way to resolve the conflict; or, alternatively, one of the parties may intend to continue the employment relationship but discover during the mediation process that his or her counterpart is unwilling to do so (Van de Griendt and Schutte 2006).

Regardless of whether or not both parties initially intend to continue the working relationship, the main distinguishing characteristic of an exit mediation is that, at the end of the process, the parties terminate the relationship and arrange a settlement. Thus, for parties in exit mediations, the terms of the agreement are likely to be more important than restoration of the relationship because the working relationship will cease (Poitras and Le Tareau 2009). This is less likely the case for parties in nonexit mediations: they hope to reconcile during the course of the mediation because they intend to work together in the future (Kressel and Pruitt 1989; Bush and Folger 1994; Poitras 2010; McCorkle and Reese 2015) and are thus socially interdependent (for discussion of social dependence theory, see Deutsch 1949; Johnson and Johnson 1989, 2005; DeOrtentiis et al. 2013).

Because parties need each other to achieve their own goals, this interdependence can produce cooperative behavior that promotes trust and reconciliation (Pruitt, Rubin, and Kim 2003; Johnson and Johnson 2005; Balliet and Van Lange 2013). The mediator role in exit mediations is thus likely to differ from the role in nonexit mediations (Poitras and Le Tareau 2009). We would expect mediators to use more solution‐oriented techniques in exit mediations and place less emphasis on mending the relationship (Coleman et al. 2014; Coleman, Kugler, and Mazzaro 2016). In nonexit mediations, in contrast, we would expect mediators to focus more on reconciliation to support a healthy working relationship (Poitras 2010). In nonexit mediations, the focus on more collaborative behavior may also positively influence how these parties evaluate the mediation. Studies focusing on collaboration in conflict situations found that cooperation and reconciliation lead to general feelings of satisfaction and self‐efficacy among conflict parties (Enright and North 1998; Pruitt, Rubin, and Kim 2003; Maltby, Day, and Barber 2005). This could mean that parties in nonexit mediations engage in more collaborative and reconciliatory behavior, leading to greater satisfaction with the mediation compared to those parties involved in exit mediations. Subordinates in exit mediations, in contrast, might experience a high level of uncertainty about their future, which could affect mediation perceptions in a negative way (Bollen, Ittner, and Euwema 2012), especially if they did not expect a termination of the work contract.

Consequently, our third hypothesis is that, compared to parties in a nonexit mediation, parties in an exit mediation perceive that the mediation was less effective in the long term.

In this study, we derived empirical data from real mediation cases involving hierarchical labor conflicts in the Netherlands. We collected data using questionnaires. Participants were supervisors and subordinates involved in a hierarchical workplace conflict, mediated by a professional mediator registered with the Dutch Mediation Federation (MfN) mediator.¹ We sent questionnaires at two different times: first, up to four weeks following the last mediation session, to collect general information about the conflict and the perceived short‐term mediation effectiveness, and a second time one year later to collect data on perceived long‐term mediation effectiveness. We collected data between January 2011 and July 2014.

We approached workplace mediators with the help of the Dutch Mediation Federation. The mediators who agreed to participate in the study recruited disputants from their own mediations. To avoid selection bias, mediators offered all parties in a hierarchical mediation at the time the chance to participate in the study. The mediators provided us with participants’ contact information and we sent a digital questionnaire by e‐mail to the disputants. We sent another survey one year later with questions about the same mediation. To provide incentive for parties to fill out the long‐term questionnaire, we awarded parties with a 10 euro gift certificate if they completed and returned the survey. We only allowed a maximum of five mediations per mediator in our data set to prevent a sample bias. Participation was voluntary and confidential; only we had access to the data and personal information of the participants.

A total of ninety‐six respondents in seventy‐nine mediations completed the first wave questionnaire. Forty‐one respondents also completed the second wave questionnaire (39 percent) including twenty‐five subordinates and sixteen supervisors. The subordinates’ characteristics were as follows: fifteen men and ten women; average age of almost 51 years; eleven in exit mediations, and fourteen in nonexit mediations. The supervisors characteristics’ were as follows: twelve men and four women, average age: almost forty‐one years, nine in exit mediations and seven in nonexit mediations. Only six participants had been disputants in the same mediation with another disputant (three dyads).

