Conflict in the boardroom: a participant observation study of supervisory board dynamics

We develop a conflict-oriented model of board task performance and argue that a common framework, that is, a shared understanding of its role, helps boards to perform well. Conflict is the mediating effect through which this plays out. We posit that a common framework increases board task performance because it reduces intragroup relationship conflicts, increases task conflicts within the board, and reduces conflicts in the relationship between board and CEO. We explore the model through a comparative participant observation study of 11 supervisory boards in action. The results show that while low levels of relationship conflict are typically considered a sign of a well-functioning board, the avoidance of relationship conflict negatively impacts board task performance and may lead to ‘cognitive blindness’. Boards of directors should manage—rather than avoid—relationship conflict. Based on our findings, we suggest an extended model of conflict in boards that takes into account the negative effect of conflict avoidance.

appearances or ''skeletal features'' (Useem and Zelleke 2006), such as board composition, became prominent variables in corporate governance research. But it is now increasingly clear that the determining factors of board task performance are not these skeletal features, but rather how board members engage and work with each other (Kemp 2006;Minichilli et al. 2012;Nicholson and Kiel 2004;Pettigrew 1992;Roberts et al. 2005;Sherwin 2003). Consequently, there is a pressing need to closely examine actual board behavior and decision-making processes in action (Van Ees et al. 2009: 314).
A particularly puzzling issue concerns the effects of conflict on board task performance. We follow the common approach in behavioral corporate governance research and define board task performance as the board's ability to perform its control and service tasks effectively (Forbes and Milliken 1999: 492). In general, conflict is considered to have a negative influence on group performance (De Dreu and Weingart 2003). Building on this insight from the group effectiveness literature, many authors in the behavioral corporate governance literature stress the importance of social cohesion and harmony for board task performance (Huse 2005;Westphal and Bednar 2005). Forbes and Milliken (1999), for example, argue that board task performance improves when members are naturally drawn to each other and want to stay on the board. At the same time, there are indications that conflict is actually beneficial for board task performance because it induces serious debate and the evaluation of alternatives. Jehn (1995) introduced the helpful distinction between relationship conflict and task conflict. While relationship conflict has a negative effect on group performance, task conflict can have a positive effect, because multiple viewpoints and a more careful evaluation of alternatives improve the quality of decision-making (Forbes and Milliken 1999). Task conflict helps boards to avoid inferior decision-making and groupthink (Huse 2007;Janis 1972).
Despite the growing literature on board dynamics we still do not know whether, how, and under what conditions conflict contributes to board task performance. Empirical studies that aim to unravel this effect, unfortunately, give mixed results. Some studies find the expected positive effect of task conflict on board task performance (e.g. Bailey and Peck 2011;Wan and Ong 2005), while other studies do not find a significant effect at all or only for particular board tasks (e.g. Minichilli et al. 2012;Zona and Zattoni 2007). From the discussion of these findings we take two ingredients that we believe should be taken into account in order to solve the puzzle of whether conflict enhances or undermines board task performance. First, whereas many studies on board processes only consider task conflict, we argue that we need to take into account relationship conflict as well. One important reason for this is that task and relationship conflict are interdependent. While high task conflict can be positive for board task performance, because it increases the quality of strategic decision making, it can also trigger relationship conflict, which in turn decreases board task performance. This interaction may explain part of the empirical and theoretical confusion. Second, combining the findings of previous board process studies on conflict with insights from the team effectiveness literature, we suggest to take into account the extent to which a board of directors has a common framework, that is, a shared understanding of its role. Stiles and Taylor concluded from their in-depth qualitative study on boards that the way in which the board defines its own purpose is a crucial prerequisite for the participation and trust development within the board. Shared beliefs on board tasks define 'the rules of the game for board activity' (Stiles and Taylor 2001: 114). In a similar vein, Nadler finds that 'selecting a level of engagement provides the philosophical framework for everything that follows' (Nadler 2004: 105). Building on consistent findings in conflict management studies that show that developing cooperative goals is an effective way to promote so called 'constructive conflict' (Tjosvold et al. 2014), we argue that an established common framework enables a board to benefit from task conflict and at the same time avoid relationship conflict. The question that inspires our research is therefore: how does conflict mediate the effect of a common framework on board task performance?
We develop a conflict-oriented model of board task performance and investigate its merits through participant observation. Boardroom observation is the most promising approach for uncovering board dynamics, but has been described as difficult due to access problems, legal considerations and confidentiality (Clarke 1998: 58;Samra-Fredericks 2000: 245). Over the past years, we have been actively involved with a considerable number of boards in the capacity of boardroom consultants. Not only did we interact closely with board members, we did so in a setting of individual and collective reflection and evaluation. We used our close involvement with 11 of these boards to investigate the role of conflict in board dynamics. This allows us to combine the benefits of a qualitatively rich research approach with the benefits of a comparative research design, and as such, go beyond thick description and theory generating. All the cases we investigated have a twotier governance structure. With a separate non-executive supervisory board it is more challenging, and thus more important, to resolve disagreements between the executive and non-executive directors (Bezemer et al. 2014b;Peij et al. 2012).
The outcomes show how the relationship between common framework, conflict and board task performance is not as straightforward as some of the literature would lead us to believe (e.g. Tjosvold et al. 2014). We find, as expected, more relationship conflict where board members disagree on their role. But a common framework does not lead to low levels of relationship conflict per se. And while a common framework can lead to benefits by promoting task conflict, we find that effective chairperson leadership is a prerequisite for such benefits. Our findings contribute to the current body of literature in at least three ways.
First, by taking into account relational conflict as a separate variable, we allow for a more nuanced view on board dynamics. We add to the literature by establishing an interaction effect between task and relationship conflict and show how the positive effect of task conflict on board task performance is dampened by the relationship conflict it triggers. Our model therefore suggests that, contrary to for instance De Dreu (2006), high levels of task conflict are not undesirable per se, but its triggering effect on relationship conflict is. Second, we extend the conclusion of Minichilli et al. (2012: 209) that the relevance of task conflict depends on the institutional context of the board. Our results show that, besides this institutional macro level context, the micro level context of the board as a social group is as relevant. Third, we show how a shared understanding of the board's role among members provides a fertile ground for task conflicts and might prevent task conflicts to trigger dysfunctional relationship conflicts. Fourth, while the literature on conflict in boards generally agrees that relationship conflict within a board has a negative effect on board task performance, we find that avoiding conflict paradoxically often hinders good governance. This calls for a reappraisal of relationship conflict and a fundamental revision of our understanding of the effects of conflict on board task performance, both in research and in practice. Our findings warn not to overstate the importance of social cohesion for boards of directors. In the conclusion, we therefore introduce a revised model of conflict in boards and introduce conflict avoidance as key element of boardroom dynamics.

