Over the last decades, scholars have increasingly focused on workplace bullying (Leymann 1996; Notelaers et al. 2006). Workplace bullying has been described as a particular type of counterproductive work behavior (Fox and Spector 2005). It reflects a situation in which negative acts at work—concerning work related (e.g., withholding information) and personal issues (e.g., gossiping)—accumulate to a pattern of systematic maltreatment. The negative acts are persistent in nature and last for at least 6 months (Notelaers et al. 2006). Workplace bullying has been linked to a range of detrimental outcomes for targets, observers, and the organization as a whole (Lutgen-Sandvik et al. 2007; Rodriguez-Muñoz et al. 2009; Trepanier et al. 2013). It has traditionally been studied from the target’s perspective (i.e., exposure to workplace bullying behaviors).

Theorizing and empirical studies in the workplace bullying research domain indicated a relationship between bullying and conflicts. Building on 800 bullying incidents, Leymann (1996) elaborated that most bullying cases originated from escalated conflicts, which he referred to as ‘critical incidents.’ Quantitative studies supported this idea and showed that (a high number of) conflict incidents correlated with being a target of workplace bullying (Ayoko et al. 2003; Baillien and De Witte 2009). In fact, conflicts were reported as one of the strongest predictors of exposure to bullying (Hauge et al. 2007; Jennifer et al. 2003; Zapf 1999). These findings correspond with the idea of ‘dispute related bullying’ that refers to being a target of bullying as a result of quarrels and highly escalated workplace conflicts (Einarsen 1999). In conclusion, evidence revealed a relationship between (the prevalence of) conflicts and bullying.

Despite their contribution to the field, these studies conceptualized conflict rather broadly as an interpersonal disagreement. They did not yet dig into the specific focus of the conflict that may evolve into bullying. Moreover, most research studied bullying from a target’s perspective, leaving voids as to how conflicts associate with bullying enactment by perpetrators (Samnani and Singh 2012).

In reply to these voids, the current study aims to (a) particularly tap into the nature of the conflicts and explore how work-related disputes relate to a highly counterproductive situation such as workplace bullying, (b) introduce conflict behavior in this relationship, and (c) investigate being a target as well as being a perpetrator of bullying. This aim follows recent developments in the field, in which scholars shifted toward the perpetrator’s perspective (e.g., Baillien et al. 2011; Balducci et al. 2012; Escartín et al. 2012). It moreover aligns with the Three Way Model (Baillien et al. 2009), in which conflicts are one of the main antecedents of being a target as well as being a perpetrator of workplace bullying. Overall, the current study combines insights from the workplace bullying research domain with knowledge developed within the conflict literature. Our hypotheses are tested in a large representative sample of the Flemish working population.

Task Conflicts Versus Relationship Conflicts

In this study, we want to shed light on how work-related conflicts may associate with workplace bullying. Albeit empirical studies in the field of workplace bullying did not dig into the exact focus of the conflict, scholars did cite Glasl’s model (1994) as a framework that could enhance understanding as to how conflicts transform into deviant behavior such as workplace bullying (Baillien and De Witte 2009; Zapf and Gross 2001). The model describes two important stages of conflict that may transform into a final stage of workplace bullying. As such, this overall framework is inspiring for our current research question.

Specifically, the first stage concerns the ‘rational conflict.’ Drawing on the earlier idea of ‘cognitive conflict’ (Pondy 1967), this stage focuses on ‘what is the problem?’ and is cognitive in nature. It aligns with—the more recent concept of—‘task conflict’ (Jehn 1995) that refers to a factual disagreement about how particular aspects of the tasks are to be accomplished (Amason 1996; Simons and Peterson 2000). The first stage may transform into the second stage of Glasl’s model, the ‘emotional conflict.’ Drawing on the broader idea of ‘affective conflict’ (Pondy 1967), this stage focuses on ‘who is the problem?’ and is affective in nature. It ties in with ‘relationship conflict’ (Jehn 1995; Simons and Peterson 2000) that is particularly focused on the individuals who are part of the conflict (Amason 1996). Relationship conflict involves perceived tension, annoyance, and animosity about personal differences such as values, attitudes, preferences, and personality (De Dreu and Van Vianen 2001). The second stage may subsequently trigger the third and final stages of Glasl’s model. This stage includes manifest reactions from the employee (i.e., manifest conflict; Pondy 1967) that may be reflected in a pattern of deviant behavior and workplace bullying (Zapf and Gross 2001). In conclusion, we may distinguish two conflict types that could relate to workplace bullying (stage 3), rational or task conflict (stage 1), and emotional or relationship conflict (stage 2). In this study, we advocate that the presence of task conflicts relate to workplace bullying through relationship conflicts.

