What Moderates the Relation Between Intragroup Conflict, Emotional Exhaustion, and Work Engagement?

Introduction To survive and thrive in an increasingly complex work environment, organizations often rely on workgroups. Workgroups are intended to provide an adaptive response to today’s work environment. Many workgroups, however, become hotbeds of conflict (Jehn and Mannix, 2001). Group members bring different values, norms, expertise, and attitudes to job-related and interpersonal issues. These differences can ignite intragroup conflict defined as real or perceived incompatibilities among group members (Jehn, 1995). Two broad types of intragroup conflict have been identified: relationship conflict and task conflict (Jehn, 1995). These conflict types are distinct, both in their nature and in their effects on workgroup outcomes (e.g., Jehn and Mannix, 2001). Even so, both types of intragroup conflict, especially relationship conflict, have detrimental effects on workgroup outcomes (De Dreu and Weingart, 2003; De Wit, Greer, and Jehn, 2012). Intragroup conflict is associated with a variety of psychological, behavioral, and physical symptoms (Bruk-Lee, Nixon and Spector, 2013; Meier et al., 2014) that threaten employees’ well-being (Dijkstra et al., 2005; Sonnentag, Unger, and Nägel, 2013). Yet, research investigating the distinct effects of relationship conflict and task conflict on the different aspects of well-being and moderators of this relationship is sparse. The purpose of the present study was to explore the relationships between intragroup conflict at the group level and employee well-being, and to examine how emotion regulation and conflict management might moderate this relationship. Employee well-being was measured via work engagement and emotional exhaustion, both of which were considered as collective (group-level) phenomena. We consider work engagement and emotional exhaustion as two distinct indicators of well-being at work (see Schaufeli et al., 2002). We assumed that studying what might increase positive aspects (work engagement) and reduce negative aspects (emotional exhaustion) of wellbeing would provide a robust representation of well-being in workgroups. Most previous research on work engagement and emotional exhaustion has examined these constructs at the individual level, treating group outcomes as statistical aggregations. In contrast, we examined and measured the constructs at the workgroup level, asking participants to assess work engagement and emotional exhaustion of their workgroup rather than of themselves. From a theoretical perspective, the psychological well-being manifested by one person can influence well-being of other people (Westman, 2001). This implies that work engagement and emotional exhaustion of one or more workgroup member can affect the work engagement and emotional exhaustion of the whole workgroup (see Bakker, Emmerik, and Euwema, 2006; Costa, Passos, and Bakker, 2014). An individual level of constructs such as work engagement ORIGINAL ARTICLE


Introduction
To survive and thrive in an increasingly complex work environment, organizations often rely on workgroups. Workgroups are intended to provide an adaptive response to today's work environment. Many workgroups, however, become hotbeds of conflict (Jehn and Mannix, 2001). Group members bring different values, norms, expertise, and attitudes to job-related and interpersonal issues. These differences can ignite intragroup conflict defined as real or perceived incompatibilities among group members (Jehn, 1995). Two broad types of intragroup conflict have been identified: relationship conflict and task conflict (Jehn, 1995). These conflict types are distinct, both in their nature and in their effects on workgroup outcomes (e.g., Jehn and Mannix, 2001). Even so, both types of intragroup conflict, especially relationship conflict, have detrimental effects on workgroup outcomes (De Dreu and Weingart, 2003;De Wit, Greer, and Jehn, 2012). Intragroup conflict is associated with a variety of psychological, behavioral, and physical symptoms (Bruk-Lee, Nixon and Spector, 2013;Meier et al., 2014) that threaten employees' well-being (Dijkstra et al., 2005;Sonnentag, Unger, and Nägel, 2013). Yet, research investigating the distinct effects of relationship conflict and task conflict on the different aspects of well-being and moderators of this relationship is sparse.
The purpose of the present study was to explore the relationships between intragroup conflict at the group level and employee well-being, and to examine how emotion regulation and conflict management might moderate this relationship. Employee well-being was measured via work engagement and emotional exhaustion, both of which were considered as collective (group-level) phenomena.
We consider work engagement and emotional exhaustion as two distinct indicators of well-being at work (see Schaufeli et al., 2002). We assumed that studying what might increase positive aspects (work engagement) and reduce negative aspects (emotional exhaustion) of wellbeing would provide a robust representation of well-being in workgroups.
Most previous research on work engagement and emotional exhaustion has examined these constructs at the individual level, treating group outcomes as statistical aggregations. In contrast, we examined and measured the constructs at the workgroup level, asking participants to assess work engagement and emotional exhaustion of their workgroup rather than of themselves. From a theoretical perspective, the psychological well-being manifested by one person can influence well-being of other people (Westman, 2001). This implies that work engagement and emotional exhaustion of one or more workgroup member can affect the work engagement and emotional exhaustion of the whole workgroup (see Bakker, Emmerik, and Euwema, 2006;Costa, Passos, and Bakker, 2014). An individual level of constructs such as work engagement

What Moderates the Relation Between Intragroup
Conflict, Emotional Exhaustion, and Work Engagement? and emotional exhaustion, might often be insufficient to address the interactive nature of workgroups. We thus attempted to examine what might be done to facilitate group work engagement and to prevent group emotional exhaustion. Bakker et al. (2006) define group emotional exhaustion as a shared feeling of depleted energy among the members of a workgroup. Consistent with our assumption, Costa et al. (2014: 6) defined group work engagement '… as a shared, positive, fulfilling, motivational emergent state of work-related well-being'. Group work engagement is shaped by the nature of members' interactions during workgroup activities.
Research suggests that the influence of conflict on workgroup outcomes is contingent upon workgroup members' reaction to conflict and how they are managed (e.g., De Dreu and Weingart, 2003;De Wit et al., 2012). We considered two means of managing conflict in the workgroups. The first was emotion regulation referring to people's efforts to impact their activated emotions during emotionally events (Gross, 1998) -a mostly emotionfocused strategy. The second one was conflict management, referring to strategies implemented by group members to reduce conflict (DeChurch and Marks, 2001) -a mostly cognitive-focused strategy. Accordingly, we measured the moderating influence of emotion regulation and conflict management as personal skills (individual level) on two indicators of employees' well-being in workgroups.
This effort extends the literature on intragroup conflict and provides insight into the use of emotion regulation and conflict management to mitigate the negative effect of intragroup conflict.