Conflict and board task performance
Conflict is typically considered to have a negative influence on group performance (De Dreu and Weingart 2003): it decreases satisfaction of group members; it decreases group productivity; and hinders the exchange of information needed for effective decision-making (Jehn and Bendersky 2003). However, there is a growing awareness that under certain conditions conflict can also be constructive (Tjosvold 2008a). Relationship conflict arises through interpersonal incompatibilities between group members and is expressed in tension, animosity and annoyance. Task conflict arises from disagreement between group members about the content of the tasks to be performed due to differences in viewpoints, ideas and opinions (Jehn 1995). While relationship conflicts are generally considered detrimental to effective group performance, Jehn (1995) argues that task conflicts actually increase group performance. Indeed, task conflicts are positively correlated with effective decision-making by top management (Amason 1996). This effect, however, only occurs under certain specific conditions (De Wit et al. 2012;O'Neill et al. 2013). First, the effect is context dependent; task conflicts produce more positive outcomes for strategic and complex decision-making and more negative outcomes for routine tasks (O'Neill et al. 2013). Second, the effect depends on the organizational level at which they occur; the higher in the organizational hierarchy, the more positive the effects of task conflicts are (De Wit et al. 2012). This suggests that boards benefit from task conflict, because they are typically involved in non-routine decisionmaking at the apex of organizational hierarchy.
Building on these insights, Forbes and Milliken (1999) developed an influential model of board task performance wherein conflict plays a central role. They argue that conflict can have a positive effect on board task performance because it requires CEOs to explain and justify their positions in light of alternative perspectives and because critical evaluation of more alternatives by the board will enhance the strategic decision-making quality. Forbes and Millliken's seminal article inspired a range of studies that seek to uncover the role of conflict and other board processes, such as the effort norms and the use of knowledge within a board. However, the empirical studies that aim to unravel the effect of task conflict on board task performance give mixed results. Bailey and Peck (2011) do find the expected positive effect of task conflict on board task performance in boards of 119 publicly traded U.S. companies. However, in their study of 301 manufacturing firms in Italy, Zona and Zattoni (2007) find no significant effect of task conflict on board task performance. They suggest that this may be because task conflict favors the emergence of negative emotions in the board, counterbalancing its positive effects. Minichilli et al. (2012) considered the effects of task conflict on board task performance for 535 industrial firms in Italy and Norway and also found no significant standalone effect of task conflict. Moreover, when they include interaction effects between board processes and country, task conflict actually has a negative effect. In order to explain this unexpected result they turn to Hambrick et al. (2008), who pointed at a customary reluctance of boards towards open and candid discussion. This, they argue, 'can make conflicts an anguished experience for board members' (Minichilli et al. 2012: 209).
These considerations point at an interaction effect: high levels of task conflict may stimulate relationship conflict and as such indirectly reduce board task performance. This is consistent with self-verification theory, which suggests that challenges of viewpoints are seen as negative assessments of abilities and competencies (De Dreu and Weingart 2003;Mooney et al. 2007;Simons and Peterson 2000). Indeed, task conflicts have positive (or at least less negative) effects on group outcomes when they are not accompanied by relationship conflicts (De Wit et al. 2012). Unfortunately, hardly any research on conflict and board task performance takes relationship conflict into account. Kerwin et al. (2011) are a positive exception and do distinguish the impact of task and relationship conflict on decision quality in their study of nonprofit sports boards. First, they find the interaction effect: increased levels of task conflict trigger relationship conflict. Second, they observe an association between intensity of conflicts on one hand and the type of leadership of the board on the other. Strong and active board chair leadership is associated with less intense conflict. They argue that this is because leadership sets a common goal, which in turn reduces the intensity of the conflict.
A common goal among a board may thus affect the interactions between conflict and board task performance. From the group effectiveness literature we know that groups are effective when there is agreement among its members about their shared goals (Wageman 1995). Wan and Ong (2005) hint at this as well when they find a positive effect of task conflict for 212 listed firms in Singapore, but only for strategic task performance and not for the monitoring and service tasks. Based on interviews with board members, they suggest that this is because the strategic role is a 'grey area' where there is a thin line between the strategic direction role of the board and the strategic implementation role of management. It is the confusion on the role of the board that triggers both task and relational conflict. Yet, in the literature, we find only scant attention for how common understanding among board members of their goal and role impacts board task performance. Following Forbes and Milliken's original model, scholars look at social cohesion instead.
Forbes and Milliken suggest that board task performance is conditioned by ''the board's ability to continue working together, as evidenced by the cohesiveness of the board'' (Forbes and Milliken 1999: 492). Furthermore, Scarborough et al. (2010: p. 14) argued that without social cohesion, task conflict could 'slip into the realm of affective conflict with all its dysfunctional consequences.' However, the effect of cohesion remains elusive. In their study on financial risk handling during the credit crisis in British firms, McNulty et al. (2013) find that high levels of social cohesion within a board hampers the positive effect of task conflict. They suggest that ''effective behavioral dynamic involves a mix of challenge and support, or control and collaboration'' (idem: 73). Conversely, Heemskerk et al. (2015) find a strong and positive relationship between task conflict and cohesion in Dutch supervisory boards of educational institutions. Brundin and Nordqvist (2008) reveal how emotions such as irritation or anger are used to increase power and status in the boardroom. Moreover, Westphal and Bednar (2005) show how low levels of cohesion within boards can prevent non-executive directors from voicing minority opinions in the boardroom. There are thus ample indications that cohesion plays an important mediating role in the effect of conflict on board task performance, but we lack a proper understanding of this interaction (see also Scarborough et al. 2010: 14). One reason for this lack of understanding is that the concept of social cohesion is difficult to distinguish from (the absence of) relationship conflicts. Rather than social cohesion, the group effectiveness literature suggests to consider the role of cohesion at the cognitive level (shared goals). We therefore introduce the concept of common framework that can account for cognitive cohesion at the level of corporate boards.

