The primary purpose of this study was to investigate the patterns of causal linkage between task conflict and relationship conflict. We comprehensively derived possible conceptual models that specify the relationships between the two types of conflict and drew several research hypotheses on causality. To investigate the causal associations thoroughly, we employed a longitudinal panel design in which both types of conflicts were measured at two points in time and then performed a series of tests.
Results of testing direct causality demonstrated that earlier relationship conflict was significantly linked to later task conflict, but the opposite direction was not. Further investigation on the mediation models showed that relationship conflict triggered task conflict through negative group affect.
Testing the moderating effect of intragroup trust produced a consistent result with previous studies: The effect of task conflict on relationship conflict was significant for groups with a lower level of trust (Peterson & Behfar, 2003; Simons & Peterson, 2000). Earlier task conflict was more likely to cause a subsequent relationship conflict when group members did not trust one another strongly. On the other hand, intragroup trust was not found to moderate the reverse direction.
In sum, our findings suggest that it is somewhat unavoidable that relationship conflict bleeds over into task conflict in organizational settings; Personal animosity naturally triggers disagreements in task-related issues. On the other hand, task conflict may not be transferred into relationship conflict when intragroup trust is high. Furthermore, as seen in Figure 2, the effects of early relationship conflict on later relationship conflict (β = 0.75) and task conflict (β = 0.43) are much stronger than those of early task conflict. Given these findings, therefore, we may argue that relationship conflict should be managed well during the early stage of a group's development; otherwise, its influence to subsequent conflicts can be detrimental. With respect to task conflict, its impact on group outcome depends on group context and group process variables: When it is supported with such process as constructive controversy built on intragroup trust, it is not easily transformed into relationship conflict.
Our findings may provide a plausible explanation for why the majority of existing empirical research presents the negative consequences of task conflict. That is, a considerable amount of task conflict can erupt as a consequence of negative group affect resulting from relationship conflict. In this sense, we concur with Schwenk's (1990) notion that affect-free cognitive conflict is scarce in reality. Therefore, we argue that task conflict may be detrimental to team functioning not because of its destructive tendencies but because it is often the result of relationship conflict; interpersonal animosity is expressed as task disagreements. In this study, however, we were not able to test this argument because of lack of performance data. Before making a hasty conclusion on the ramifications on performance of task conflict in an organization, we need to investigate thoroughly when the two types of conflict are correlated and how we can separate the two. Future studies need to focus more on investigating the mechanisms and conditions that can reduce the dysfunctional effects of relationship conflict on task conflict.
Although it is well-recognized that task conflict and relationship conflict are related, how they are related is relatively studied less. Our study provides insight into the pattern of associations between the two variables by showing evidence on the direction of causality between them. Our findings provide important insights to conflict literature by indicating that the direction of causality is reciprocal for some specific groups, particularly those with lower trust among members. However, for groups with higher levels of trust, relationship conflict can cause subsequent task conflict, whereas task conflict is not transformed into subsequent relationship conflict.
The result of testing cross-lagged effects reveals that relationship conflict is apt to cause task conflict, which suggests a promising research direction for future studies. Previous studies have exclusively tested the task conflict → relationship conflict linkage, but not the opposite. This was mainly because researchers were less interested in the relationship conflict → task conflict path or cross-sectional data were inadequate to test both directions. Our findings suggest that the relationship conflict → task conflict path is a more universalistic phenomenon and merits thorough investigation. As this study presents, the relationship conflict → task conflict causal flow has a valid theoretical background and is empirically significant.
We would like to mention several advantages in considering relationship conflict as a precedent of task conflict. First, relationship conflict is known to be related to individual differences such as personality (Asendorpf & Wilpers, 1998; Varela et al., 2008), attitude, or visible demographic characteristics (De Dreu & Van Vianen, 2001; Pearsall, Ellis, & Evans, 2008). In contrast, less is known about the source of task conflict (e.g., where it comes from, how it is proscribed or promoted, or who are more likely to perceive it). Therefore, if we start conflict management interventions from the relationship conflict, we are able to draw more implications that will help prevent or manage an appropriate level of conflict through direct interventions on the source of relationship conflict, from which indirect interventions on task conflict are also possible.
The result of moderation analysis provides important implications for trust literature. Our findings provide further support to an argument proposed by Dirks and Ferrin (2001) that, in many situations, trust should be expected to have a moderated rather than a direct effect on desired organizational outcomes. Second, although trust has been linked to a range of desired organizational outcomes (e.g., Dirks & Ferrin, 2001; Kramer, 1999), we are only beginning to understand its effects on task and relationship conflict. Our results indicate that trust can be expected to influence relationship conflict (a moderated effect), but not task conflict.