Relationship Conflict and Group Emotional Exhaustion
Relationship conflict, also known as affective conflict or social-emotional conflict occurs when differences in personalities, values, and norms create interpersonal emotional tension, anger, and animosity in a group (Jehn, 1995). Relationship conflict tends '…to be more interpersonal and emotional' (De Dreu and Weingart, 2003: 747) and is accompanied by stress, tension, and frustration (Jehn, 1995). Relationship conflict is associated with negative affective responses and a greater risk to one's well-being (Sonnentag et al., 2013).
There is consensus on the adverse effects of relationship conflict on group outcomes (e.g., De Wit et al., 2012;Shaukat, Yousaf, and Sanders, 2017). The main supportive reason is that the mere experience of discord and divergence of interests, norms, and values elicits negative emotions and threatens oneself (Jehn et al., 2008). This is consistent with self-verification theory (Swann et al., 2004), which posits that whenever an individual's selfviews and social views are not confirmed by colleagues, it is perceived as a threat to oneself. Lack of confirmation jeopardizes the self-verification process (Jehn et al., 2008) and increases group member anxiety and hostility (e.g., Dijkstra et al., 2005).
From a work stress perspective, relationship conflict is a threatening work stressor (Spector and Bruk-Lee, 2008) as it generates tension and aggravation, with psychological costs increasing strain and emotional exhaustion (Dijkstra et al., 2009). When relationship conflict persists or intensifies, it results in a decline in physical and psychic functioning, which could lead to burnout (De Drue and Beersma, 2005). Relationship conflict is related to symptoms of increased strain De Dreu, Dierendonck, and Dijkstra, 2004;De Dreu, 2008), increased psychological stress, and decreased well-being (Dijkstra et al., 2011;Meier et al., 2014;Sonnentag et al., 2013). Relationship conflict is also associated with emotional exhaustion, a core element of burnout (De Dreu and Beersma, 2005;Dijkstra et al., 2009).
Emotional exhaustion is defined as depletion of one's physical and emotional resources (Maslach and Jackson, 1981) leading to emotionally charged interactions in workgroups. These emotional interactions evoke strong emotions (e.g., Dijkstra et al., 2009) and can then lead to emotional exhaustion. Shaukat et al. (2017) reported a significant, positive correlation between relationship conflict and emotional exhaustion. Furthermore, research illustrates that there is a tendency to automatically copy and synchronize one's emotional states with those of another member (e.g., Hatfield et al., 1994). This suggests that one workgroup member's emotional exhaustion could spread to the whole group (Bakker et al., 2006). Recent studies have found that frequency of exposure to such emotional interactions at workplace predicts group emotional exhaustion (Benitez, Francisco, and Lourdes, 2018). We thus argue that emotionally charged interactions and conflict-induced negative emotions elicited by relationship conflict generate additional emotional demands. When such demands exceed group members' resources, group members feel chronic fatigue, exhaustion, and depleted of energy (e.g., Dijkstra et al., 2009), leading to emotional exhaustion. When these feelings are shared amongst group members, this could escalate negative emotions (Jehn and Mannix, 2001) and spread emotional exhaustion in workgroup. Thus, we proposed hypothesis H1: H1: Workgroup relationship conflict will be positively related to group emotional exhaustion.