A common framework for boards of directors
The role of a board is not straightforward and boards differ in the roles they take on (Finkelstein and Mooney 2003). In the absence of an established common framework, considerable differences may exist between board members about what the board's tasks actually are. First and foremost, a board has its (legal) duty to monitor management. This control task is typically most pronounced in the literature and refers to activities such as selecting and replacing executives and monitoring strategic initiatives (Forbes and Milliken 1999). Second, boards have the service task of providing advice and counsel for the organization and top management (Forbes and Milliken 1999). Some argue that boards have a distinct strategic task as well (Stiles and Taylor 2001), but others include this task in either the control task (Forbes and Milliken 1999) or in both the control and service task (Huse 2005). The board's role may be ambiguous when its members do not share similar views on what their role is, but the members themselves are not aware of this or there may be outright disagreement on the appropriate role of the board. Both ambiguity and disagreement can be important drivers for conflict and, in turn, lead to poor board task performance. In addition, even if the members agree on what their tasks are, a board can be more or less involved with their tasks.
Boards differ in how engaged they are in influencing management decisions and their companies' directions. Nadler (2004) introduced a simple and useful typology of boards along a continuum from least to most involved. In a passive board, board activity is minimal and the main job is to ratify (or rubber-stamp) managerial decisions. This role is reminiscent of the passivity of boards Mace noticed in the early 1970s (Mace 1971). The certifying board is most concerned with its control tasks and certifies that the organization is managed properly. The engaged board not only takes into account its control tasks, but also attends to its service tasks and strategic tasks. This board helps shape strategy. The intervening board is involved in making key executive decisions about the company. Finally, the operating board makes all key decisions, which are then implemented by management. In practice, boards may slide back and forth across this scale as their roles change with different issues and circumstances (see also Bezemer et al. 2014a;Machold et al. 2011;Nicholson and Newton 2010;Roberts et al. 2005). Notwithstanding the contextual nature, boards will have more-or less-cohesive collective understandings of their roles.
A board of directors may agree or disagree to a certain extent about their current role and involvement. Because roles are not static and corporate governance is under continuous pressure to develop and improve, directors can also agree or disagree on their desired role. Table 1 shows the four possible frameworks if we take a binary perspective on agreement and disagreement. In an established framework all board members share an understanding of their current and their desired roles. When a board does agree on the current role but disagrees on the desired role, the board is static and lacks a proper course. The forward-looking board does share a common understanding of how the board should work, but disagrees on their assessment of the current role. Finally, a board lacks a common framework when it disagrees on both the current and the desired roles.

A conflict model of board dynamics
We argue that a common framework helps boards to perform well, and that conflict is the mediating effect through which this plays out. We posit three ways in which the effect of a common framework on board task performance is mediated by conflict. First, it reduces intragroup relationship conflicts within the board. As mentioned previously, agreement among group members about their level of involvement and shared goals makes a group effective (Wageman 1995). In contrast, the lack of a common framework generates relationship conflict (Wakefield et al. 2006) and relationship conflicts hinder the necessary information processing as well as lead to indecisiveness (De Dreu 2008;De Wit et al. 2012). High levels of relationship conflict therefore have an expected negative effect on board task performance. With this comes the interaction effect: relationship conflicts use cognitive resources that can then not be used for task performance.
Second, a board with a common framework has a higher level of task conflicts because it can more easily discuss differences in viewpoints. Task conflicts are positive for group performance as long as they do not escalate into relationship conflicts. Groups that are able to prevent task conflicts from turning into relationship conflicts outperform others (Ensley et al. 2002). Eisenhardt et al. (1997), for instance, show how the management teams of tech companies that are able to keep constructive task conflict about issues from degenerating into dysfunctional interpersonal conflict are the most effective both in terms of speed and quality of decision-making. A common framework provides a common ground that allows for frank and open discussion and for diverging ideas. As such, it can contribute to high levels of task conflict. We therefore expect that a common framework positively affects task conflict and as such increases board task performance. Third, a common framework reduces conflicts in the relationships between board and CEO. Conflicts between executives and non-executives are generally considered to negatively impact board task performance (Westphal 1999;Westphal and Zajac 2013). Tension and conflict in the relationships between boards and CEOs negatively impact both the control and service tasks. When cooperation between boards and CEOs becomes difficult, teamwork and joint problem solving are less likely to emerge, leading to a decrease in the boards' task performance (Zhang 2013). A common framework within the board creates stability and reduces uncertainty vis-à-vis the executives. This in turn stimulates an effective working relationship (Sundaramurthy and Lewis 2003). If relationships between executives and non-executives are tense, executives are less inclined to seek advice from their boards and more likely to regard their boards' control task with suspicion (Bezemer et al. 2014b;Peij et al. 2012). We therefore expect that a more established common framework among (supervisory) board members leads to less conflict between the supervisory board and the executive(s) and, as such, improves board task performance. Figure 1 illustrates the expected relationships between a common framework, conflict, and board task performance.

RelaƟonship Conflict
Task Conflict

Board -CEO Conflict
Common Framework Board Task Performance ---+ + + - Fig. 1 A conflict-oriented model of board task performance 2 Setting, methods and data

A qualitative comparative research approach
The lion's share of the board process studies rely on indirect access to boards through surveys (Bezemer et al. 2014b;Calabrò and Mussolino 2013;Minichilli et al. 2012;Zattoni et al. 2012), documents such as board-meeting minutes (Schwartz-Ziv and Weisbach 2013), and legally prescribed disclosure forms (Agrawal and Chen 2011). This research is valuable but necessarily limited at the same time. Documents only reveal the outcome of board discussions and say little about the actual group dynamics during the discussions and, while surveys help to reach a large population, they cannot uncover unforeseen aspects of behavioral dynamics. Here, we used qualitative research methods to answer our research question. We conducted interviews with all individual supervisory board members and chief executives of 11 organizations as well as participatory observation of their interactions during one or several extraordinary board meetings.
A key reason why we resort to qualitative methods is that we need to establish the extent to which the theoretical expectations derived from the group effectiveness literature are in fact applicable to (supervisory) boards. Boards constitute a very special kind of professional group. They only meet a few times a year, their output is purely cognitive, and compared to top management teams boards have few opportunities to diminish or smooth over the differences that might separate them (Finkelstein and Mooney 2003;Forbes and Milliken 1999). In addition, much of the group effectiveness literature discusses how conflict affects information sharing within the group. Contrary to more standard teams, boards are not so much engaged with sharing information as they are with the critical questioning of the information they receive from the executives.
Because of difficult access many studies only observe a single board (Parker 2007(Parker , 2008Winkler 1974Winkler , 1987. Recent studies contribute to a more integrative understanding of board dynamics using multiple-case study analyses (Bezemer et al. 2014a;Huse 1998;Machold and Farquhar 2013). Comparing across cases allows us to go beyond single-case 'storytelling' and 'thick description' of board dynamics. We follow this line of research and present results based on a comparison of 11 cases. Increasing the scope of the study to multiple cases means that we lose some of the qualitative richness of our observations. However, this is compensated by the fact that comparing across cases allows us to explore the validity of the conflict model of board task performance as outlined in Fig. 1.