Firstly, task conflicts may be contaminated with relationship conflict, based on empirical observations and arguments. Empirical studies revealed that relationships between task conflicts and negative outcomes such as deceased organizational citizenship behavior and counterproductive work behavior depend on relationship conflicts: Whereas relationship conflicts consistently associated with negative outcomes (see De Wit et al. 2012), task conflicts only related to negative outcomes when these conflicts substantially correlated with relationship conflicts (Mooney et al. 2007; Simons and Peterson 2000). Conflict scholars explained these findings by stating that task conflicts are less closely associated with negative emotions than relationship conflicts (Jehn et al. 2008). Task conflicts tend to carry fewer personal connotations (Greer and Jehn 2007). Consequently, scholars argued that—not always detrimental—task conflicts associate with negative outcomes when they trigger or transform into—clearly negative—relationship conflicts (Mooney et al. 2007; De Wit et al. 2012). That is, in view of our current study, task conflicts may be indirectly related to workplace bullying through their transformation into relationship conflicts. Following arguments further strengthen our position: First, the parties in conflict may perceive the task-related disagreement as evidence of personal rejection owing to a process of emotional misattribution (Simons and Peterson 2000; Torrance 1957). Second, based on behavioral arguments, task-related disagreements often contain emotionally harsh language that can be taken personally (Pelled 1996; Ross 1989; Simons and Peterson 2000). Third, task conflicts may create tensions that have an influence on the interpersonal relations, possibly even creating attributions with regard to the cause of task conflict, which puts even further weight on interdependent relations. Finally, research has pointed out that during conflict, people not only do not always explicitly stick to task-related arguments but also refer to personal aspects. This means that task conflicts contain elements that touch upon relationship conflict (Peterson and Behfar 2003). In conclusion, task conflicts are positively associated with relationship conflicts.

Secondly, relationship conflicts may escalate into workplace bullying in line with the following arguments. Relationship conflicts contain strong negative emotions and increase tensions and strain (Dijkstra et al. 2005). Such conflict could therefore wear out the employee’s resources, and the employee may become an ‘easy target’ offering little resistance against workplace bullying (Baillien et al. 2011). The tensions and strain following from relationship conflicts could also transform into being a perpetrator of bullying through a process of frustration–aggression (Baillien et al. 2009; Berkowitz 1989). From a perpetrator’s perspective specifically, relationship conflicts are strongly intertwined with the self-concept and may therefore cause blame, which leads to grievances. People may address such grievances through bullying toward a co-worker in an attempt to protect the self and redress injustice (Bies and Tripp 1996). In conclusion, relationship conflicts are positively associated with being a target as well as being a perpetrator of bullying.

Following this reasoning, in which task conflicts associate positively with relationship conflicts, and relationship conflicts associate positively with (being a target and being a perpetrator) of bullying, we hypothesize:

H1

There is a positive indirect association between task conflict and being a target of workplace bullying through relationship conflict.

H2

There is a positive indirect association between task conflict and being a perpetrator of workplace bullying through relationship conflict.

Conflict behavior

A further distinction between being either a target or a perpetrator could be established building on the Three Way Model (Baillien et al. 2009). Targets can be differentiated from perpetrators based on the employee’s conflict behavior which reflects the amount of power he/she claims in the conflict situation. Specifically, doing little effort to defend oneself (i.e.. taking a less powerful, submissive stance) will increase the probability of being a target of bullying. Not giving in and aiming to win the fight (i.e., taking a powerful, combatting stance) will particularly lead to being a perpetrator of bullying (Baillien et al. 2009).

These reactions refer to distributive—win/loose—conflict behavior which is defined as conflict management styles that either minimize the outcomes for the employee in conflict in favor of others (yielding) or maximize the outcomes for the employee in conflict at the expenses of others (forcing)(Baillien et al. 2014; Van de Vliert 1997). Yielding is characterized by giving into others and ignoring one’s own interests (Van de Vliert 1997). Yielding has been described as less assertive and powerful conflict behavior that aims to soothe the other party (Yang and Mossholder 2004). This conflict behavior relates to targets of bullying, as targets typically indicate perceiving a power imbalance in which they cannot defend themselves (Leymann 1996; Notelaers et al. 2006). This means that yielding may strengthen the employee’s position as an easy target. That is, tensions and strain that accompany relationship conflicts would particularly associate with the employee offering little resistance against workplace bullying as a target if the employee responds through yielding. We therefore hypothesize:

H3

The (positive) association between relationship conflict and being a target of workplace bullying is boosted by yielding.