Task Conflict and Group Work Engagement
Task conflict, also known as cognitive conflict, centers on disagreements about task views; the content of the task being performed, the distribution of resources, procedures, or guidelines and the interpretation of facts (De Dreu and Weingart, 2003).
Findings about the association between task conflict and workgroup outcomes are less conclusive than those of relationship conflict and group outcomes. Some studies report that task conflict is beneficial for workgroup performance (e.g., Jehn and Mannix, 2001) and for making better decisions (e.g., Behfar et al., 2011). Other studies indicate that task conflict impairs work group performance, and group member satisfaction (De Dreu and Weingart, 2003). Jungst and Blumberg (2016) demonstrated that task conflict is negatively associated with work engagement.
Recent studies have examined the circumstances influencing positive or negative outcomes. For example, De Wit et al. (2012) have shown that task conflict is beneficial when task conflict and relationship conflict are weakly correlated or when the conflict arises among top management groups rather than groups at lower levels of the organizational hierarchy. Nevertheless, some studies suggest that employees who engage in task conflict tend to experience negative emotions and to be less satisfied with their group, colleagues, and work (De Dreu and Weingart, 2003).
A normal reaction to any type of conflict is frustration and dissatisfaction irrespective of outcomes (Ross, 1989). According to Jehn (1997), such negative emotions can be present with any type of conflict, including task conflict. Still, research shows that positive emotions positively influence the intensity of work motivation or the amount of effort contributed, (e.g., Staw, Sutton, and Pelled, 1994). For example, Staw et al.'s findings (1994) indicate that positive emotions are associated with work achievement. Seo, Barret, and Bartunek (2004) theorized that unpleasant affective experiences produce negative motivational states. Jungst and Blumberg (2016) suggested that task conflict probably demotivates individuals, due to increased cognitive load as a cost in the process of task conflict. Their findings suggest that when employees experience or perceive task conflict, they are less engaged with their work. Schaufeli et al. (2002) defined work engagement as a positive, fulfilling, work-related state of mind, characterized by vigor (high energy), dedication (inspiration and pride in one's work), and absorption (full concentration on one's work). As task conflict intensifies, arousal and tension increases (e.g., Carnevale and Probst, 1998). This places additional burdens on group members and increases their cognitive load, often draining most of the group members' cognitive resources. As a result of increased cognitive load, group members' cognitive resources become diverted towards the task conflict. Increased cognitive load also reduces the ability to process information and this often leading to freezing the process (Carnevale and Probst, 1998).
According to threat rigidity theory (Staw et al., 1981), conflict can produce rigid thinking, which restricts judgment and reduces ability to consider alternative perspectives (Carnevale and Probst, 1998). Drawing on threat rigidity theory, task conflict may pose a threat to self and increases arousal, anxiety, and frustration (Carnevale and Probst, 1998). These feelings could be shared amongst group members. Consequently, when group members feel threatened and fail to process information from others, they may be less likely to question their initial views and be more likely to hold onto their initial opinion rigidly. Restriction in information processing may reduce attention to their fellow workgroup members' views while increasing reliance on their initial views and prior expectation. As a result, workgroup members might freeze up, withdraw, or confine their perceptual field of input -all characteristics of work disengagement. To adapt, group members often behave by withdrawing from task involvement and withholding effort (Jehn, 1995). When disengaging from workgroup activities is shared amongst the workgroup members, it leads to disengaging from work. Thus, we deduced a second hypothesis: H2: Workgroup task conflict will be negatively related to group work engagement.

The Moderating Effect of Emotion Regulation on Group Emotional Exhaustion
Research suggests that conflict, especially relationship conflict, is emotionally laden (Jehn and Bendersky, 2001) and is closely linked to conflict management (Nair, 2008). Handling emotions also moderates the effects of relationship conflict on group outcomes (Griffith, Connelly, and Theil, 2014) and it also has important consequences for well-being (Gross and John, 2003). According to the model by Gross (1998), emotion regulation (ER) is an efficient means of managing negative emotions. ER is defined as '…the processes [strategies] by which individuals influence which emotions they have, when they have them, and how they experience and express these emotions' (Gross, 1998: 275). Gross (1998) discusses one means of emotion regulation, called antecedent-focused strategies, which can occur before the emotional responses have been completely developed. Antecedent-focused strategies promotes psychological well-being (John and Gross, 2004).
Cognitive reappraisal and distraction (attentional depletion) are two types of antecedent-focused strategies. Both involve taking action prior to or during an emotional response. These two strategies, in particular distraction, are the most likely and cognitively healthy emotion regulation strategies in the context of organizations (Griffith et al., 2014;Gross and John, 2003), yet no study has directly measured or compared the effects of these strategies on emotional exhaustion in the face of workgroup conflict.
Cognitive reappraisal involves changing the meaning attributed to a disagreement and thus the emotional reaction associated with it (Gross, 2008). Cognitive reappraisal is an effective strategy for moderating the experiential and behavioral consequences of emotion (Gross, 2008). It occurs before the emotion response tendencies have been fully generated. That is, cognitive reappraisal can efficiently alter the entire subsequent emotion trajectory. The usage of cognitive reappraisal allows one to exhibit interpersonal behavior that is appropriately focused on social interaction and is perceived by the others as emotionally engaging and responsive (John and Gross, 2004). Research shows cognitive reappraisal improves interpersonal functioning and well-being (John and Gross, 2004).
Distraction involves a mental turning away from the emotion generating event. When using an emotional strategy of distraction, group members redirect their thoughts to distract themselves from an emotional response towards a less emotional one. Examples include thinking of something pleasant and invoking thoughts that are inconsistent with the undesirable emotional state. By shifting their attention elsewhere, group members use distraction to regulate their emotions. Bushman (2002) has shown that distraction reduces anger and other negative emotions. Griffith et al. (2014) showed that emotion regulation, in particular distraction rather than cognitive reappraisal, has an important role in moderating the negative consequences associated with relationships conflict.
Considering emotion regulation as a personal resource, we expect that workgroup members who are able to regulate their emotions would manage the negative effects of emotions caused by relationship conflict and thus avoid becoming emotionally exhausted. This expectation was derived from the Triple-Match Principle (TMP, de Jonge and Dormann, 2006) proposing that stressors, moderators, and the outcomes should be qualitatively similar to produce a buffering/moderator effect. We assumed there is a match between relationship conflict, emotion regulation, and emotional exhaustion, because emotions play a critical role in each of them. We thus propose that emotion responses to relationship conflict can be managed by using emotion regulation strategy. This led to the following hypotheses: H3a: The association between relationship conflict and group emotional exhaustion will be moderated by emotion regulation -such that the positive association between relationship conflict and group emotional exhaustion is weaker when group members' use of distraction is higher. H3b: The association between relationship conflict and group emotional exhaustion will be moderated by emotion regulation -such that the positive association between relationship conflict and group emotional exhaustion is weaker when group members' use of cognitive reappraisal is higher.