Setting
We collected information on 11 different organizations and over 90 board members and executives. All organizations are based in the Netherlands. This means that any differences we find between the board dynamics did not stem from legal, political and/or economic issues at the national level. The Netherlands is a small country with a strong international orientation and large numbers of multinational corporations. Its longstanding transatlantic political, cultural and economic orientation puts the country at the forefront of, for instance, shareholder-oriented corporate governance in continental Europe (Akkermans et al. 2007;De Jong et al. 2005). Dutch boards are well connected in the emerging European network of board interlocks (Heemskerk 2011;Heemskerk 2013), reflecting an open and international orientation that makes the Netherlands an appropriate site for our research. In the Dutch two-tier structure, an executive board is responsible for a firm's daily operations and a separate supervisory board meets regularly (typically six to ten times a year) to oversee the executives. There is no CEO-chair-duality. The chief executives are typically present at the supervisory board meetings. In all our cases, the non-executive supervisory directors are outsiders to the organization. Rather typical for the Dutch setting, none of the non-executive supervisory directors represent shareholders and none are former executives of the organization whose supervisory board they now serve on.
The organizations are private medium-to large-sized firms. Some are among the 30 largest corporations in the country. However, none of the organizations are publicly listed and some are nonprofit organizations. All organizations operate in a market environment with ongoing attention to practices of good corporate governance. Table 2 gives some descriptive information on the 11 organizations. Privacy and professional ethics prevent us from using some of the information we gathered, such as all information that might reveal the identity of the board members and their organizations. Therefore, we use dummy names and do not relate the outcomes of the evaluations with organization-specific indicators.

Participatory research as a method for corporate governance research
We observed boards in action in our capacity as boardroom consultants. This means that we actively participated in our own research setting. Thus, while our method was one of participant observation, we were not merely a 'fly on the wall', nor 'one of the lads' (Huse 1998(Huse , 2008Winkler 1987). We were outsiders asked by the boards to assist with their self-evaluations. This process always consisted of a research phase (interviews, document studies) and a board meeting aimed at evaluating board task performance and generating interventions to improve it. This approach allowed us to follow the recommendation of Samra-Fredericks (2000: 245) that if we wanted to have a richer understanding of boardroom activities, we not only need to ask board members about their behavior through surveys and interviews but also need to observe their interactions. Consequently, our approach resembles action research: a form of collective, self-reflective enquiry undertaken by participants in order to improve the rationality, coherence, satisfactoriness or justice of their own practices (Baard 2010;Jönsson and Lukka 2006;McTaggart and Kemmis 1988). In action research, the researcher brings special knowledge to the group and 'intervenes to encourage the community with whom the researcher is conducting research' to self-reflect and subsequently change its practices (Melrose 2001:160). In our role as consultants, we did precisely this; we brought special knowledge and made interventions that were aimed at stimulating self-reflective enquiry and improving board task performance. Our approach, however, differed from action research in a number of ways. Contrary to typical action and interventionist research, we did not include the effects of our interventions in this research; we only used our observations. Our approach is also different because we do not adhere to a full-fledged interpretivist or grounded perspective. Thus, when Melrose writes that in action research 'the researcher researches with, not on, other people, and does not treat the group merely as objects or sources of data' (Melrose 2001: 162), we agree with the former but not necessarily with the latter. We used our data collection as input for a comparative research design, aimed at validating and extending existing theory. Our perspective is close to postpositivism, as we compare our observations to prior theories and findings in an attempt to falsify them (see Gephart 2004). We thus adopt the perspective that qualitative data can be used to test or validate theories as well as extend them (George and Bennett 2005;Lee 1999).

Involvement during data collection
Our involvement with the research objects was different at various stages. This is in line with the best practice for effective participant observation; the study needs to include interviews with respondents, observation and direct participation, document analysis, and introspection (Conger 1998: 112). Our involvement was as follows: after being contacted by one of the board members (typically the chair) we had an intake meeting, sometimes by phone, but typically face-to-face, where we discussed the setup of the self-evaluation process, the goals, and our role. The self-evaluation always consisted of two phases. In the first phase, we studied the relevant documents, such as by-laws, regulations, strategic and organizational development documents, and outcomes of previous evaluations, and we conducted semistructured interviews with each individual board member as well as with the top executive(s). The interviews included a set of open-ended questions as well as a limited set of structured questions, which will be discussed in the next paragraph.
The aggregated results of these interviews were used as input for the next phase. The directors were aware that no individual names would be mentioned in the aggregated report; instead we indicated how often particular observations or statements were made. In a few cases the interviews were conducted by telephone. During this phase, our involvement is best described as 'observatory-participation' on the Participation and Observation Continuum (Vinten 1994: 30-31). We observed the board members and invited them to reflect frankly and openly on the board dynamics as they observed them. The second phase was the actual self-evaluation typically in the form of an extraordinary meeting of the board. Ranging from 3 h to a full day, these meetings were intensive and engaging. Typically, the board members appreciated the opportunity to discuss their own performance seriously and without time constraints. The aggregated results of the first phase served as input. In about half of the cases, the executive was present during this session. In terms of our involvement, we were fully situated at the participant end of the continuum. In each case, after the meeting, we produced a final report and one of us had an exit conversation with the chair.
This approach allowed us to combine direct observation of board dynamics with the results from the interviews about these dynamics. Together, this approach gave us an extraordinarily rich qualitative empirical basis. Not only did we have information about what the individual board members think about their boards, we also observed what happened when entire boards was confronted with the outcomes of the interviews and we observed the ensuing discussion. The data we used for analysis included the interview notes, the preparatory memos that we made before each meeting, the reports from these meetings, the notes we took during the entire process, and our final evaluations.
A well-known issue with participatory research is that the mere presence of the observer may affect the actions of the observed. In this case, we believe the potential bias is relatively limited. First, we did not solely rely on our observations during the meeting, but also on the rich body of material that we acquired during the interview phase. Second, we are not solely interested in what happened during those particular meetings, but also in how the members reflected on their past (and future) performance. In that sense, the board meetings partly functioned as focus groups. In some cases, the observations during the meetings served as extra evidence of, for instance, relationship conflict. But we never used the observations we made during the meetings as the sole indicators of board dynamics.

Measuring the common framework
We measured common frameworks through self-reported scores from the individual directors. In the interviews, we asked each board member to indicate both the current involvement and the desired involvement of the board as either passive, certifying, engaged, intervening or operating (Nadler 2004). The board members often responded with in-between categories ('we are in between passive and certifying'). To accommodate for this, we coded all responses on a nine-point scale that also included the four in-between positions. When we received multiple answers ('we are certifying and intervening') we coded them all. For each board, we calculated the mean position and the deviation of each response from the mean. The average of these deviations gave a disagreement indictor for the board (normalized by the number of responses). Each board had a disagreement indicator for its current situation and its desired situation. The next step was to determine if agreement was high or low. Because there was no outside benchmark, we considered boards with an above-average score on the disagreement indicator to have a disagreement, and those with below-average scores to be in agreement. For one of the 11 organizations, we only had information on its board's current common framework but not on its desired role. Therefore, we could not determine the type of its common framework. Table 3 shows the distribution of the cases over the types of common frameworks.