Forcing, in contrast, contains the need to prevail at the expense of others. It is defined as assertive and competitive conflict behavior characterized by exerting pressure, intimidating and irritating others (Yang and Mossholder 2004). It reflects a powerful reaction from the employee (Aquino et al. 2006; Fitness, 2000). This conflict behavior aligns with the perpetrator’s perspective. This is because, first, bullying behavior is typically enacted by employees with high formal or informal power who perceive a high effect/danger ratio: Being a perpetrator of bullying depends on the perception that the results of negative behavior toward others (i.e., effect)—as may be expressed through competitive conflict behavior—exceed the risks of being penalized for these acts (i.e., danger) (Keashly and Neuman 2010). Second, social interaction theory of aggression (Tedeschi and Felson 1994) states that people use bullying behavior as a matter of saving face or regaining power. Through bullying, one can affect the reputation of those who initiate or are part of the conflict situation. By doing so, bullying harms the social standing of others in favor of securing one’s reputation (Archer and Coyne 2005). From this, we expect:

H4

The (positive) association between relationship conflict and being a perpetrator of workplace bullying is boosted by forcing.

Following our arguments and assumptions regarding the mediating role of relationship conflict and the moderating role of conflict behavior, we moreover hypothesize a moderated mediation model (Preacher et al. 2007) (Fig. 1):

Fig. 1
figure 1

Model of the research hypotheses

Full size image

H5

The indirect effect of task conflict on being a target of workplace bullying through relationship conflict will be stronger for high levels as compared to low levels of yielding.

H6

The indirect effect of task conflict on being a perpetrator of workplace bullying through relationship conflict will be stronger for high levels as compared to low levels of forcing.

Method

Sample

Data collection was organized in October 2011 via a market research company, specialized in web-based market research surveys, which distributed an anonymous on-line questionnaire among their panel. Panel members were recruited via a mix of channels, including visual and print media, and off-line and on-line recruitment. Participants were motivated based on lottery-based incentives. Using advanced stratified sampling (Singh 2003), aiming for a representative sample of the Flemish working population in view of gender and age, only those panel members were selected who match the target population. Of the 14.000 panel members who were invited 14 % (N = 2,029) provided full answers.

The final sample contained 60 % white-collar workers, 29 % blue-collar workers, and 12 % respondents held a managerial position. About 55 % followed higher education. About 54 % were male, which aligns with the Flemish working population in 2011 (http://statbel.fgov.be). As in the Flemish working population, the majority of the respondents were between 25- and 49-year old (69.1 %); 5.7 % were between 18- and 24-year old; and 25.2 % were between 50- and 64-year old. The representativeness of the sample in terms of age and gender is a particular advantage of this study, as these demographic variables have been shown to relate to workplace bullying. Specifically, women are more likely to become targets (Björkvist et al. 1994), while men are mostly perceived as perpetrators (Einarsen and Skogstad 1996). Research also established age differences in workplace bullying (Einarsen and Raknes 1997; Notelaers et al. 2011).

Measures

The questionnaire measured all core variables using well-established and internationally validated scales.

Task conflict and relationship conflict were assessed using the items of Jehn (1995). The respondent indicated on a five-point Likert scale (1 = never; 5 = almost always) how frequently he/she experienced conflicts with his/her colleagues. Task conflict (α = 0.89) was tapped using 3 items such as ‘My colleagues and I disagree about opinions regarding the work being done.’ Relationship conflict (α = 0.92) was measured by means of three items such as ‘There are personality conflicts evident between my colleagues and I.’

The Dutch Test for Conflict Handling (DUTCH; Van de Vliert 1997; De Dreu et al. 2001) was used to investigate the employee’s distributive conflict behavior. Response categories ranged from ‘never’ (=1) to ‘almost always’ (=5). Forcing (α = 0.79) was measured by means of 4 items such as ‘I aim at winning the conflict.’ Yielding (α = 0.82) consisted of 4 items such as ‘I adapt to the other party’s goals and interests.’

Being a target of bullying (α = 0.87) was measured by means of the 9 items Short Negative Acts Questionnaire (S-NAQ; Notelaers and Einarsen 2008). The items examine how often the respondent was confronted with a list of bullying acts during the last six months (e.g., ‘being withheld information’). The response categories varied from ‘never’ (=1) to ‘now and then’ (=2), ‘monthly’ (=3), ‘weekly’ (=4), and ‘daily’ (=5). In line with the bullying literature, all items were included in one scale (for a discussion, see Einarsen et al. 2009; Nielsen et al. 2009). Being a perpetrator of bullying (α = 0.75) was measured by means of the same nine items of the S-NAQ, however, using an active formulation (see Baillien et al. 2011; Escartín et al. 2012). The respondent rated how frequently during the last 6 months (1 = never; 5 = daily), he/she had engaged in the bullying acts (e.g., ‘withholding information’).