The Moderating Effect of Conflict Management on Group Work Engagement
The effect of task conflict is not just a function of disagreement, but also of how group members manage disagreements (e.g., Friedman et al., 2000). Conflict management could be an important moderator of the association between task conflict and work engagement. It refers to the strategies implement by group members to reduce or solve conflict (DeChurch and Marks, 2001;Jehn and Mannix, 2001).
Much research on conflict management is built upon Pruitt and Rubin's (1986) dual concern theory. It posits that reactions to disagreement derives from two motivational underpinnings: concern for the self and concern for the other party. We focused on two strategies for conflict management in workgroups: cooperative and competitive. Based on dual concern theory these two strategies are in accordance with the two main motivational underpinnings. Individuals may communicate cooperative or competitive intentions to other parties in the conflict (Tjosvold, 1998). Furthermore, cooperative and competitive conflict management strategies are both active, conflict-engaging strategies and have a high concern for self (e.g., Dijkstra et al., 2009).
A cooperative style of conflict management is high on concern for both self and others. It is characterized by exchanging information on priorities and preferences, problem solving, and making trade-offs between important and unimportant issues. Cooperation is considered to be the most effective conflict management strategy for reducing both conflict and stress (Friedman et al., 2000).
In contrast, a competitive conflict management style is high on concern for self and low on concern for others. It is characterized by threats, bluffs, punishments, and intimidation (e.g., De Dreu and Beersma, 2005). Compared to the cooperative style, the competitive style frequently leads to conflict escalation and negative outcomes (e.g., Alper, Tjosvold, and Law, 2000). Research indicates a competitive style increases conflict and contributes negatively to group functioning and group effectiveness (De Dreu and Van Vianen, 2001).
Drawing on the dual concern theory (Pruitt and Rubin, 1986) group members who handle conflict cooperatively are more likely to engage in constructive communication. They encourage full exchange of task-related opinions with their fellow group members, leading to an improved understanding of the key perspectives and issues and an improved understanding of the standpoint of other members (Tjosvold, 1998). We argue that engaging with task conflict constructively via a cooperative style leads to unfreezing of the processing of task-related information and an opened mindset that produces less rigid cognitive structures (Carnevale and Probst, 1989). As a result, a cooperative style mitigates the adverse effects of task conflict on work engagement.
In contrast, when group members adopt a competitive style of conflict resolution, the result is usually fractious debate, escalating threats and deadlock. Group members who wish to outdo one another rarely compromise and instead block each other's efforts (e.g., Tjosvold et al., 2003). Managing conflict competitively tends to be associated with withholding information, win-lose interactions, and closed-mindedness (e.g., Chen, Liu, and Tjosvold, 2005). Engaging with task disagreements competitively leads to freezing of the process of taskrelated information and a closed mindset that produces rigid cognitive structures (Carnevale and Probst, 1989). Holding onto self-centred work views and prioritizing personal gain of all advantages impairs constructive interactions and intensifies disagreements and stress (Friedman et al., 2000), leading workgroup members to withhold their task effort (Jehn, 1995).
Based upon the TMP, we argue there should be a match between task conflict, conflict management, and work engagement. Task conflict is initially a cognitive disagreement based on facts, evidence, interpretations, or applications. However, it is emotion laden as well (Jehn, 1995). Conflict management is an approach to conflict incorporating cognitions and emotions associated with conflict and resolutions (e.g., Huang, 2010). Work engagement also includes cognitive and emotional components (Schaufeli et al., 2002). These two components indicate a qualitative match among task conflict, conflict management, and work engagement as defined by the TMP. As Figure 1 illustrates, Esbati and Korunka: What Moderates the Relation Between Intragroup Conflict, Emotional Exhaustion, and Work Engagement?
Art. 5, page 5 of 15 we expect conflict management moderates the adverse effects of task conflict on work engagement, leading us to the following hypotheses.
H4a: The association between task conflict and group work engagement will be moderated by conflict management -such that the negative association between task conflict and group work engagement is weaker when group members' use of a cooperative strategy is higher. H4b: The association between task conflict and group work engagement will be moderated by conflict management -such that the negative association between task conflict and group work engagement is greater when group members' use of a competitive strategy is higher.

Method Participants
We collected data from 106 workgroups drawn from five Iranian organizations. Fifteen of these workgroups had a research questionnaire return rate below 80 percent. These workgroups were excluded from the data analysis. The remaining 91 workgroups (N = 595 participants: 272 women and 323 men; average age = 38 year; SD = 4.87 years) constituted our research sample. The participants were drawn from one industrial organization (34%), two professional research centers (29%) and two health care centers (37%). Seventy-nine percent of the participants had a bachelor's degree or higher. The average length of time workgroup members had been working together was 4.1 years (SD = 1.97, ranging between 1 year and 11 years. The average group size was six members (SD = 1.8).

Procedure
Because the questionnaire items of all but one measure were in English, they were back-translated. The conflict management items were taken from a previously constructed Farsi scale.
Prior to data collection, several steps were taken to address ethical concerns and to ensure participants' commitment to the study. First, we sent managers an email describing the research as a study of intragroup conflict. Consenting managers nominated workgroups in which the members had been intact for at least one year. A research assistant distributed the questionnaires to members of these groups. The managers also wrote a memo to their employees requesting their cooperation. Next, the research assistant visited each selected group, explained the purpose of the study and distributed the questionnaires to the group members. The written instructions accompanying the questionnaires again included a brief explanation of the study's purpose.
Employees were given assurance of confidentiality and were allowed to respond to the survey anonymously during work hours. It was emphasized that the responses would be anonymous and that data would be aggregated for analyses.