Subordinates were relatively highly educated, with thirteen of them (52 percent) having received a higher education without university degree, and five (20 percent) holding a university degree. Six supervisors (37.5 percent) had received a higher education without university degree, and eight supervisors (50 percent) held a university degree.

Forty respondents (97 percent) had a work contract of indefinite duration at the time of the mediation, with only one person, a supervisor, holding a fixed‐term contract. Fifteen subordinates (60 percent) were on sick leave before the mediation started and twelve of them (49 percent) had reported ill for longer than two months. None of the supervisors were absent from work.

In this sample, twenty mediations (49 percent) fit the definition of exit mediation. In these exit mediations, six supervisors (30 percent) and six subordinates (30 percent) intended at the start of the mediation to end the employment contract, while three supervisors (15 percent) and five subordinates (25 percent) initially intended to continue working together. In the nonexit mediations, six supervisors (28 percent) and 13 subordinates (62 percent) had the intention to continue the employment relationship, while only one supervisor (5 percent) and one subordinate (5 percent) initially intended to cease the working relationship. Data showed that the conflicts were perceived as highly escalated with an average escalation level of 3.95 on a five‐point scale. Regarding the mediation outcome, thirty‐two out of forty‐one mediations ended in an agreement (78 percent).These settlement rates are in line with other research indicating that agreement rates for (workplace) mediation vary from 60 percent to 80 percent (McDermott et al. 2002; Poitras and Le Tareau 2009).

In this study, hierarchical position has been operationalized according to who held a position of authority in relation to the other party involved (“What is your position in the conflict?”). Possible answers were “employer,” “employee,” and “other…” “Employee” refers to subordinates and “employer” to supervisors (subordinate = 0, supervisor = 1).

We ascertained whether it was an exit or nonexit mediation by asking the mediators and disputants if the particular mediation ended in the termination of the employment relationship (nonexit = 0, exit = 1).

We relied on the five‐dimensional mediation effectiveness model developed by Jean Poitras and Aurelia Le Tareau (2009). This fifteen‐item scale comprises five subscales of three statements each:

1. reconciliation between the parties (e.g., “I feel like my relationship with the other party has been restored”),

2. satisfaction with mediator (e.g., “The mediator’s intervention was determinant in advancing discussion”),

3. satisfaction with mediation process (e.g., “Mediation was run in a neutral and objective manner”),

4. satisfaction with the mediation outcome (e.g., “I am happy with the solution we came to”), and

5. confidence in the agreement (e.g., “I believe our agreement will be applied”).

Responses for the different items were coded on a five‐point Likert scale (1 = strongly disagree; 5 = strongly agree), with a high score indicating a high level of agreement with the statements. The Cronbach’s alpha (measure for internal consistency) for each of the five subscales was as follows: reconciliation = 0.90, satisfaction with the mediator = 0.86, satisfaction with the process = 0.91, satisfaction with the outcome = 0.97, and confidence in the agreement = 0.78. A Cronbach’s alpha above 0.70 is expected to reflect internal consistency, meaning that the three items we used to measure the subscales of mediation effectiveness were, indeed, reliable in the sense that they measured the specified construct.

With the exception of “confidence in the agreement,” we used the same scales in the second wave questionnaire, but we slightly adapted the statements to reflect the time that had passed: (e.g., “Looking back, the mediator’s intervention was determinant in advancing discussion”). For the second set of surveys, the Cronbach’s alphas were as follows: reconciliation = 0.88, satisfaction with the mediator = 0.93, satisfaction with the process = 0.72, satisfaction with the outcome = 0.95.

Instead of measuring “confidence in the agreement” as we had in the short‐term measure, in the long‐term measure (when it would no longer have been relevant), we used “compliance with the agreement by the other party.” In our study, following the work of Dean Pruitt and his colleagues (1993), we assessed compliance with the agreement by the other party with the statement, “The other party complied with the mediation agreement.” These responses were also coded on a five‐point Likert scale (1 = strongly disagree; 5 = strongly agree) because it was the perception of compliance that we measured.

