Organizational Behavior and Human Decision Processes 116 (2011) 116128

Contents lists available at ScienceDirect

Organizational Behavior and Human Decision Processes

journal homepage: www.elsevier.com/locate/obhdp

The bigger they are, the harder they fall: Linking team power, team conict, and performance
Lindred L. Greer a,, Heather M. Caruso b, Karen A. Jehn c
a

Work and Organizational Psychology, University of Amsterdam, The Netherlands Booth School of Business, University of Chicago, USA c Melbourne Business School, University of Melbourne, Australia
b

a r t i c l e

i n f o

a b s t r a c t
Across two eld studies, we investigate the impact of team power on team conict and performance. Team power is based on the control of resources that enables a team to inuence others in the company. We nd across both studies that low-power teams outperform high-power teams. In both studies, higher levels of process conict present in high-power teams explain this effect fully. In our second study, we show that team interpersonal power congruence (i.e., the degree to which team members self-views of their individual power within the team align with the perceptions of their other team members) ameliorates the relationship between team power and process conict, such that when team interpersonal power congruence is high, high-power teams are less likely to experience performance-detracting process conict. 2011 Elsevier Inc. All rights reserved.

Article history: Received 25 March 2008 Accepted 29 March 2011 Available online 19 May 2011 Accepted by Linn Van Dyne Keywords: Power Intragroup conict Teams

Introduction Given the rise of work teams in recent decades (Cohen & Bailey, 1997), enhancing team effectiveness has become a signicant concern of modern organizations. Although an abundance of literature exists on the hallmarks of effective teams (for a review, see Kozlowski & Ilgen, 2006), research has yet to systematically assess whether teams at different levels in the organizationthat is, teams with different levels of powerare equally effective. This is surprising given the pervasiveness of power in social interaction (Fiske, 1993; Magee & Galinsky, 2008) and the large impact power can have on individual behavior (e.g., Keltner, Gruenfeld, & Anderson, 2003). Understanding the impact of team-held power on intra-team interactions and performance may be critical to understanding how to create effective organizational teams (cf. Mannix & Sauer, 2006). In this paper, we hope to help break new ground in understanding how team-held power affects the internal dynamics and performance of organizational teams. In particular, we are interested in comparing teams with high and low levels of team power within an organization. Team power is dened as the collective capacity of a team to modify others states by administering (i.e., providing or withholding) actual resources or punishments to others (French &
Corresponding author. Address: University of Amsterdam, Work and Organizational Psychology, Roetersstraat 15, Amsterdam 1018 WB, The Netherlands. Fax: +31 (0)20 639 0531. E-mail address: L.L.Greer@uva.nl (L.L. Greer).
0749-5978/$ - see front matter 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.obhdp.2011.03.005

Raven, 1959; Keltner et al., 2003). Our denition of team power thus aligns closely with past denitions of power held by individuals (with the sole difference being our focus on the collective capacity of a team to inuence others), and we believe the main effect of team power on the individual team members will be similar to that of individually held power on an individual. Indeed, research has shown that manipulating power by telling participants they belong to a high-power team leads to individual behaviors that are similar to those that result from manipulations of power at the individual level (e.g., Guinote, Brown, & Fiske, 2006). However, studies have not yet conclusively established this parallel, and additional uncertainty exists regarding what is likely to happen when multiple high-power individuals interact in a highpower team setting (cf. Mannix & Sauer, 2006). Therefore, a central focus of our paper is on what team power does to team interaction. Although team power can be rooted in various sources (e.g., expertise, information, structural), we focus primarily on team power as stemming from the teams position in the formal organizational hierarchy (Astley & Sachdeva, 1984; Ragins & Sundstrom, 1989). Research has suggested formal hierarchical level forms an important basis of team power within the organization (e.g., Astley & Sachdeva, 1984; Fombrun, 1983; French & Raven, 1959; Mintzberg, 1983). Teams with a high level of formal power in an organization include, for example, management teams or advisory teams that are in a position to control the outcomes of others in the organization; low-power teams include teams of entry-level employees who have less control over the outcomes of others in the organization.