Measures
In this study, we measured the independent (relationship and task conflict) and dependent variables (group emotional exhaustion and group work engagement) at the group level. Theoretically, the construct of intragroup conflict resides at the group level. In line with research (e.g., Bakker et al., 2006;Costa et al., 2014;Benitez et al., 2018), and with crossover theory (Westman, 2001) proposing when a stressor or psychological strain experienced by one person affects the level of stress of other persons, we measured emotional exhaustion and work engagement at the group level. The moderators (individual perception of emotion regulation and conflict management styles), as personal skills and resources were measured at the individual level. We assumed that the level of these skills and resources varied among group members.
All independent, moderating, and dependent variables represented by items on the questionnaire were measured on seven-point rating scales e.g., from 1 (strongly disagree) to 7 (strongly agree) and from1 (none) to 7 (a great deal). The following scales were included in the questionnaire.

Intragroup conflict
Jehn's three-item scale (1995) of relationship conflict and three-item scale of task conflict were used. The participants were asked to rate how much conflict they perceive in their workgroup. The sample items: 'How much relationship tension is there in your work group?' and 'How much conflict of ideas is there in your work group?' measure relationship conflict and task conflict respectively.  To assess group work engagement, the Utrecht Work Engagement Scale (17 items, UWES; Schaufeli et al., 2002) was included. However, we modified this scale based on the referent-shift consensus to refer to group members' perceptions of their colleagues' work engagement rather than to themselves (Chan, 1998). As such, the items were transformed from an individual referent to group-level referent that reflected the perceptions of the individual of their colleagues' work engagement. A sample item was: 'in our workgroup, my colleagues find the work that they do full of meaning and purpose'.

Group emotional exhaustion
We employed the emotional exhaustion subscale of Maslach's Burnout Inventory-General Survey (8 items MBI-GS; Schaufeli et al., 1996). As with items in UWES, emotional exhaustion items were reworded, replacing individual references (I, me, my, mine) to group-level references that asked for judgments on all workgroup members. A sample item was: 'My fellow group members are burned out from our workgroup'.

Conflict management style
The questionnaire items used for this assessment were taken from those employed by Ghahremani et al. (2015) in their evaluation of group conflict in Iranian management teams. Ghahremani et al. measured the cooperative (7 items, α = 0.82) and competitive (5 items, α = 0.81) dimensions of conflict management based on Rahim's (1983) organizational conflict inventory form C (ROCI-II). One such item was: 'I try to explore ideas to come up with a decision collectively' and 'I utilize my authority to make decisions in my favor'.

Emotion regulation
We included five cognitive-reappraisal items from Spaapen et al.'s (2014) Revised Emotion Regulation Questionnaire. One of them, for example, was: 'I control my emotions by changing the way I think about the situation I am in'. We also developed three items to assess attention depletion or distraction (α = 0.81). Items included: 'When I am feeling negative emotions, I make sure to distract my mind by thinking about something exciting or challenging', 'When I am faced with a stressful situation, I get my mind off the situation by concentrating on something pleasant', 'When I am feeling negative emotions, I keep myself busy with something else (such as giving a phone call to a friend or surfing on the net)'.

Control variables
Because the workgroup was the unit of analysis in our study, we considered workgroup size and tenure as control variables. These variables have been shown to increase the likelihood of incompatibilities and differences in personal and work views (e.g., Jehn, Northcraft, and Neale, 1999;Lichtenstein et al., 1997), which, in turn, affect outcomes such as work engagement, workgroup creativity, and decision quality (Hu et al., 2016;Parayitam and Dooley, 2004;Selmer, Jonasson, and Lauring, 2013).

Statistical analysis
We examined the structure of each measure used in this study by performing Confirmatory Factor Analysis (CFA, AMOS 24). Chronbach's alphas were calculated for all scales. Given that our study is about the perception of each group member of their workgroup, theoretically our unit of analysis is workgroup. To assess the appropriateness of aggregating individual responses to the workgroup level, we computed within-group agreements and intraclass correlations coefficient (ICC) for all variables except for moderators. To explore the moderator hypotheses, we conducted moderation analysis using SPSS macro (PROCESS; Model 1) developed by Hayes (2012). Finally, significant interactions were plotted, and simple slopes tests were conducted at both low (-1 SD) and high levels (+1 SD) of the moderator variables.