Finally, for practitioners to make use of research findings on the linkage between task conflict and relationship conflict, it is critical to know which is the determinant or the consequence of a particular type of conflict. Groups suffering from a negative spiral of task and relationship conflicts are likely to be those who are low in trust. Our results indicate that if trust is repaired in these groups, then the vicious cycle between task and relationship conflict can be broken due to the trustworthiness-cooperation spiral (Ferrin, Bligh, & Kohles, 2008). In addition, managers should not expect to reduce relationship conflict by focusing on improving the conflicts revolving on tasks because our results suggest that task conflict is unlikely to influence relationship conflict. Instead, it may be necessary to address interpersonal-related conflict among group members directly. Unless relationship conflict is handled well, its detrimental effect is more likely to be transformed into task conflict.
Limitations and future research directions
Although the present study attempted to shed light on the existing body of conflict literature by suggesting a rigorous prediction of the causal link from relationship conflict to task conflict, this study is not without limitations.
First, although a longitudinal panel design is often considered the preferred method to explore causal relations among variables, it is still limited in eliminating alternative interpretations attributable to a third or unmeasured variable. For example, social desirability or common method variance remains an unknown factor that could potentially influence causal relations between study variables. We tried to minimize the effect of unspecified variables by including age, sex, tenure, and value diversity as control variables and by allowing residual correlations in the test models. Replication of this study through a research design that enables maximal control over other factors (e.g., field experiment) is indeed desirable for future studies.
Second, self-report measures have been used exclusively in previous research on task and relationship conflicts. We also employed Jehn's (1994, 1995) self-report measures for this study. As Torrance (1957) noted, however, individuals tend to perceive all disagreements as evidence of personal rejections whether they are personal relationship- or task-related (318). Despite empirical separation of the two factors, this general tendency of human perception (i.e., difficulty in differentiating relationship conflict from task conflict) may have caused inflated correlations between the two constructs and could be an explanation for the high intercorrelations in existing research that relies on self-report measures. Thus, future researchers are encouraged to employ methods other than self-report measures.
Third, among many conditions that possibly influence the transformation between relationship conflict and task conflict, we tested only one possible moderator (i.e., trust). Future study merits identifying other moderating variables that may amplify or suppress the relationship. One promising research area is the role of conflict-relevant interactional norms (Yang & Mossholder, 2004). It is expected that some behavioral norms may reduce the strength or speed of relationship conflict transformed into task conflict. For example, under the influence of group norm that is intolerant to conflict among members or that proscribes open debates, group members are more likely to avoid situations that might bring conflict, whether it is a personal bicker or a task-related argument. If any type of conflict happens, then even appropriate interactions among group members will be halted, which blocks one type of conflict mutating into the other. In addition to its effect on the relationship conflict → task conflict flow, conflict-relevant norms may directly influence the magnitude of members' perception of conflict. For example, a group norm that promotes constructive controversy (Tjosvold, 1985) is more likely to encourage group members to accept others' perspectives without feelings of being threatened, which leads to weaker conflict perception. Thus far, there has been little empirical research on the role of group norms on conflict-relevant interactions. Further studies on the role of conflict-relevant interactional norms or conflict management strategies (Behfar, Peterson, Mannix, & Trochim, 2008; DeChurch, Hamilton, & Haas, 2007) will extend our understanding of the nature of the association between task and relationship conflict.
Fourth, the nature of the sample and the tasks the student groups performed may limit the generalization of our findings. Although the student groups worked like “real” teams while engaged in group projects for 13 weeks, they may have different characteristics compared with actual work teams in organizational settings. The student groups in this study were relatively homogeneous in terms of education, age, work experience, power status, and so on. They were also newly formed groups rather than already existing groups where the group members have a history of working together. The setting of the present study provides an empirical baseline for exploring how the relationship between task conflict and relationship conflict changes over time in initially formed groups because a newly formed group is relatively free from established relationships among its members. However, one may argue that causal directions may depend on the nature of their past work experiences together, which was not considered in this study. In particular, member familiarity affects task-relationship conflict associations. Familiar group members are more likely than strangers to form favorable first impressions, trust one another (Davis & Todd, 1985), accept other members' ideas at face value (Uzzi, 1996), and be comfortable disagreeing with one another (Gruenfeld, Mannix, Williams, & Neale, 1996). Furthermore, the groups in our study differ from existing work groups in the sense that they had a clearly defined life. If relationships among group members persist over time without delineated deadlines, the escalation of conflict is likely to happen, especially in low-performing teams (e.g., Jehn & Mannix, 2001) because one conflict type (either task or relationship conflict) is apt to stimulate the other conflict type. Thus, our findings are likely to generalize to a limited set of work teams in organizations (e.g., ad-hoc project teams) not necessarily concerned with establishing social relations and anticipating future relationships among team members.
Fifth, past performance feedback which was not considered in our study is likely to trigger emotional responses among group members and affect subsequent group interactions. Peterson and Behfar (2003) found that poor performance feedback increased later task and relationship conflict in groups. One may argue that positive feedback intervention may prevent task conflict from evolving into relationship conflict and vice versa. Therefore, future studies need to replicate our findings in teams where members have previously worked together and received performance feedback.
Despite the limitations, our study makes a contribution to the literature on task and relationship conflict by suggesting a causal flow from relationship conflict to task conflict. We hope this study will help advance our understanding of the dynamic nature of conflict in groups.