Item-level confirmatory factor analyses (CFA; Lisrel 8.72) distinguishing six factors (i.e., task conflict, personal conflict, forcing, yielding, being a target of bullying, and being a perpetrator of bullying) revealed a good fit, χ2(390) = 2,180.48, p < 0.001; RMSEA = 0.05, CFI = 0.97, SRMR = 0.04, and NNFI = 0.96. This six-factor model showed a better fit to the data compared to the common method factor model, χ2(405) = 9,157.14, p < 0.001; RMSEA = 0.10, CFI = 0.91, SRMR = 0.13, and NNFI = 0.90. It also fitted better compared to alternative models in which concepts that might be related were combined into one factor (see Table 1).

Table 1 More-detailed results of the confirmatory factor analysis establishing the measurement model
Full size table

Analyses

The Hayes (2013) macro was used to assess both the interaction effects (Hypotheses 3 and 4) and the (conditional) indirect effects (Hypotheses 1, 2, 5 and 6). Following Aiken and West (1991), we mean centered the independent variables for testing the interaction effects. Additionally, all hypotheses were tested using heteroscedasticity-consistent standard errors (HC3) in order to control for a possible violation of homoscedasticity (Hayes and Cai 2007) which is particularly important in view of a large sample size. All analyses controlled for gender, dummy coded as ‘female’ (0 = male; 1 = female), age (in years), and supervisory position (0 = no; 1 = yes). In addition to gender and age, which have been identified as important personal characteristics related to bullying, supervisory position was chosen as supervisors were identified as the perpetrator in a majority of bullying cases (Nielsen, 2013).

Results

The scales’ means, standard deviations, and correlations are reported in Table 2. There were positive correlations for task conflict, relationship conflict, and distributive conflict behavior with both being a target and being a perpetrator of bullying. Task and relationship conflicts were also positively correlated (r = 0.48, p < 0.001). Notably, being a target of workplace bullying was significantly related to being a perpetrator of workplace bullying (r = 0.44, p < 0.001).

Table 2 Descriptive statistics and correlations
Full size table

Hypotheses

Hypotheses 1 and 2 were tested by calculating bootstrapping confidence intervals using 5,000 replications (Preacher and Hayes 2008; Shrout and Bolger 2002). First, our results revealed an indirect effect of task conflict on being a target of workplace bullying through relationship conflict (b = 0.14, boot SE = 0.01, 95 % CI [0.12:0.16]) supporting Hypothesis 1. Additionally, the results also showed a significant direct effect between task conflict and being a target of workplace bullying (b = 0.10, SE = 0.02, p < 0.001). The R2 of the total effect model was 0.12. Second, our results revealed an indirect effect of task conflict on being a perpetrator of workplace bullying through relationship conflict (b = 0.04, boot SE = 0.01, 95 % CI [0.03:0.06]) supporting Hypothesis 2. They also indicated a direct link between task conflict and being a perpetrator of workplace bullying (b = 0.05, SE = 0.01, p < 0.001). The R2 of the total effect model was 0.07.

In view of Hypothesis 3, regression analyses showed a main effect of both relationship conflict (b = 0.27, SE = 0.02, p < 0.001) and yielding (b = 0.06, SE = 0.02, p < 0.005) and a significant interaction effect (b = 0.07, SE = 0.03, p < 0.005) on being a target of workplace bullying. The R2 of this model was 0.24; the interaction term accounted for a ∆R 2 of 0.01 (p < 0.001).Footnote 1 Simple effects at low (i.e., 1 SD below the mean) and high (i.e., 1 SD above the mean) levels of yielding revealed that the positive association between relationship conflict and being a target of workplace bullying is stronger for employees scoring high (b = 0.31, SE = 0.02, p < 0.001) rather than low (b = 0.22, SE = 0.03, p < 0.001) on yielding. This confirmed Hypothesis 3. Regarding Hypothesis 4, regression analyses revealed a main effect of both relationship conflict (b = 0.09, SE = 0.01, p < 0.001) and forcing (b = 0.06, SE = 0.01, p < 0.001) and a significant interaction effect (b = 0.04, SE = 0.01, p < 0.01) on being a perpetrator of workplace bullying. The R2 of this model was 0.12; the interaction term accounted for a ∆R 2 of 0.01 (p < 0.001). Simple effects revealed that the positive association between relationship conflict and being a perpetrator of workplace bullying is stronger for high (b = 0.11, SE = 0.02, p < 0.001) as compared to low (b = 0.06, SE = 0.01, p < 0.001) forcing. Hypothesis 4 was thus confirmed.