Preliminary analysis
We tested the hypothesized two-factor structure of the intragroup conflict measure, with the first factor representing relationship conflict and the second reflecting task conflict in the total research sample (N = 595). The first CFA with two factors representing relationship conflict and task conflict accounted good for the data [(χ 2 (8, N = 595) = 27.22, p < 0 .001. (GFI = 0.96, CFI = 0.99, TLI = 0.98, RMSEA =0 .06)]. For cognitive reappraisal and distraction types of emotion regulation, we tested the hypothesized two-factor structure of emotion regulation with the first factor representing cognitive reappraisal and the second reflecting distraction. The first CFA did not account well for the data, χ 2 (19, N = 595) = 53.8, p < 0.001, GFI = 0.97, CFI = 0.97, TLI = 0.96, RMSEA = 0.06. However, removing one item (out of eight items) based on standardized residual covariance (>2.58) resulted in a model that provided a much better fit of the data, χ 2 (13, N = 595) = 24.87, p < 0.02, with fit indices largely exceeding 0.97 (CFI = 0.99, TLI = 0.98, GFI = 0.99), and an RMSEA of 0.03. Therefore, the model with seven items was preferred as the model that best fits the data.
For conflict management, we tested the hypothesized two-factor structure of the conflict management measure, with the first and the second factor representing the cooperative and competitive style, respectively. The first CFA with two factors, representing the cooperative and competitive styles accounted good for the data: χ 2 (53, N = 595) = 91.29, p < 0.001, with good fit indices (CFI = 0.98, TLI = 0.97, GFI = 0.98), and an RMSEA of 0.03.
We also tested the hypothesized a single structure of the emotional exhaustion measure. The first CFA did not account well for the data, χ 2 (20, N = 595) = 514.3, p < 0.001; GFI = 0.80. CFI = 0 .84, TLI = 0.77, RMSEA = 0.20. However, removing four items based on standardized residual covariance (>2.58) resulted in a model that provided a much better fit of the data: χ 2 (2, N = 595) = 7.77, p < 0.02, with following fit indices CFI = 1.00, TLI = 0.99, GFI = 0.99, and an RMSEA of 0.06. Furthermore, the contents of these four items showed that they addressed individual action or feelings not shared with the groupmates. For example, 'Workgroup members feel fatigue when they get up in the morning and have to face another day on the job'. In light of the CFA results and inspection of these four items, the model with for remaining items was judged as the model that best fits the data.
For work engagement, we tested the hypothesized three-factor structure of the work engagement measure, representing dedication, vigor, and absorption, respectively. The first CFA with three factors did not fit. After inspecting the modification indexes and allowing the errors of some items to correlate, the model did not fit. Thus, an exploratory principal component analysis with varimax rotation on group work engagement was performed to verify whether the three factors could be retained, and the items loaded on their intended factor. The analysis revealed the three-factor model, explained 62.07 percent of the variance. The factor solution was interpretable. The cut-off point for considering an item for a scale was set at a component loading higher than 0.70.

Data aggregation
As the unit of analysis in this study was the workgroup rather than the individual, we computed r wg values (James, Demaree, and Wolf, 1993) to determine the amount of agreement among the judgments made by the members of each workgroup for all variables, except for moderators measured at the individual level ( Table 1). The r wg values were all above the conventionally acceptable value of 0.70 (James et al., 1993). Furthermore, considerable variance between groups is required to justify aggregation at group level. To do so, intra-class correlation (ICC; Bliese, 2000) was estimated by ICC (1). As indicated by James (1982), if the resulting ICC (1) is between the conventional cutoffs of 0.05 and 0.25, the aggregation of individual scores at workgroup level can be safely justified. As Table 1 displays all scales, except that for emotional exhaustion, exceeded them slightly. Table 2 displays the descriptive statistics (mean, standard deviations) and correlation matrix.

Main effect
Consistent with H1, Table 3 illustrates the positive and significant effect between relationship conflict and group emotional exhaustion (b = 0.54, p < 0.001). Therefore, this finding supported our Hypothesis 1.
H2 predicates a negative and significant correlation between task conflict and group work engagement. As Table 4 illustrates there is a negative effect between task conflict and work engagement, yet this effect is not significant. Thus, the second hypothesis did not gain support. Although, a simple linear regression analysis supported the negative and significant correlation between task conflict and group work engagement (F = (1, 89) = 10.68, p = 0.002, β = -0.33, p < 0.01) with an R 2 of 0.11.

Moderating effect
H3a predicts that distraction strategy of emotion regulation would moderate the association between relationship conflict and group emotional exhaustion. We conducted a moderation analysis for the distraction as follows.
Controlling for task conflict, workgroup tenure, and size, the interaction of distraction and relationship conflict was significantly negative ( Table 3, b = -0.57, P < 0.001). Thus, hypothesis 3a suggesting the usage of distraction moderates the association between relationship conflict and emotional exhaustion, was supported.
A simple slope analysis was performed to test the significance of the relationship in low and high distraction strategy usage. The results revealed that at low levels of distraction usage (1 SD below the mean distraction, that  is, below 0.62), the association between relationship conflict and group emotional exhaustion is positive and significant (b = 0.90, p < 0.001, CI 95% [0.63; 1.17]). This is also true at average level (b = 0.54, p < 0.001, CI 95% [0.35; 0.73]), but not at high level of distraction strategy usage (1 SD above the mean distraction, that is over 0.62) (b = 0.19, p > 0.05, CI 95% [-0.08; 0.45]). The graphical presentation of the interaction was derived using standard regression coefficient of the regression lines for workgroup high and low (+/-1SD of the mean) on the moderator variable of distraction (Figure 2). The figure suggests that utilizing high distraction in a situation with high relationship conflict is not beneficial. Perhaps this is because the tension and resentment caused by such conflict restrict emotional resource. In a highly emotional situation, the restriction could create a gap between the group members' emotional arousal (e.g., high animosity) and their wishes and efforts to redirect their attention from the relationship conflict to a neutral or pleasant issue. This gap might vitiate using a high distraction strategy in a high relationship conflict situation.
H3b predicts that cognitive reappraisal strategy of emotion regulation would moderate the association between relationship conflict and group emotional exhaustion.
H4a postulates that a cooperative style would moderate the association between task conflict and group work engagement. Controlling for relationship conflict, workgroup tenure, and size indicated that the interaction of a cooperative style and task conflict on group work engagement was significant ( Table 4, b = 0.47, p < 0.05). Thus, hypothesis H4a was confirmed.
A simple slope analysis was performed to test the significance of the relationship in low and high levels of a cooperative style usage. The results revealed that at low level (1 SD below the mean a cooperative style, that is, below 0.34), the association between task conflict and group work engagement was negative and significant (b = -0.35, t = -2.91, p < 0.001, CI 95% [-0.59; -0.11]). For average (b = -0.19, p > 0.05, CI 95% [-0.44; -0.05]) and high levels of using a cooperative style (b = 0.03, p > 0.05, CI 95% [-0.35; 0.29]), the association was also negative yet not significant.
The graphical presentation of the interaction was derived using standard regression coefficient of the regression lines for workgroup high and low (+/-1SD of the mean) on the moderator variable of a cooperative style (Figure 3). The figure shows that utilizing a high level of cooperation in a high task conflict situation is not beneficial. At high level of task conflict, the workgroup members' cognitive and emotional loads and tension increase, thereby drawing resources away from the processing of information associated with critical debates of differing perspectives. This may block an open and constructive discussion and create reluctance and withdrawal, while a cooperative style requires tackling conflict actively and openly.
H4b postulates that a competitive style would moderate the association between task conflict and group work engagement. Controlling for relationship conflict, workgroup tenure, and size revealed that the interaction of competitive style and task conflict on group work engagement was not significant (b = -0.09, p > 0.05, [95%CI: -0.55; 0.36]). Thus, H4b did not gain support.