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128

117

Note that although a position of power may afford teams a certain level of status, or respect and prominence, in the organization (e.g., Anderson & Berdahl, 2002; Guinote, Judd, & Brauer, 2002; Guinote et al., 2002; Ridgeway, 1991, 1997; Sidanius & Pratto, 1999), power and status are conceptually distinct constructs (Anderson, Srivastava, Beer, Spataro, & Chatman, 2006; Keltner et al., 2003; Magee & Galinsky, 2008). For example, a management team that gives itself large bonuses at the same time it asks the rest of the company to cut costs might have formal control over the company (high power) but is not likely to have the respect of its employees (low status). In this paper, our focus will be on the effects of team power (based on the resources a team controls via its position in the organizational hierarchy), although we acknowledge that much of our theorizing would likely be applicable for team status as well. We develop and test the general hypothesis that teams with high levels of formal power in an organization perform less well than teams with low levels of formal power within an organization, because of higher levels of conict within high-power teams. We tested this hypothesis in two eld studies of existing organizational teams. In our rst study, we nd that teams with high, rather than low, formal power have substantially more intragroup conict and lower team performance. Our second study replicated this nding utilizing a controlled information-sharing task with an objective team performance outcome. The use of this task allowed us to control for potential variation in task type and performance expectations between low- and high-power organizational teams, as well as to better assess the causal linkages in our theoretical model. We also introduced power congruence as a moderator of the relationship between team power and intragroup conict. Our ndings suggest that when members of high-power teams agree with their teammates on the amount of individual power they themselves hold within the team, those teams experience less intragroup conict. This integration of team and individual power offers managers and researchers alike an important potential pathway by which high-power team performance can be improved. Linking team power to intragroup conict and team performance At rst blush, one may expect teams with a high level of formal power in an organization to function better than low-power teams. High-power team members may have better education, more functional experience, and more highly developed political skills (Lazear & Rosen, 1981). Yet despite these positive qualities, we suspect that, all else kept constant, high-power teams may actually perform worse than low-power teams due to dysfunctional intragroup processes such as conict (cf. De Dreu & Weingart, 2003). Intragroup conict is dened as the process arising from perceived incompatibilities or differences between team members (e.g., De Dreu & Weingart, 2003; Thomas, 1992; Wall & Callister, 1995). Research has suggested three types of conict exist in teams: task conict (regarding task goals or outcomes), process conict (regarding the logistics of task accomplishment), and relationship conict (regarding personal clashes over topics such as values or personality) (Jehn, 1997; Jehn & Bendersky, 2003). We propose that high-power teams will have higher levels of all three types of intragroup conict than low-power teams, driven by competition between members of high-power teams for positions of power within their team (e.g., Bendersky & Hayes, in press; Chattopadhyay, Finn, & Ashkanasy, 2010; Greer & Van Kleef, 2010). We suggest that this competition for power within highpower teams stems from the addictive nature of power (Mulder, 1977). Possessing power can drive individuals to seek ever greater levels of power and control (Bruins & Wilke, 1992; Mulder, 1977), implying that membership in a high-power team can lead individ-

ual members to value and pursue power. In high-power teams, however, individual pushes for greater power may encounter frequent hindrances in the form of similarly power-hungry teammates, who can potentially challenge ones pursuit of power. In support of this idea of tense power dynamics in high-power teams, research has shown that high-power teams, such as top management teams (TMTs), are especially sensitive to inequities in the distribution of within-group pay, power, and prestige (Eisenhardt & Bourgeois, 1988; Greer & van Kleef, 2010; Siegel & Hambrick, 2005). These heightened anxieties about power may also predispose high-power teammates to lash out against each other (Baumeister, Smart, & Boden, 1996; Martorana, Galinsky, & Rao, 2005). Indeed, research shows that when high-power people feel threatened, as in the case of high-power groups in which highpower members interact with one another, they become defensive, dominant, aggressive, and rude (e.g., Chattopadhyay et al., 2010; Georgeson & Harris, 2006; Morrison, Fast, & Ybarra, 2009; SmithLovin & Brody, 1989). Furthermore, research has shown that those in power are less likely to behave politely toward others (e.g., Keltner, Young, Heerey, Oemig, & Monarch, 1998; Smith, Jost, & Vijay, 2008), attend to others perspectives (e.g., Galinsky, Gruenfeld, & Magee, 2003), feel compassion for others (e.g., Van Kleef et al., 2008), and are more likely to engage in aggressive, conictual behaviors (e.g., Anderson & Berdahl, 2002; Keltner et al., 1998; Smith et al., 2008). These negative tendencies are all likely to be amplied when high-power teammates threaten and interrupt each others quest for power. Thus struggles for individual power are likely to produce high levels of intragroup conict (task, process, and relationship) in high- more than in low-power teams. We propose the following: Hypothesis 1. High-power teams will have higher levels of conict (task, process, relationship) than low-power teams. In line with recent literature reviews (De Dreu, 2007) and a meta-analysis (De Dreu & Weingart, 2003), we further suggest that all three types of intragroup conict (task, process, and relationship) will be negatively related to team performance. The theoretical mechanisms explaining this reasoning center around the cognitions and emotions that conicts may provoke in team members. First, conicts can distract members from task accomplishment, decreasing team productivity and task efciency (Evan, 1965). In addition, research suggests that conicts of all types are often linked to negative affect (cf. Jehn & Bendersky, 2003). The perceived incompatibilities that give rise to conict can lead to a variety of negative emotions (cf. Bell & Song, 2005; Lazarus, 1991; Roseman, Antoniou, & Jose, 1996) such as frustration, resentment, and anger (e.g., Guetzkow & Gyr, 1954; Russell, 1978; Stearns, 1972). These negative emotions can impair the cognitive functioning of team members (Brief & Weiss, 2002). Therefore, based on these theoretical mechanisms (decreased task focus, negative emotionality, and decreased cognitive capacity), we propose that conict and team performance will be negatively related: Hypothesis 2. Conict (task, process, relationship) will have a negative relationship to team performance. Furthermore, we believe the performance disadvantage highpower teams experience relative to low-power teams will be primarily attributable to the higher levels of conict in high-power teams. Therefore, we propose the following: Hypothesis 3. Conict will mediate the relationship between team power and team performance, such that high-power teams will have higher levels of conict than low-power teams, and conict will have a negative relationship to team performance.