Finally, we tested Hypotheses 5 and 6 regarding moderated mediation following the method described by Hayes (2013). This method uses a bootstrapping sampling method and calculates bias-corrected confidence intervals at different values of the moderating variable in order to explore conditional indirect effects. First, we tested whether the indirect effect of task conflict on being a target of workplace bullying through relationship conflict was moderated by yielding. We found that the indirect effect of task conflict on being target of workplace bullying (through relationship conflict) was stronger at high compared to low levels of yielding (see Table 3). Hypothesis 5 was supported. Similarly, we tested whether the indirect effect of task conflict on being a perpetrator of workplace bullying through relationship conflict depended on the levels of forcing. Our results indicated that the indirect effect of task conflict on being a perpetrator of workplace bullying (through relationship conflict) was stronger at high as compared to low levels of forcing (see Table 4). Hypothesis 6 was also supported.

Table 3 Analyses for the conditional indirect effect (through relationship conflict) between task conflict and being a target of workplace bullying at different values of yielding
Full size table
Table 4 Analyses for the conditional indirect effect (through relationship conflict) between task conflict and being a perpetrator of workplace bullying at different values of forcing
Full size table

Post-hoc Analyses

Given our results regarding Hypotheses 1 and 5—revealing both a direct effect between task conflict and being a target of workplace bullying and a (conditional) indirect effect through relationship conflict—we additionally analyzed whether yielding also moderated the direct effect between task conflict and being a target of workplace bullying. Interestingly, this additional test (see Table 5) not only revealed a non-significant task conflict × yielding interaction but also a non-significant relationship conflict × yielding interaction. Similarly, given the findings regarding Hypotheses 2 and 6, we explored whether forcing also moderated the direct effect between task conflict and being a perpetrator of workplace bullying. Our additional test (see Table 6) indicated that—in contrast to targets—this is indeed the case. That is, both the indirect effect and the direct effect of task conflict on being a perpetrator of workplace bullying are stronger at high levels of forcing compared to low levels of forcing.

Table 5 Analyses for the conditional indirect effect (through relationship conflict) and direct effect between task conflict and being a target of workplace bullying at different values of yielding
Full size table
Table 6 Analyses for the conditional indirect effect (through relationship conflict) and direct effect between task conflict and being a perpetrator of workplace bullying at different values of forcing
Full size table

Discussion

The current study holds a number of contributions to the literature to date. First, we tested how work-related disagreements associate with workplace bullying. As such, we add to the existing literature in which scholars advanced that workplace bullying may grow from conflicts (e.g., Baillien and De Witte 2009; Hauge et al. 2007). Specifically, our study sheds light on the focus of the conflicts that relate to bullying by drawing on Glasl’s model (Glasl 1994; Pondy 1967) and arguments from the more recent conflict literature (Jehn 1995; Simons and Peterson 2000). Most importantly and in line with our expectations, our results revealed that task conflicts are indirectly related to workplace bullying trough relationship conflicts.

Second, this indirect effect was confirmed for being a target as well as being a perpetrator of bullying. By focusing on both targets and perpetrators, we align with recent evolutions in the workplace bullying research domain (Baillien et al. 2011; Balducci et al. 2012; Escartín et al. 2012). We not only advanced knowledge on the factors contributing to exposure to bullying but also regarding aspects contributing to the enactment of bullying behaviors.

Third, while the indirect mechanism did not differentiate between targets and perpetrators, we moreover tapped into what could particularly mold a conflict situation into being either a target or a perpetrator of bullying. Inspired by the Three Way Model (Baillien et al. 2009) and the broader bullying and conflict literature, this was established by introducing distributive conflict behavior—forcing for perpetrators and yielding for targets—as a boosting moderator of the association between relationship conflict and workplace bullying. Our results showed that the indirect effect between task conflict and being a perpetrator of bullying through relationship conflict was stronger for high levels of forcing. Similarly, the indirect effect between task conflict and being a target of bullying through relationship conflict was stronger for high levels of yielding. In sum, our findings indicate that both the specific focus of the conflict and conflict behavior are to be combined when predicting workplace bullying from the two perspectives. This attests to the plea in the literature to combine various research perspectives for a throughout understanding of how exactly—through what mechanisms—conflicts relate to (experience versus enactment of) workplace bullying (see Zapf and Einarsen 2011). Our findings seem to indicate that the model of Glasl (1994) and the Three Way Model (Baillien et al. 2009) together may shed further light on the relationship between conflicts and bullying. Conflict behavior defined in line with the Three Way Model (Baillien et al. 2009)—as powerful versus submissive reactions to the conflict—may particularly fit between stage 2 (i.e., emotional conflict) and stage 3 (i.e., deviant behavior and bullying) of the Glasl (1994) model. Additionally, combining the presence of conflicts at work (a work environmental factor) with how employees deal with conflicts (an individual factor) as predictors of workplace bullying also particularly adds to the bullying literature. Indeed, research regarding bullying antecedents has predominantly developed in two separate directions: studies tapping into work environmental aspects (e.g., Hauge et al. 2007) versus studies investigating the impact of individual factors (e.g., Glaso et al. 2009). To further insight into reasons to bully or become bullied, scholars in the field have therefore called for a multi-causal view that combines work environmental and individual factors when investigating antecedents of bullying (Zapf and Einarsen 2011); an issue we addressed in our current study.