Coding conflict in the boardroom
Our purpose is to compare conflict across boards. Therefore, we apply what Lee (1999) called a 'factor analytic' approach: the reduction of large amounts of observations and data to produce meaning from this data. Coding the observations on conflict helps to make them more comparable between the various cases. First, we made a qualitative summary for each board concerning relationship conflict, task conflict and executive-board relations. Subsequently, we coded the results on a three-point scale. Relationship conflict could be considered low, medium or high. High conflict means that there were sustained, observable conflict and tension recognized as such by at least a considerable portion of the board. Low relationship conflict means the (near) absence of observable conflict. Conflict between the board and the executive(s) was coded similarly. For task conflict, we used the three categories: low, moderate, and high. Two of the authors coded the data individually, compared scores and discussed the few differences we encountered (high intercoder reliability). Looking back on the cases, the final scores on the board dynamic indicators are a good reflection of what we witnessed in our direct interactions with the boards (good face validity).
In the next section, we present the findings of our analysis in a two-step approach. First, we look at the patterns that emerge from a comparison of the coded data. Second, we use the qualitative richness of our material to further investigate these patterns in the context of the individual boards. In particular, we contrast examples where the patterns are clearly visible with examples where the patterns are not so evident.

Established frameworks and relationship conflict
How does a common framework affect conflict in the boardroom? Table 4 shows the aggregated findings. The two boards with established frameworks provide contrasting views. The case of Curefit shows how an established framework goes together with low levels of relationship conflict. The board of Curefit agreed that they should be more involved than they currently were. Most notably, they felt that they were not able to offer adequate countervailing power to that of the CEO. In this board, relationship conflict was low. The members described the board as harmonious and cooperative. Likewise, the relationship with the CEO was friendly and polite. They perceived the harmonious relationship as somewhat of an achievement, because there had previously been fierce debates with personal attacks. Strikingly, one member revealed that the observed low level of relationship conflict had a darker side as well. In his view, he and his colleagues were somewhat apprehensive about expressing their opinions, as they were afraid that doing so 1 board would stir up conflict. As we will see, such avoidance of relationship conflict turns out to be an important yet hitherto overlooked element of board dynamics. Northernhome, on the other hand, is the exact opposite of Curefit. Here, an established framework was coupled with high levels of relationship conflict. Agreement on their role was in part due to a tense relationship with the CEO. In their animosity toward the CEO, the board members joined ranks and had a clear understanding of how they wanted to function. Yet, this did not prevent high levels of relationship conflict within the board. Here, the relationship conflict was a classic example of the negative impact of board diversity. Within the board, there was a strong delineation between the two female members and the rest of the board. The female members felt insecure about their roles and contributions to the board, partly because they did not have much governance experience. In their efforts to make contributions they often pressed issues in their fields of expertise (HR and legal); issues that the other, male members of the board considered too operational. One of the male members said, 'HR and legal expertise is not really necessary for the board'. The male members dismissed the contributions of the female directors in both verbal and non-verbal ways. When this issue emerged during the meeting, the male members were surprised and somewhat embarrassed at how strongly the female members felt that they were part of a sub-group; the male members had never realized how brusquely they were accustomed to behaving. These two cases suggest that there is no straightforward relationship between common framework and relationship conflict.

How disagreement on roles can stir up relationship conflict
The cases of Novocare, SecureYouth and NewPsy show how relationship conflict relates to disagreement on the role of the board. The board members of Novocare disagreed on both their current and desired roles. This disagreement was related to an ongoing discussion about board independence. One of the non-executives was also an executive at another organization that operated in the same sector and market. The CEO was concerned about a potential conflict of interest and voiced this concern to the board. And while the member in question saw no problem whatsoever, a number of his fellow board members agreed with the CEO. For them, 'integrity is an issue for this board'. But the matter remained unresolved because their diverging views on the role of the board made it very difficult to determine how problematic the position of this particular board member was. As a consequence, the issue was regularly discussed during meetings, leading to relationship conflict. In the end, they decided to consult a governance association in their sector in order to resolve the matter. However, the board members recognized that the issue had already had a negative impact on board task performance. Perhaps most importantly, they failed to press the CEO to formulate strategic options at a time when the continuity of the corporation was at stake. 'We realize that we let opportunities go by to press the CEO to come forward with a clear strategy and make decisions', one board member said. The lack of a common framework and the ensuing discussion on board independency used precious cognitive resources and increased relationship conflict. How different views on the role of the board fuel relationship conflict is further illustrated by the case of SecureYouth. Its board had a forward-looking framework; members agreed that they should take on both their control and service tasks without becoming too intervening. However, the board was in disarray as to its current role; they sometimes acted certifying, sometimes intervening or even operational, but in other instances engaged-and between themselves they frequently disagreed about when who fulfilled which role. During the self-evaluation meeting, it became clear that disagreement about the current role was an important source of the high level of relationship conflict we observed. One of the board members took on a purely certifying role. All of the other board members were very critical about his contribution to the board. In their opinion, he was too critical. He was hardly ever satisfied with answers from the CEO and often asked the CEO to provide more 'evidence' to back up his arguments. While this is in itself a valid role for a board member, high levels of relational tension emerged because the other board members considered his behavior and style to be rude, unpleasant and undesirable. However, because they had not explicitly agreed on their role as a board, they found it very difficult to settle this issue. The issue was finally resolved under the leadership of a new chair. The new chairperson made it clear that a collective understanding of their role was important. He also decided that he wanted to 'foster a culture of respect between board members and with the CEO wherein domineering and disrespectful behavior was not accepted'. However, the level of relationship conflict among the board members was already so high that there was no room left for a proper discussion. The board member at the focus of the dispute called in sick moments before the self-evaluation meeting and left the board shortly after the board evaluation. The case of SecureYouth shows how the lack of a common framework can cause relationship conflict. However, it is not true that an established common framework necessarily leads to low relationship conflict. A common framework helps to resolve relationship conflict, but board chair leadership is an important additional prerequisite.
The third case that underscores the relationship between relationship conflict and a common framework is that of NewPsy. Similar to SecureYouth, the board had a forward-looking framework. Disagreement over the current role went together with an unfortunate combination of high task conflict and high relationship conflict. The source of both was that the organization was the result of a recent merger, and the two initial boards were very different in styles and cultures. One was businessminded and close to the certifying role, while the other was very involved and leaned more towards the intervening role. The composition of the new board and, in particular, the choice of the chairperson were thorny issues during the negotiations. According to one of the executives, it took the board over 20 meetings to come to a decision on this matter. But they did not spend time discussing what they saw as the role of the new board. The solution was a compromise: the chair of one of the organizations would be chair for 3 years, at which point the other chair would take over. In the meantime, the second man was appointed as vice-chair. In practice, they both acted as chairs but with different opinions on what a board and its executives should do. The two men were highly different in character. One was rather formal and introverted; the other a prominent figure in the sector, with a flamboyant personality and outspoken opinions on almost all matters ranging from strategy to operations. The other members generally saw the 'two-chair' solution as a bad compromise. The chairs acted too much on their own, while the other board members felt left behind. The result was a situation with high task and relationship conflict and an ineffective board. The executives, in turn, did not know what to expect from their board. As a result they tried to keep their board at a distance, further contributing to low board task performance. These cases show that disagreement on board roles causes relationship conflict, although agreement alone is not enough to ensure low relationship conflict.