In this study, we define an exit mediation as one that ended in the termination of the employment relationship. What parties intended to be the outcome at the start of the mediation with regard to continuing the employment relationship, however, may have differed from what the outcome actually was, and this might have had an impact on how they perceived mediation effectiveness. For example, based on expectancy theory (Vroom 1964; Miner 2005; Kanfer and Chen 2016), parties who want an exit at the start of the mediation might evaluate an exit mediation outcome more positively than parties whose intention it was to continue working together. For this reason, we controlled for this by asking participants what their intention was regarding their employment relationship at the start of the mediation (we stop working together = 0 versus we continue working together = 1). In addition, we also used gender (male = 0, female = 1), the objective mediation outcome (no agreement = 0, agreement = 1), escalation level of the initial conflict (on a five‐point scale), and age (in years) (Bollen, Euwema, and Müller 2010; Bollen and Euwema 2014) as control variables in the analyses.

We managed and analyzed the data using the Statistical Program for the Social Sciences (SPSS) 24.0. To test our hypotheses, we conducted a series of hierarchical regression analyses. We ran these analyses separately to gauge the impact of each subscale of short‐term mediation effectiveness (reconciliation, satisfaction with mediator, process, and outcome, as well as trust in the agreement) on its long‐term equivalents. In the first step, we entered the control variables (age, gender, intention employment, conflict escalation, and agreement). In the second step, we entered hierarchical position and (non)exit mediation to examine their effect on long‐term mediation effectiveness. In the third step, we entered the relevant subscale of short‐term mediation effectiveness.

To control for the risk of multicollinearity (Cohen et al. 2003), we tested the variance inflation factor (VIF) and tolerance of all predictors. The VIF of the predictors varied between 1.01 and 2.41; the tolerance of the predictors varied between 0.41 and 0.98. Both were within the acceptable range for multicollinearity risk (Coakes 2005).

The mean scores, standard deviations, and intercorrelations among control and research variables are listed in Table One. We conducted five three‐stage hierarchical regression analyses for the five different subscales measuring short‐ and long‐term mediation effectiveness. Results of the first hierarchical regression analysis, on reconciliation, show no effect from any of the control variables, the hierarchical position, nor the exit or nonexit nature of the mediation. The only important significant predictor of long‐term reconciliation was short‐term reconciliation. The beta value (β) was 0.63 and significant (β = 0.63, p ≤ 0.001). Together the independent variables accounted for 48.6 percent of the variance in long‐term reconciliation with an adjusted R squared (adjusted R²) of 0.486.

The second hierarchical regression analysis, on satisfaction with the mediator, shows that only short‐term satisfaction with the mediator was a significant predictor of long‐term satisfaction with the mediator (β = 0.38, p ≤ 0.05). There was no effect from the control variables, hierarchical position, or (non)exit mediation. All the independent variables together accounted for 39.8 percent of the variance in long‐term satisfaction with the mediator (adjusted R² = 0.398).

The third hierarchical regression analysis, on satisfaction with the mediation process, shows that none of the control variables, nor the independent variables (hierarchical position, exit or not, short‐term satisfaction with the mediation process) affected satisfaction with the mediation process in the longer term. Altogether, in stage three of the hierarchical regression analysis the independent variables only accounted for 11.5 percent of the variance in long‐term satisfaction with the mediation process (adjusted R² = 0.115).

The fourth hierarchical regression analysis, on satisfaction with mediation outcome, shows that neither any of the control variables, hierarchical position, nor exit or status affected satisfaction with the mediation outcome in the longer term. We did find an effect, however, for short‐term satisfaction with the mediation outcome on long‐term satisfaction with the mediation outcome (β = 0.80, p ≤ 0.001). All the independent variables together explain 43.6% of the variance in long‐term satisfaction with mediation outcome (adjusted R² = 0.436).

The final hierarchical regression analysis, on perceptions of compliance with the agreement by the other party, shows that neither the control variables nor being in an exit mediation or not were significant predictors of perceptions of compliance with the agreement by the other party. Hierarchical position (β = 0.65, p ≤ 0.001), however, as well as short‐term trust in the mediation agreement (β = 0.51, p ≤ 0.001) were significant predictors of parties’ perceptions that the other party was in compliance with the agreement. Together, the independent variables accounted for 46.6% of the variance in long‐term compliance with the agreement by the other party (adjusted R² = 0.466).