118

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128

Study 1 methods We tested the hypotheses in a multi-source eld study of preexisting workgroups within the sales unit of a telecommunications company located in the Netherlands. The sample included 22 teams (94 employees). The average team size was 4.27 (SD = 1.78), with team size ranging from 3 to 9 members. The average age was 41.9 years old (SD = 4.6), and 24% of the participants were female. The average team had worked together for 3.0 years (SD = 1.8).

Controls We initially controlled for team size, team tenure, task interdependence (four items, e.g., I have to work closely with my colleagues to do our work properly), organizational tenure, and diversity in gender, nationality, and work function. However, these variables did not signicantly affect the relationships in our model nor did they signicantly vary between high- and low-power teams, so we removed them from ensuing analyses. Treatment of multi-level nature of dataset Because our dataset involves individuals within teams, two potential levels of analyses existthe team and individual levels. However, as we theorized all of our variables to be team-level variables (and, methodologically, we worded the items to reect the group level of analysis [i.e. how much conict of ideas is there in this team?]), we explored the appropriateness of aggregating individual responses to the team level. We examined the interrater agreement (rwg) and the inter-class correlation coefcients (ICCs) as well as their corresponding F-tests. For all three team conict types, as well as perceived team performance, F-tests indicated signicant between-group variance, and ICCs and rwgs exceeded acceptable levels for aggregation (Klein & Kozlowski, 2000; LeBreton & Senter, 2008), indicating sufcient within-team agreement (task conict: rwg = .76, ICC[1] = .31, ICC[2] = .66; relationship conict: rwg = .80, ICC[1] = .22, ICC[2] = .55; process conict: rwg = .70, ICC[1] = .28, ICC[2] = .62; perceived team performance: rwg = .82, ICC[1] = .26, ICC[2] = .60). This evidence was sufcient to justify averaging individual responses to create team-level scores, which we then used in our hypothesis testing. Study 1 results Table 1 displays means, standard deviations, and correlations. Team power is positively related to process and relationship conict and negatively related to both measures of team performance.1 Table 2 shows the means and standard deviations for each of the variables at low and high levels of team power. In Hypothesis 1, we proposed that high-power teams would have higher levels of conict than low-power teams. In line with this hypothesis, we found that high-power teams had signicantly higher levels of relationship conict (b = .57, t(20) = 3.11, p < .01) and process conict (b = .54, t(20) = 2.89, p < .01). We did not nd a signicant effect on task conict (b = .27, t(20) = 1.24, n.s.). For the results of hierarchical regression analyses testing all our proposed relationships, please see Table 3. In Hypothesis 2, we proposed that all three forms of conict would be negatively related to team performance. In line with this hypothesis, we found that all three forms of conict were negatively related to team performance. Process conict was negatively related to both leader-rated performance (b = .60, t(16) = 2.97, p < .01) and member-rated performance (b = .72, t(20) = 4.68, p < .001). Relationship conict was also negatively related to leader-rated performance (b = .51, t(16) = 2.37, p < .05) and member-rated performance (b = .58, t(20) = 3.14, p < .01). Task
1 When coding the organizational charts, we spoke with representatives from the organization about the potential ways of rating team power, and they felt a very clear differentiation existed in their organization between teams that did and did not have power. However, as power is often viewed as a continuous variable, we also included a self-report measure of team power in our survey based on a continuous scale of 17. We ran supplemental analyses using this variable as a predictor variable, instead of the externally rated dichotomous variable. We compared linear and quadratic t for the relationship between team power and conict and performance, and found also that when using this continuous measure, team power was linearly, positively related to conict and linearly, negatively related to team performance, in line with the results from our coder-rater classication of teams. We did not nd the quadratic terms to reach signicance.