Task conflicts and common frameworks
We now turn to the issue of what kind of board dynamics fosters a stable basis for in-depth and open discussion in a board and as such induce task conflict. Two of the cases under consideration show moderate levels of task conflict: Northernhome and SecureYouth. First, in the case of Northernhome, we see that high relationship conflict went together with moderate task conflict and an established framework. Notwithstanding the relationship conflict due to the male/female factions, the members judged the discussions to be open and to have ample room to disagree. We observed a rather healthy mix of critical questioning and diverse viewpoints during the discussions. However, an established framework alone is not enough for moderate task conflict, as the case of Curefit shows.
Second, at the board of SecureYouth, moderate task conflict went together with high relationship conflict. The board members disagreed on their current role and had fierce debates about this, as described above. But this went hand in hand with moderate task conflict. The main reason for this was that the board had little confidence in the CEO. This kept the board members acutely aware of their responsibilities concerning the continuity of the organization. This sense of urgency contributed to open discussions with room for different viewpoints. These observations beg the question whether relationship conflict is always detrimental to board task performance.
The three boards of NewPsy, CapitalCare and ForYou all displayed high levels of task conflict. We already mentioned how the recent merger and lack of common framework troubled the board of NewPsy. During board meetings, members typically approached problems from different angles and discussed them from their own perspectives. This did not result in effective decision-making, however, mainly because of the high relationship conflict and the conflict between the chairperson and the vice-chairperson. The lack of a common understanding of the role of the board hampered board task performance. A striking example of this inadequacy was that the executives were reluctant to present their ideas for the firm's strategy. They feared fuzzy input and a broad set of new requests for information to underpin their choices instead of a helpful strategic discussion.
In the other two cases, contrary to what we expected, high task conflict did not stir up high relationship conflict. At CapitalCare, the board had recently been renewed with a new chairperson and a new member. The board was still searching for its role and this led to fierce discussions. Members expressed highly different views on their role and tasks. For instance, one of the members, who had a legal background, saw it as his task to make a draft of new bylaws for the organization, while others (including the executives) saw that as too operational. On the question of how each member would describe the culture of the board we got seven very different answers varying form 'friendly and laid-back' to 'members keep their distance' to 'in flux'. However, this disagreement did not lead to high levels of relationship conflict. In fact, the members were hesitant to take fierce positions in the discussions because they thought this might negatively affect their working relationships. This also resulted in a situation where task conflict was not translated into board task performance.
At ForYou, we observed a very similar dynamic of conflict avoidance. Here, the task conflict was the result of a rather fundamental difference in how certain board members perceived the role of the board. ForYou had been formed over the previous years through a number of acquisitions and mergers. Some of the more senior board members still felt that they had specific responsibilities related to those parts of the organization where they had originally started as board members. In addition to these differences in opinion, two of the members refused to adhere to the terms of maximum tenure stated in the corporate governance code. This led to fierce debates with newer board members who thought that not adhering to the terms was an outdated and odd perspective. Some members suggested that personal interests were prevailing over the interests of the organization. At first glance, this did not, however, appear to be leading to relationship conflicts. In fact, it seems that relationship conflicts were purposefully avoided.

Avoiding relationship conflict
We often noticed a strong sense among board members that they should avoid relationship conflict. This was the case in four out of six boards with low relationship conflict. While this might sound laudable, there is a downside as well. When relationship conflict is actively avoided, it becomes increasingly difficult for a board to engage in a critical role and to have thorough debates on strategic issues. At ForYou, there was opposition to any form of relational tension. It was of the utmost importance that relationships be 'good'. Therefore, they framed the fierce and fundamental disagreement about governance roles as a technical issue of good corporate governance. First, they obtained advice from the governance committee of their professional organization. When this ruling was in favor of one faction, the second faction got a second opinion from a governance lawyer who ruled in their favor. During this time-consuming process, the board avoided the issue. But in the individual interviews we had with the board members, everybody-except for the two members in question-mentioned this issue as standing in the way of effective decision-making. Three members thought that the board as a whole was 'too nice and too friendly'. 'We are avoiding the actual issue', they told us.
In other boards, we found similar inclinations to avoid conflict. At Brightside, the board was eager to foster a 'good' relationship with the executives and therefore avoided conflict. They were afraid that relationship conflict among themselves would negatively impact their relationship with the executives. A conflict with a former CEO over financial issues just after a merger was still fresh in the memories of a number of board members. And in the example of the board of Reliable, the avoidance of relationship conflict led to low task conflict. As one board member said, 'we could say what we want, but we just don't do it'. Some boards were actually quite aware of this problem. At Blueroad, a board member told us that there was 'a danger that we place too much value in unity of opinions'. They remained aloof and did not get to the heart of matters in their controlling and advising tasks.
These observations suggest a more complicated interaction between relationship and task conflict than commonly assumed. First, while task conflict can indeed arouse negative emotions and lead to high levels of relationship conflict, for instance at NewPsy, we now observe another strategy wherein relationship conflict is actively avoided. When relationship conflict is avoided, there is often little room for task conflict. We saw this at Reliable, Blueroad and Brightside, where avoiding relationship conflict resulted in low levels of task conflict. Avoiding relationship conflict results in a situation of 'coerced cohesion,' wherein task conflict suffocates and groupthink may easily emerge.