Fourth, besides the workplace bullying research domain, the current study also contributes to the conflict literature. To date, the conflict field has developed considerable understanding of how task and relationship conflict affect organizations and teams in terms of productivity, performance, commitment, and turnover intentions (De Dreu and Weingart 2003; De Wit et al. 2012; Tidd et al. 2004). Less attention has, however, been paid to destructive behavior such as workplace bullying. We further this line of research by adding workplace bullying—both from the perspective of perpetrators and targets—as a valuable outcome of task and relationship conflicts.

Regarding our results, we first of all detected an indirect relationship between task conflicts and workplace bullying through relationship conflicts (Hypotheses 1 and 2), in line with Glasl (1994). This indirect process corresponds with suggestions that task conflicts are more likely to be associated with negative outcomes when they show a rather strong positive correlation with relationship conflict (De Wit et al. 2012). Such a correlation would be needed as task conflicts—compared to relationship conflicts—are less closely associated with negative emotions (Jehn et al. 2008), tend to carry fewer personal connotations (Greer and Jehn 2007), and may therefore be less strongly related to negative outcomes (Bayazit and Mannix 2003; Gamero et al. 2008). Scholars thus argued that task conflicts relate to negative outcomes because the tensions that arise from these conflicts encourage negative interpersonal relations leading to relationship conflicts (De Wit et al. 2012; Simons and Peterson 2000; Tidd et al. 2004; Yang and Mossholder 2004). In other words, task conflicts may translate into negative outcomes when they convert into relationship conflicts.

Notably, our results still revealed a significant direct effect between task conflict and being a target/perpetrator of workplace bullying. Consequently, although the association between task conflicts and bullying is indirect through relationship conflicts, task conflicts could also relate directly to bullying without transforming into relationship conflicts. A possible explanation could be that, as relationship conflicts, task conflicts may yield strain too, which might particular be so for high levels of task conflicts (Giebels and Janssen 2005). Consequently, task conflicts could also wear out the employee’s resources, making the employee an ‘easy target’ for workplace bullying (Baillien et al. 2011). Such reasoning would be in line with the Conservation of Resources Theory (COR; Hobfoll 1989): individuals who possess little resources may be more prone to demanding circumstances and have less tools to overcome this problematic situation. Alternatively, task conflicts may initiate a process of frustration–aggression in which the employee becomes a perpetrator of bullying (Baillien et al. 2009; Berkowitz 1989). This also aligns with COR Theory in the sense that employees experience stress when (they feel that) their resources are threatened. They will subsequently strive to protect or regain their resources. Becoming a bully may then be a successful strategy to regain resources such as status and power: the creation of a resource spiral for oneself at the expense of others (Wheeler et al. 2010).

Another interesting finding concerns the moderating impact of distributive conflict behavior (a) on the association between relationship conflict and bullying (Hypotheses 3 and 4) and (b) on the indirect association between task conflict and bullying (Hypotheses 5 and 6). Regarding perpetrators, post hoc analyses showed that—while the conditional indirect effect is present—also the direct effect between task conflict and being a perpetrator of workplace bullying was stronger for high levels of forcing. This underlines the importance of forcing for perpetrators of bullying, both in view of the conflict escalation process in which task conflicts are linked to bullying through relationship conflicts as well as in view of the direct association between task conflicts and bullying. Regarding targets, the post hoc analyses could not reveal a simultaneous effect of yielding on the indirect and direct association between task conflicts and bullying. From a statistical perspective, this could indicate that the interaction between relationship conflict and yielding does not uniquely contribute to being a target of workplace bullying when one accounts for the interaction between task conflict and yielding (e.g., substantially shared amount of variance). One explanation could be that yielding has an ‘overall’ impact on being a target on workplace bullying and cannot be assigned to specific direct or indirect conflict escalation processes. Or, targets of workplace bullying are generally ‘powerless’—and, as put forward in the bullying literature, perhaps even ‘passive’ (Baillien et al. 2009; Salin and Hoel 2011)—when confronted with a range of demanding and problematic situations at work. Nevertheless, besides its role as an overall predictor of bullying, distributive conflict behavior particularly distinguishes between the perpetrator and target’s perspective. We therefore argued that employees may become either a perpetrator or a target based on how they react to the conflict. Table 3, however, shows a positive correlation between being a perpetrator and a target of bullying (r = 0.44), indicating that both perspectives may be related. This corresponds with debates in the literature in view of ‘provocative/aggressive targets’: targets of workplace bullying may engage in bullying because of a retaliation process or in an attempt to shift the perpetrator’s attention to a ‘new’ scapegoat. Alternatively, perpetrators may see themselves as targets or may become targets through retaliation by others (Rodriguez-Muñoz et al. 2012). Additional analyses in which we tested Hypotheses 1, 3, and 5 controlling for being a target of workplace bullying and Hypotheses 2, 4, and 6 controlling for being a perpetrator of workplace bullying did, however, not significantly alter our results.