Discussion
The aim of this study was to examine the effects of the distraction and cognitive reappraisal strategies of emotion regulation on the association between relationship conflict and group emotional exhaustion. In addition, we attempted to examine how cooperative versus competitive styles of conflict management moderate the association between task conflict and group work engagement. Prior to testing the moderator hypotheses, we investigated whether relationship conflict is positively related to group emotional exhaustion and whether task conflict is negatively related to group work engagement.
In our study, relationship conflict was positively related to group emotional exhaustion, which supported our expected association (H1). The result is congruent with previous findings that showed a positive association between relationship conflict and emotional exhaustion at the individual (De Dreu et al., 2004;De Dreu and Beersma, 2005;Dijkstra et al., 2009) and group levels (Benitez et al., 2018). It suggests that relationship conflict, which is emotional in its core (Greer and Jehn, 2007) depletes group members' emotional resources as a whole. One plausible reason is that group members perceive relationship conflict as a threat triggered by colleagues, a threat that drains them of energy. When this negative experience is shared within the workgroup members, it increases the level of group emotional exhaustion. Furthermore, the results revealed that workgroup size is related to relationship conflict, because when workgroups are larger, social/personality clashes, which can engender hostility and tension, are more likely.
Our results also extend previous research by examining the moderating influence of emotion regulation strategies during the occurrence of relationship conflict in workgroups. The results revealed that the distraction strategy is a plausible moderator, supporting H3a. This finding is consistent with previous studies (e.g., Bushman, 2002). For example, Zillman's studies (1988) suggest that evoking negative emotions can be effectively mitigated by engaging in activities that are highly absorbing and entertaining. Griffith et al. (2014) showed that using distraction could mitigate the adverse effects of relationship conflict. Shifting attention away from relationship conflict is a readily available means of mitigating susceptibility to negative emotion (e.g., Eisenberg and Fabes, 1992). It easily prevents the mind from wandering back to the source of negative emotion. Thus, using distraction may serve as a venting when experiencing relationship conflict as Griffith et al. (2014) noted. It allows negative emotions to subside. Distraction may also provide group members with the opportunity for rest and detachment from the negative emotions. As a result, it is likely that the depleted resources are replenished which, in turn, buffer the effects of conflict on emotional exhaustion.
Contrary to our expectation, a cognitive reappraisal strategy did not gain support as a moderator in the association between relationship conflict and emotional exhaustion, not supporting H3b. One plausible reason may be that cognitive reappraisal requires cognitive resources to re-evaluate the situation, which might not be available in the heat of relationship conflict. Furthermore, cognitive reappraisal requires application before or after experiencing an emotion (Sheppes and Meiran, 2007), while distraction could be applied at any point during the experience of an emotion. Due to its simplicity and rapid effect, distraction strategy received support as a moderator.
In our study, task conflict was negatively related to group work engagement, yet this association was non-significant in a robust testing. Therefore, H2 did not gain support. A plausible reason for this might be related to controlling for relationship conflict, as there is a correlation between relationship and task conflicts. In fact, research has shown the average correlation between these two constructs is approximately +0.54 (De Dreu and Weingart, 2003;De Wit et al., 2012).
Our study extends the intragroup conflict literature by examining the moderating influence of cooperative and competitive styles of conflict management on the association between task conflict and group work engagement. The results show that a cooperative style is a moderator, but a competitive style is not. Consistent with previous studies (e.g., DeChurch and Marks, 2001;Lovelace, Shapiro, and Weingart, 2001), this finding indicates the benefit of a cooperative conflict management style over the competitive style in managing task conflict.
One plausible reason is that handling task disagreements cooperatively increases intrapersonal self-efficacy and satisfaction as well as harmony in the workgroup (see De Dreu, 2008). This, in turn, mitigates negative emotions and encourages group members to openly discuss issues. Workgroup members can then become more confident that their opponents are pursuing mutually beneficial solutions. This could conserve their cognitive and emotional resources to buffer the negative effects of task conflict on their work engagement. In contrast, a competitive style of conflict management does not buffer the adverse effects of task conflict. Perhaps when group members view task conflict as a win-lose struggle, negative emotions would escalate and if deadlocks prevail, group members would block each other's efforts (e.g., Tjosvold et al., 2003). Holding rigidly to initial work task views blocks open discussion. This could intensify disagreements and stress (Friedman et al., 2000) leading members to withhold their effort from work due to incompatibilities (Jehn, 1995).
Although we developed general rather than crosscultural predictions, a few comments about cultural patterns are in order as data were collected in Iran. Iranian culture emphasizes values of preserving harmony and saving face in order to maintain long-term relationships. Group members usually have an implicit, indirect, and abstract style of communication (Hofstede, 1983). This style has two common consequences. First, it can restrict open discussion among group members; too much open discussion can violate social norms, leading group members to maintain their initial views. The second consequence of the Iranian communication style is related to distraction as a strategy to manage the negative effects of relationship conflict. Distraction allows group members to maintain their relationships in their group by taking attention away from conflicts. Thus, it could serve as an escape to circumvent the emotional situation faced.
In summary, by focusing on group emotional exhaustion and group work engagement and the distinct effects of relationship and task conflicts in non-western workgroups, the results confirm the significantly negative impact of the relationship conflict on emotional exhaustion. The results suggest that the effects of task conflict might be adverse for group work engagement, as it showed a negative impact, even so the effect was not significant. Furthermore, the results reveal how distraction and a cooperative conflict management style reduce the adverse effects of relationship and task conflicts. The results provided evidence that Western's findings can be generalized and are relevant to a collectivist culture.