Conflict between the CEO and the board
We expected that when a board has a cohesive common understanding of its role, it would be more predictable and reliable from the CEO's point of view. This would reduce conflict between the board and the CEO and improve board task performance. However, this assumption does not hold. In fact, conflict between CEOs and boards can also be a sign of a performing board.
The two boards with established common frameworks (Curefit and Northernhome) did not have low levels of conflict with their CEOs. At Curefit, the board had an occasionally tense relationship with the CEO. The CEO was a strong leader with an outspoken perspective on the organization and in-depth knowledge of the sector. The board had difficulty creating enough countervailing power to the CEO. The members were somewhat hesitant and reserved, in part because they recently had a fierce conflict among themselves. The CEO, for his part, basically did not accept any direction or advice from the board. He considered the meetings ''a waste of time'' and a ritual. The chairperson of the board was unable to provide effective leadership and a counterbalance to the CEO. He was a soft-spoken and risk-avoiding team player. He strived for a good and harmonious climate and this added to the imbalance in the CEO-board relationship that was very similar to the conflict avoidance we noticed in a number of boards. As a result, low relationship conflict combined with low task conflict led to a low-performing board.
In the case of Northernhome, an established framework went together with a complicated relationship between the board and the CEO, but here there was moderate task conflict. The situation was somewhat extraordinary; the CEO performed well in his executive capacity, but his position was negatively affected because he was involved (and indicted) in a court case concerning personal tax issues. The case was heavily covered in the media and that complicated the relationship between the CEO and the board to a great extent. The CEO only wanted to be accountable for his functioning as an executive of the organization. The board, however, had to take into account that his position had become troubled within the firm in regard to employees, customers and stakeholders. As a result of the high tension and strained relations, the CEO was reluctant to share information in an open and constructive manner and meticulously tried to avoid discussion. All this resulted in even more tension. In the end, the relationship between the board and the CEO became so strained that the CEO was suspended from his duties. The board members of Northernhome were genuinely, and most likely rightfully, concerned about the continuity of their organization and the CEO's ability to keep performing his duties. The complicated CEO-board relationship therefore signaled a wellperforming board.
SecureYouth is another example where the high level of conflict between a CEO and a board was a symptom of fundamental doubts about the ability of the CEO to perform his duties. SecureYouth was acting in a turbulent environment. Farreaching changes in the market demanded fundamental changes to the business model, but the CEO was not able to convince the board that he was doing the right things. He was strong-willed and tried to keep the board at a distance as much as possible, as he thought the board was too intervening. He scorned the Audit Committee as operational and without a feeling for sector-specific issues. The majority of the non-executives defined the relationship with the CEO as 'unstable and lacking trust'. The board concluded that 'it was not convinced that the CEO is in control'. Shortly after our observations, the CEO was dismissed. An effective board is willing to make tough decisions when necessary and this may very well be reflected in a certain amount of conflict between the board and the executives.
A crucial relationship is that between the non-executive chair and the CEO. At Blueroad, for instance, the chair judged the executives to be too operational and to be paying too little attention to strategy. He perceived this as a direct threat to the organization, because the environment was extremely turbulent and redefining strategy was necessary. The chair's perspective often led to tension with the chief executives. At Brightside, the medium-level conflict was mainly caused by a dominant chair together with a CEO who 'pre-cooked' many decisions. As mentioned above, this organization had just come off of a period of heavy tension and conflict with the previous CEO. In that crisis situation, the chair of the board moved from the certifying position to the intervening position. When the crisis was over and the CEO had been replaced, the chair found it somewhat difficult to go back to the certifying position that the rest of the board members had adhered to. As a consequence the other board members felt somewhat distanced and were often not able to fulfill their board tasks. Among board members relational tension was avoided, but the relationship with the CEO became occasionally tense.
The absence of conflict between a board and its executives does not necessarily mean a well-performing board, as our observations illustrate. The managing of both control and collaboration in the relationship between boards and executives requires a balanced-or even paradoxical-interplay of both trust and conflict (Sundaramurthy and Lewis 2003). Complicated board-executive relationships are more indicative of a board that takes its control task seriously than of an under-performing board.

Discussion and conclusions
Useem reminds us that 'what transpires behind the boardroom's closed doors is what ultimately counts' in corporate governance research (Useem 2003: 242). Here we lifted the lid off the black box of board task performance and found that conflict plays a crucial role. From the literature, we derived three mechanisms to explain how a common framework among board members concerning their role affects conflict and subsequently influences board task performance. We distinguished between Board-CEO conflict, task conflict and relationship conflict and also expected that task conflict triggered relationship conflict (see Fig. 1). Our findings partly support our expectations, but also suggest a number of extensions to our conflict-oriented model of board task performance.

Conflict in the boardroom
When the board disagrees on its role, and in particular on its desired role, we register more relationship conflict, as was expected. But at the same time, an established common framework is not sufficient to ensure low relationship conflict. The prevailing suggestion in the group effectiveness literature that a shared perception of roles contributes to reducing relationship conflict thus requires further additions and contextualization for boards. In the cases we observed, we found board chair leadership to be an important additional prerequisite in resolving relationship conflict. In addition, our findings suggest that relationship conflict can also be spurred by diversity in board composition. Minichilli et al. (2012: 209) have argued that task conflict is 'context-specifically relevant' for boards, referring to the national institutional context. Our study shows that the micro context of the board as a group is as relevant. The positive effect of task conflict on board task performance occurs more easily in an environment where a shared task perception exists, but also where critical discussion and exchange is stimulated and guided by active board chair leadership. The latter is in line with previous behavioral research on board chair leadership, wherein good leadership by the board chairperson was found to lead to a constructive team culture in the boardroom (Gabrielsson et al. 2007). This implies that for task conflicts to be fruitful, a fertile soil and proper board chair leadership are required.
We found that task conflict often goes together with relationship conflict and, in several cases, it was clear that task conflict in fact induced relationship conflict, which in turn reduced board task performance. We believe it is important to separate this effect of task conflict on relationship conflict because it can confuse the positive effects of task conflict on board task performance. Thus, where De Dreu (2006) argues that high levels of task conflict are undesirable, we argue that it is not task conflict itself that is undesirable, but rather its triggering effect on relationship conflict.
For CEO-board conflict our findings are not in line with our expectations. We did not find a clear pattern between common framework and board-CEO conflict. And perhaps more fundamentally, our findings suggest that we have to nuance the assumption in previous studies that conflict between the board and the CEO is negative per se (Zhang 2013). Whether this conflict has a negative effect on board task performance depends on the circumstances in which the firm and its board are operating. In a number of our cases, conflict was in fact a sign of a critical and controlling board. A well-functioning board is willing to make tough decisions and to refute the CEO when necessary, even if this leads to tension in the relationship with the CEO. Harmonious relations between boards and CEOs, therefore, do not always signal high board task performance. This is in line with earlier research that found that board members being overly trusting of CEOs leads to 'cognitive blindness' (Van Ees et al. 2008), and with research that found that challenging and questioning executives to account for their conduct is the most effective means of intervention and influence by non-executives (Roberts et al. 2005: S19).