Following our theoretical arguments, we focused on distributive—win/loose—conflict behavior as the moderator that distinguished between targets and perpetrators. That is, we elaborated and investigated the boosting effects of yielding for targets and forcing for perpetrators.Footnote 2 Future studies could advance our insights regarding the role of conflict behavior by, for example, investigating hypotheses regarding the role of integrative conflict behavior (Van de Vliert 1997). Similarly, drawing on our theoretical and empirical arguments, our analyses were linear in nature assuming more/less associates with more/less hypotheses. Future studies could build on our findings by, for example, investigating curvilinear associations between task conflicts, relationship conflicts, and bullying and assessing which model—linear versus non-linear—fits best; an aspect that has recently been studied in view of psychosocial safety climate and bullying (Escartín et al. 2013).

Limitations and future research

As any other study, the current study shows some drawbacks that could be improved in future research. First, our data rely on self-reports and might be subjected to common method bias. We are, however, confident that our results were not significantly affected by common method bias, as we followed instructions that may prevent this bias. We, for example, underlined that there were no right or wrong answers and guaranteed confidentiality (Podsakoff et al. 2003). Confirmatory factor analyses moreover revealed these precautions were effective. Additionally, given that method bias deflates rather than inflates interaction effects (Podsakoff et al. 2012), we are confident that common method bias may not have significantly altered our findings regarding the interaction effects. Nevertheless, in reply to this concern, future studies could account for common method bias by retesting our hypotheses using multi-source data, for example, provided by colleagues and the employees’ contact person familiar with the definition of workplace bullying. Note, however, that from a bullying perspective, collecting third party scores on workplace bullying raises practical concerns. Given the subtle nature of most bullying acts, third parties may be unable to perceive a situation as bullying. Moreover, assessing third party scores on workplace bullying without trying to counteract such behavior raises ethical concerns, as workplace bullying yields many negative consequences for the parties involved as well as for the work unit and the organization (Einarsen et al. 2009).

On a related matter, our results could perhaps be influenced by social desirability, particularly in view of workplace bullying. Our findings may therefore relate to a specific subgroup of respondents willing to admit exposure to or enactment of workplace bullying. Social desirability, however, generally causes a lack of variance that typically leads to underestimation of statistical effects (Spector, 2006). As such, the relationships found in the current study may become even stronger when taking into account social desirability.

Second, our study relies on cross-sectional data, and any conclusion regarding causality should be handled with care. That is, even though we build our hypotheses based on theoretical arguments indicating that conflict may translate into bullying, we cannot entirely be sure that bullying is a consequence—and not an antecedent—of conflicts. A recent longitudinal study exploring different types of causal relationships between conflicts, conflict management styles, and bullying may, however, shed more light on this aspect (Baillien et al. 2014): cross-lagged analyses revealed that the presence of conflicts at work and conflict management styles are antecedents rather than consequences of workplace bullying. Based on our current theorizing and findings, future studies could, however, investigate this causal matter in more detail. An interesting aspect could then, for example, be to investigate whether daily exposure to task conflict, relationship conflicts, and the way the employee responded to these conflicts escalates into being a target or a perpetrator of bullying over time.

Notably, a particular strength of this study is that all hypotheses were tested in a representative sample of Flemish employees. Moreover, we used three important control variables in view of workplace bullying: gender, age, and supervisory position (e.g., Björkvist et al. 1994; Notelaers et al. 2011). Omitting these control variables from the analyses did, however, not alter our conclusions, suggesting that our findings are fairly robust. However, future studies may want to further test the generalizability of our results, also in view of cultural differences. For example, being part of Belgium, power distance in Flanders is somewhat higher than average, compared to other countries where power distance is low (e.g., Norway) or high (e.g., China). Future work could test whether similar effects may arise for the conflict management styles in other countries.