Theoretical and practical implications
The first finding of theoretical importance is related to examining the distinct effects of intragroup conflict (relationship and task conflicts) on two independent dimensions of well-being (group emotional exhaustion and group work engagement). Few studies have addressed the possible relations between the two major types of Esbati and Korunka: What Moderates the Relation Between Intragroup Conflict, Emotional Exhaustion, and Work Engagement?
Art. 5, page 11 of 15 intragroup conflict and workgroup well-being. Moliner et al. (2008) stated that a complete understanding of wellbeing at work requires both reducing the negative aspects of well-being and increasing the positive aspects of wellbeing. Our results suggest that researchers should focus more on factors that can buffer the negative consequences of workgroup conflict. The moderating effects of distraction and cooperative approach for dealing with workgroup conflict indicate that the effects of conflict depend on how it is managed.
Our findings also map well with previous findings at the individual level reporting the adverse effects of relationship. Consistent with related research, our findings suggest that future research should address the means of managing conflict at the workgroup level. There might be a broad range of manageable and trainable variables for alleviating the effects of conflict on workgroup outcomes.
The results also have important practical implications. As relationship and task conflicts are inherent to workgroups, group members cannot escape their effects. A major implication is that workgroup creators should prepare group members to manage the dynamics of relationship and task conflicts by providing training that supports the regulation of emotions and a win-win approach to conflict management. These skills can be learned (e.g., Tjosvold et al., 2003), although they might be a part of the group members' personalities and behavioral repertoires. For example, Druskat and Wolff (2001) stated that workgroup training can improve emotional awareness and emotion regulation among workgroup members. Lovelace et al. (2001) also showed that collaborative communications help alleviate conflict before it becomes personalized.
With practice, both emotion regulation and cooperative conflict management could become habitual. We suggest that workgroup leaders train group members to develop more effective emotion regulation skills and teach group members intragroup communication techniques and how to adopt a cooperative approach that helps them frame task and relationship disagreements as challenges to the whole group rather than to particular individuals.

Limitation and future research
Our study was based on self-reports and it is likely that responses were influenced by social desirability or common method variance (Ganster and Schaubroeck, 1991). However, some researchers believe that the magnitude of any such effects is often overstated (Spector, 1994). For example, Evans (1985) states that common method bias is less of a problem when testing interaction effects as we did in our current study. Furthermore, the choice of standardized and validated measurements used in this study (i.e., CFAs) could reduce the possibility of this problem (Spector, 1987).
Another limitation is the cross-sectional nature of our data. Cross-sectional studies do not allow the establish ment of causal relationships between the variables analyzed. Developing longitudinal designs could address this issue. As the data were collected from multiple organizations, specific organizational characteristics may have influenced the measured variables. However, the size and heterogeneity of our sample (N = 91 workgroups that involved 595 participants) is a better alternative to the unique characteristic of data captured from a single organization. Indeed, the generalization of results is improved in that they are based on data from different organizations. One advantage of the heterogeneous respondents is that it helps reduce the effects of individual biasing factors. Nevertheless, researchers are careful about causality and develop longitudinal designs to address this issue.
The study finding may need to be interpreted cautiously. One reason is related to the CFAs results of group emotional exhaustion and group work engagement. In line with Costa et al. (2014) one possible explanation could be related to using group-referent measures when studying group-level constructs. One might assume that aggregating individual-referent items accurately does not represent a collective construct. Although it is plausible, because workgroups have specific dynamics and that promoting well-being in individuals within workgroups may demand a different action. Thus, we suggest that future study needs to investigate whether aggregating individual-referent items accurately represents a collective construct.
The results of this present study suggest that conflict within workgroups is manageable. Thus, future research could address the variables that buffer the negative effects of conflict. Additionally, group members who base their daily interactions on the context of their own workgroup could develop a perception of conflict in the workgroup. The context of the workgroup has a major impact on group members (George and Jones, 1997). Therefore, to provide a better understanding of intragroup conflict researchers should take into account the contexts under which the negative effects of conflict can be diminished, if managed correctly. Of note, due to the nature of the survey study, there is further needed for development of experimental designs to asses which emotion regulation, i.e., distraction or cognitive reappraisal is more effective, as well as to evaluate their usage frequency.