Conflict avoidance
Going beyond what we expected from the extant literature, we found that the avoidance of conflict plays a prominent role in board task performance. It is an important, but often overlooked factor in board dynamics (Kuhn and Poole 2000). Counter intuitively, we found that striving for low levels of conflict often hinders board task performance. The role of conflict avoidance in board dynamics has consequences for the corporate governance literature. While intra-board relationship conflict and CEO-board conflict is generally considered to have a negative influence on board task performance (Jehn 1995;Mooney et al. 2007;Simons and Peterson 2000;Zhang 2013), the avoidance of these conflicts is even more harmful.
We must therefore also reconsider the importance of social cohesion for board task performance. In the literature, the importance of social cohesion is often overstated. Neill and Dulewicz (2010), for example, maintain that harmonious personal relationships between board members are a crucial, yet often neglected, driver of board task performance. And Forbes and Milliken (1999) reason that board task performance is improved when members are naturally drawn to each other and want to stay on the board. They seem to assume that when board members like each other this automatically signals a good set of relationships. However, our findings suggest that a board can certainly exist wherein the members are quite fond of each other and surely want to continue working together, but yet the board as a whole is hardly effective. In a rare qualitative example, and one that is more extreme than our cases, Golden-Biddle and Rao observed how the board members of a non-profit organization saw themselves as a family. Due to being a 'family' and seeing a need to relate in a friendly manner, board members very rarely expressed disagreement during meetings. Their commitment was to continuing relationships based on friendliness and conflict avoidance (Golden-Biddle and Rao 1997: 599).
Our findings serve as a warning not to simplify the idea of attraction and to confuse it with low relationship conflict. Rather, we should habitually become suspicious when dissent is absent (Schulz-Hardt et al. 2006: 170). This means that at the least we need to take into account the role of relationship conflict in boards. We find it remarkable that relationship conflict is so often ignored in research on processes within boards. Forbes and Milliken (1999)-although they made use of the distinction of Jehn (1995)-left relationship conflict out of their model of board processes. Minchilli et al. (2012: 196) even argue that relational conflict is not relevant because 'corporate boards represent a context in which relationship conflicts and personal antagonisms are less likely to take place than in other organizational teams'. Our findings not only show that this assumption is certainly not generally valid, but also suggests that the superficial absence of personal antagonisms may be the result of conflict avoidance.

An extended model of conflict in the boardroom
The outcomes of our qualitative comparative study suggest a revision of the model on how conflict, a common framework, and board task performance are interconnected. Figure 2  avoidance as a critical factor in board dynamics. On the right side are the relationships between conflict and performance. As in our original model, relationship conflict has a negative impact on board task performance while task conflict increases board task performance. Task conflict also increases relationship conflict, which in turn leads to lower performance. A common framework among a board of directors allows for task conflict and at the same time reduces relationship conflict. As such, it cushions the triggering of relationship conflict by task conflict. While relationship conflicts are harmful, avoiding conflict may be even worse. Thus, we introduce conflict avoidance as a hitherto neglected variable in board research. Avoiding conflict theoretically hinders board task performance in four related ways. First, avoiding conflict leads to less task conflicts, and suppressing task conflict makes a board underperform through a lack of exchange of different views. Second, there are also good reasons to suggest a direct effect of conflict avoidance on board task performance. This direct effect follows from the astute observation that 'the absence of conflict is not harmony, it's apathy' (Eisenhardt et al. 1997: 77). Conflict avoidance inhibits the decision-making performance of a board because 'it encourages the suppression of differences and either fast, unreflective decisions or indecisiveness' (Kuhn & Poole 2000: 563). It can lead to 'escalating indecision' where a board keeps investing (cognitive) resources in a decision process but where closure appears illusive (Denis et al. 2011). This is indeed what we observed in some of the cases we examined. Third, in similar vein, conflict avoidance can lead to procrastination when directors keep searching for information to assess alternatives under consideration (Ferrari and Dovidio 2000). As such, conflict avoidance is connected to an inability to deal with a crisis (Mordaunt and Cornforth 2004). All this implies that conflict avoidance negatively impacts board decision-making and hence board task performance. Finally, Tjosvold argues that conflict avoidance can actually aggravate differences in a group, because without interaction members are unlikely to come to understand the reasons for others' positions (Tjosvold 2008a, b; see also O 'Neill et al. 2013).
We found that a number of contextual factors play a role as well. Three antecedents are included at the left side of the model: board chair leadership, personality and context. First, a common framework requires leadership from the board chair to get the board to debate about and agree on their desired role. Board leadership by the chairperson is a driving force in enabling open and critical debate within a board (Leblanc 2005; Van den Berghe and Levrau 2004). Our findings suggest that a common framework is the conduit for the board chair leadership to produce constructive debate within a board. But the precise causality in the relationships between board chair leadership, a common framework and task performance remains an issue for future research. This calls for studies on board leadership that go beyond leadership structure or CEO duality and that investigate board leadership as behavior (e.g. Elsayed 2011;Zona 2014). Second, at the micro level of board behavior, the personalities of board members are likely to influence conflict avoidance. Some persons are more conflict avoiding than others. From the work of Westphal andStern (2006, 2007), Stern and Westphal (2010), we know that personality traits such as opinion conformity and a tendency to use flattery are beneficial to individual board members because it increases the chances of receiving board appointments. The tendency to appoint and retain conforming, non-critical board members with similar views (Westphal and Zajac 1995)-even when they are demographically different (Zhu et al. 2014)-can be an important driver for conflict avoidance.
Third, we discern three sets of additional contextual factors. First, the general factors related to the business environment, such as corporate governance rules and legislation. A second category of contextual factors concerns the firm-specific circumstances-such as recent mergers, firm performance, etc. The third category covers the history of the board, such as previous conflicts or newly appointed members. Recent experiences with high relationship conflict in the board and/or recent mergers can cause boards to avoid conflicts. While we are able to show how contextual factors are important for the boards we looked at, it is difficult to generate contextual patterns from this series of observations. Comparative studies across a larger number of cases, wherein the context varies for a restricted number of elements, would be a helpful next step in uncovering these contextual patterns.
We do not include board-CEO conflict in the revised model because our results do not provide us with a clear understanding of its impact on board task performance. The findings do suggest that an absence of conflict reflects blind trust rather than a good professional relationship. And, as a corollary, professional tensions between the executives and non-executives are signs of a well-performing board. Others have taken a different perspective altogether and considered the board-CEO relationship to be an element of board task performance rather than an antecedent (Sundaramurthy et al. 2014;Westphal 1999). In sum, it is not yet clear how board-CEO relationships relate to conflict, common framework and conflict avoidance, and more detailed observations on the matter are warranted. More indepth longitudinal research is necessary to unravel how CEO-board conflicts and common frameworks affect boards and firms. Further research should distinguish between relationship conflict and task conflict between a board and a CEO and pay specific attention to the role of the board chair in managing the relationships between the CEO and the other executives and board members.
Future studies that build on our revised model would do well to further integrate the intermediate factors between conflict and board task performance. The first step would be to integrate variables that are known to play a role in board dynamics as well, such as board diversity (Dobbin and Jung 2011;Miller and del Carmen Triana 2009;Olson et al. 2007), the use of knowledge and skills, effort norms (Forbes and Milliken 1999), and trust (Van Ees et al. 2008). Moreover, the findings we presented here are restricted to one institutional and legal setting that makes use of a two-tier governance system. In a one-tier system similar dynamics of conflict may well occur between executives and non-executives. Minichilli et al. (2012) argue that national context moderates the relationship between board processes and board task performance. Voordeckers et al. (2014), on the contrary, show how national board structures are to a large extent decoupled from actual board behavior in SMEs in Belgium, the Netherlands and Norway. Comparative behavioral governance research should take diversity in board dynamics across different contexts seriously.