Hypothesis One stated that perceptions of short‐term mediation effectiveness are positively related to perceptions of long‐term mediation effectiveness. The results of the different hierarchical regression analyses show, as expected, significant and positive main effects of short‐term reconciliation (β = 0.63, p ≤ 0.001), short‐term satisfaction with the mediator (β = 0.38, p ≤ 0.05), short‐term satisfaction with the mediation outcome (β = 0.80, p ≤ 0.001), and short‐term trust in the mediation agreement (β = 0.51, p ≤ 0.001) on their long‐term equivalents. There was no significant main effect of short‐term satisfaction with the mediation process on long‐term satisfaction with the mediation process. This means that our findings generally support Hypothesis One.

Hypothesis Two stated that supervisors are likely to view mediation as more effective in the long term than are subordinates. Results show that hierarchical position is a significant unique predictor of perceptions of compliance with the agreement by the other party one year after the mediation (β = 0.65, p ≤ 0.001), but not of the other long‐term mediation effectiveness subscales: reconciliation, satisfaction with the mediator, satisfaction with the mediation process, and satisfaction with the mediation outcome. Hypothesis Two is, therefore, only partly supported.

Hypothesis Three stated that compared with parties in a nonexit mediation, parties in an exit mediation would perceive less long‐term mediation effectiveness. Our findings show no difference in perceptions of long‐term mediation effectiveness according to whether the mediation was for an exit or not. Therefore, Hypothesis Three was not supported.

The goal of this study was to offer more insight into the long‐term effects of mediation in hierarchical workplace conflicts, with special attention to the role of hierarchical position and being involved in an exit mediation or not.

Previous research on workplace mediation focused mainly on short‐term outcomes and therefore ignored how participants in workplace mediation perceive mediation effectiveness in the long term. Also, it is surprising that, although some workplace mediations are exit mediations, this type of mediation was never the topic of earlier studies. With this study, we hoped to fill in this research gap and offer new insights on mediation theory and practice.

We have tested our hypotheses using real‐life mediation cases that involved hierarchical labor conflicts. We ran hierarchical regression analyses to see if perceptions of mediation effectiveness immediately after the mediation predicted perceptions of long‐term mediation effectiveness, while differentiating between perceived reconciliation; satisfaction with the mediator, process, and outcome; and compliance with the agreement by the other party.

First and foremost, our results show that mediation is considered effective in the long run by both supervisors and subordinates in mediation. Generally, short‐term measures of mediation effectiveness such as reconciliation and satisfaction with the mediator and outcome, as well as trust in the agreement, predicted how mediation clients would perceive long‐term mediation effectiveness. We found one exception, however: perceived satisfaction with the mediation process immediately after the mediation did not predict how satisfied mediation clients would feel with the process in the long term. Two possible explanations for this are that, in the long run, tangible results such as the mediation outcome were more salient and memorable than the specific mediation process, and/or participants may simply have forgotten the details of the process while the outcomes and their feelings about the mediator remained more vivid.

With regard to hierarchical position, both supervisors and subordinates reported satisfaction with the mediation process, the mediator, and the outcome in the long term. For both supervisors and subordinates, reconciliation was the lowest rated subscale of long‐term mediation effectiveness. This supports earlier research on (workplace) mediation that found reconciliation between parties was not a common result (Kressel and Pruitt 1989; Poitras and Le Tareau 2009; Poitras 2010).

In addition, hierarchical position affected how parties perceived compliance with the agreement over the longer term: supervisors experienced substantially more compliance with the agreement by the other party than did subordinates. This may reflect the fact that subordinates feel more constrained in their behavior when compared to supervisors (Keltner, Gruenfeld, and Anderson 2003; Galinsky, Rus, and Lammers 2011; Galinsky, Rucker, and Magee 2016) and might be less willing or capable of speaking out to make sure that the supervisor behaves according to the mediation agreement than the other way around. It is also possible that supervisors evaluate compliance more positively because the psychological experience of power increases a general sense of optimism (Anderson and Galinsky 2006; Galinsky, Rus, and Lammers 2011). We did not find that hierarchical position had a significant impact on the other subscales of mediation effectiveness.