Third, owing to the nature of our data—in which respondents were not sampled within organizations but rather within the general working population—we could unfortunately not investigate who was having a conflict with whom and which employees were involved in bullying incidents as perpetrators or targets. Such information could, however, be valuable, as both conflicts and workplace bullying refer to social phenomenon in which two or more employees are in interaction. Future research could therefore benefit from assessing dyadic or team data that may, for example, investigate whether the employees involved in the conflict escalation are also the ones involved in the bullying incident and aim to differentiate their position as perpetrators versus targets based on their reactions to the conflict. Another fruitful avenue in this respect could be to take into account team cohesion and power aspects within the team in view of conflicts and conflict escalation: both team cohesion and power have been suggested, but not yet specifically investigated, as valuable factors for bullying (Keashly and Nowell 2011). Recent findings did, however, reveal that a high individual identification with the workgroup as well as a higher average level of social identification within the workgroup related to less workplace bullying (Escartín et al. 2013). These findings underline the role of contextual effects on a group—and thus broader social—level in which strong average levels of identification may create a positive atmosphere against bullying.

Finally, the current study aimed to strengthen knowledge regarding conflicts and workplace bullying by empirically integrating insights from both research domains. By doing so, we aligned with the current state in these research domains in view of validated measurements and analyses. In a recent debate regarding concepts and measurements in the field of unethical organizational behavior, some scholars, however, indicated that this usual approach might not entirely capture our constructs’ true meaning (e.g., Hershcovis, 2011). Adhering to these remarks, we additionally explored whether our results could be replicated for respondents belonging to the ‘very top’ groups of being a target and being a perpetrator of workplace bullying. We defined these groups based on Mikkelsen and Einarsen’s (2001) 2-act criterion: targets (n = 322) are confronted weekly with at least 2 acts from the S-NAQ (Notelaers and Einarsen 2008); perpetrators (n = 88) enacted at least 2 acts on a weekly base.Footnote 3 The analysis among these subgroups replicated our resultsFootnote 4 yet could not re-confirm the indirect effect through relationship conflicts for perpetrators: an observation that could be owed to theoretical (e.g., from a certain point, other factors besides conflict escalation may become more important, such as personality) and statistical reasons (e.g., decrease in statistical power). Future studies could therefore dig more into this aspect by, for example, retesting our hypothesis in groups of victims/perpetrators versus a comparable control group of employees not involved in bullying incidents.

A related conceptual debate in this respect could concern the difference between conflicts, conflict behavior, and bullying. Besides many theoretical arguments in the literature (Ayoko et al. 2003; Baillien et al. 2014; Keashly and Nowell 2011; Samnani 2013), the current study further underlines the conceptual difference between these constructs based on empirical findings. Several models tested in a CFA revealed a better fit for our hypothesized model as compared to alternative models that, for example, grouped the conflict and bullying items into one factor (see Table 1). These findings align with studies in which (personal) conflicts were empirically distinguished—as antecedents—from abusive supervision as another related construct (Tepper et al. 2011) and from workplace bullying (Baillien et al. 2014) particularly.

Implications for practice

Our findings contain valuable suggestions for organizations and practitioners that wish to prevent workplace bullying. First, they could be specifically aware of situations in which task-related, cognitive conflicts convert into personal conflicts as this escalation may further trigger workplace bullying. Second, they should also take care of task conflicts per se as these conflicts may directly associate with incidents of bullying. Third, policy workers and practitioners may particularly invest in introducing ‘healthy’ conflict behavior within the organization and amongst employees. Our results showed that forcing and yielding both boosted the association between task conflicts, relationship conflicts and bullying. Organizations could therefore avoid these conflict management styles. One additional option could then be to encourage, for example, problem solving conflict behavior: a form that has been negatively linked with workplace bullying both for targets (e.g., Ayoko et al. 2003) and perpetrators (e.g., Baillien et al. 2014). They could do so by providing employee training or by encouraging supervisors to set the right example by assisting the employees in a problem-solving way in case of task or relationship conflicts. Note, however, that our results revealed that the majority of explained variance in bullying is owed to the presence of task and relationship conflicts at work, which underlines that investing in conflict behavior is a valuable route to prevent bullying yet not as successful as investing in a climate of constructive conflicts that counteracts negative and destructive outcomes.

Conclusion

The current study investigated how conflicts relate to workplace bullying. We found an indirect positive association between task conflicts and being a perpetrator or a target of bullying through relationship conflicts. Our results moreover showed that the association between relationship conflicts and bullying is boosted by distributive conflict behavior, being forcing for perpetrators and yielding for targets. Even more so, distributive conflict behavior also moderated the indirect task conflicts—bullying association. Additionally, we also found a direct effect from task conflicts to bullying, indicating that the presence of task conflicts—without having any personal connotation—may encourage bullying as well. This relationship was also moderated by forcing for perpetrators and by yielding for targets.