We also found that whether or not the mediation involved terminating or continuing the employment was not a predictor of perception of long‐term mediation effectiveness. Parties in both exit mediations and nonexit mediations reported long‐term satisfaction with the mediator, process, and outcome, and that they believed the other party had complied with the agreement. Compared to these outcomes, parties felt to a lesser extent reconciled with the other. This is not what we expected because social interdependence theory (Deutsch 1949; Johnson and Johnson 1989, 2005; DeOrtentiis et al. 2013) predicts that parties in a nonexit mediation, who are more interdependent because of a future relationship, will engage in more collaborative behavior that stimulates trust and reconciliation (Pruitt, Rubin, and Kim 2003). A possible explanation for our result might be that, although in exit mediations the focus is on terminating the relationship and not on reconciliation, parties might still perceive the mediation as effective because it enabled them to leave on good terms or perhaps perceived the mediation as effective simply because they were so relieved that the conflict had come to an end (Moore 2014). It is also possible that social dependence is less relevant because collaboration is less necessary in an exit mediation because mediators adopt a strategy in exit mediations that is less dependent on disputants (Coleman et al. 2014; Coleman, Kugler, and Mazzaro 2016).

This study supports the effectiveness of workplace mediation even in hierarchical conflicts. Not only do parties consider mediation effective in the short term, but also in the longer term, thereby supporting the durability of mediation. Furthermore, this study underscores the important effect of hierarchical position on compliance with the agreement.

Our findings suggest that mediators should consider, when mediating hierarchical conflicts, that supervisors and subordinates may have different perceptions of agreement compliance. These findings suggest that mediators should help parties work on implementation and monitoring of the agreement as part of the mediation process (Moore 2014). Because it is more difficult for subordinates to address supervisors when they feel the agreement is not respected than the other way around (Keltner, Gruenfeld, and Anderson 2003), mediators should pay special attention to these differences and even make disputants aware of these power dynamics when discussing the agreement.

The third implication of our study is that exit mediations have value and mediators should not avoid them. We found, counterintuitively, no significant difference in levels of reconciliation between exit and nonexit mediations. Consequently, reconciliation was just as possible for disputants in both exit and nonexit mediations. Our results show that parties in exit mediation consider this type of mediation just as effective as parties in nonexit mediations.

This study has several limitations. First, the sample size of forty‐one participants is small. This is a common limitation for real‐life mediation research: in this case, as in others, it was difficult to find mediators willing to ask parties to participate because of their concerns about confidentiality. And, as is often the case, long‐term participation was difficult to achieve: more than half of the participants “dropped out” between the first and second questionnaire, which could have caused a selection bias (Hogan, Roy, and Korkontzelou 2004; Deng et al. 2013).

In addition, participation in the study was voluntary and may thus have attracted greater interest from those who had positive experiences. This could partially account for the high agreement rate in our data set (78 percent). Although a 78 percent agreement rate is within the range reported by other studies of workplace mediation (Mareschal 2005; Wood and Leon 2005; Poitras and Le Tareau 2009), it is on the high side of that range.

Further, in our hypothesis development we made the assumption that mediators in different types of mediations use different mediation techniques. For example, we surmised that mediators probably use more solution‐oriented techniques in exit mediations and put less emphasis on mending the relationship (Coleman et al. 2014; Coleman, Kugler, and Mazzaro 2016). We did not test for this, however. Future research could examine whether different mediation techniques are used in these two different types of mediation, and, if they are different, whether it makes any difference.

Finally, we did not examine the long‐term impacts of the mediation on feelings of well‐being or improved functioning. Especially in the case of hierarchical workplace conflicts, it would be interesting to see if mediation leads to improved well‐being in the long term, especially for subordinates as they are more affected by hierarchical conflicts in a negative way compared to supervisors (Giebels and Janssen 2005; Dijkstra 2006). Because unemployment has negative impacts on health and well‐being (McKee‐Ryan et al. 2005; Griep et al. 2016), it would also be interesting to find out whether workers who participated in exit mediation found a new job and whether what they learned via mediation helped them in their next job.

In this study of long‐term workplace mediation effectiveness, we found that disputants’ perceptions of reconciliation and satisfaction with the mediator and the mediation outcome, as well as the level of trust in the mediation agreement as reported immediately after the mediation, were good predictors for similar results in the longer term. Unexpectedly, we found few differences between supervisors and subordinates, with the exception that subordinates perceived less compliance with the agreement on the part of the supervisor. We were also surprised to find no differences between exit and nonexit mediations. In general, parties found mediation effective in the long term regardless of their hierarchical position or whether the employment relationship was destined to continue or to cease.