Measures Team members completed questionnaires while seated together in a conference room on the day of the study, with at least one empty seat between each member. Team leaders were not present and lled in the survey separately. Experimenters assured anonymity to all participants. In return for their participation, team members and leaders received reports on how their team functioned compared with other teams in the organization. On the survey, all items utilized a 7-point Likert scale, with 1 indicating low agreement and 7 indicating high agreement.

Team power We assessed team power based on the organizational level of the teams within the business unit. We obtained this information from company records provided to us. In line with our denition of team power, we classied teams that had control over other teams or over important decisions that would impact the rest of the company as high-power teams. Seven teams were classied as high power and 15 as low power. Examples of high-power teams included the management teams of several departments within the business unit, as well as an internal steering committee. In addition, as a measure check, we also asked team members whether their team occupied a high-power position within the company (specically, members responded to the question How much inuence does your team have in your company?). This measure was successful, as high-power team members reported their teams as having signicantly more inuence within the company than members of low-power teams reported their teams as having (t(21) = 1.82, p < .05).

Conict We measured intragroup conict by asking team members to respond to Jehns (1995) 8-item scale measuring task and relationship conict, and Jehn and Mannixs (2001) 3-item scale measuring process conict. Each scale exhibited sufcient internal reliability (task conict a = .90; relationship conict a = .76; process conict a = .91). A conrmatory factor analysis revealed that a 3-factor structure best t the data (v2(94) = 336.33, p < .001, CFI = .96, NNFI = .95, RMSEA = .10).

Team performance To assess team performance, we had each team leader rate his or her teams performance on the basis of three items (I believe this group performs well at work, This group is effective in getting things done in time, and I think in general this group is effective with respect to work). This scale had sufcient internal reliability (a = .93). For four of our teams, we were unfortunately unable to obtain leader ratings. As an additional measure, we also had all team members rate their own performance based on the same items. This scale also had sufcient internal reliability (a = .87).

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128 Table 1 Study 1 means, standard deviations, and correlations.a Variable 1. 2. 3. 4. 5. 6.
+ * ** *** a b

119

Mean
b

SD .82 .56 .79 .84 .71

1 .30 .65*** .56** .50** .46**

2 .65*** .72*** .41+ .66***

Team power Task conict Relationship conict Process conict Team performance leader Rating Team performance team Rating

3.26 2.00 3.01 5.25 5.54

.75*** .51* .58**

.60** .72***

.44+

p < .10. p < .05. p < .01. p < .001. n = 22. Dichotomous variable.

Table 2 Study 1 mean comparisons. Dependent variable F Low-power teams M Control variables Team size Team tenure Organizational tenure Task interdependence Gender diversity Nationality diversity Functional diversity Hypothesized dependent variables Task conict Relationship conict Process conict Team performance (team-rated) Team performance (leader-rated) n = 22. * p < .05. ** p < .01. .41 .20 2.21 .01 .01 1.15 .01 1.53 9.67** 8.32** 4.35* 5.01* 4.53 3.31 10.91 5.52 .19 .21 .58 3.15 1.87 2.74 5.76 5.50 SD 1.81 1.75 5.61 .55 .21 .21 .30 .20 .44 .63 .57 .50 High-power teams M 3.40 2.17 11.37 4.99 .23 .14 .77 3.65 2.57 3.78 5.00 4.50 SD 1.52 1.69 8.15 .58 .22 .21 .28 .36 .33 .84 .89 1.41

Table 3 Study 1 results of hierarchical regression analysis. Intragroup task conict Step 1 Task conict Relationship conict Process conict F R2/Adj. R2 DR2 Step 2 Team power F R2/Adj. R2 .27 1.53 .07/.03 .03 .57** 9.67** .33/.29 .29 .54** 8.32** .29/.26 .26 .49* 5.01* .24/ .19 .19 Intragroup relationship conict Intragroup process conict Leader-rated team performance Perceived team performance

.41 .51* .60** 3.19 .17/.11 .11 .38 2.96+ .28/.19 .08
+

.05 .23 .40 2.14 .32/.17 .17 -.26 1.83 .36/.16 .00 .42* 4.35* .18/.14 .14

.66*** .58** .72*** 15.54 .44/.41 .41 .27 9.60** .50/.45 .04
**

.39 .17 .22 5.78** .49/.41 .41 .24 4.73** .53/.42 .01

5.62 .26/.21 .21 .29 3.50 .32/.23 .02

8.80 .36/.32 .32 .13 4.29 .36/.28 .00

**

9.86 .33/.30 .30 .14 4.97* .34/.27 .00

**

21.92 .52/.50 .50 .04 10.47** .52/.47 .00

***

DR2
n = 22. + p < .10. * p < .05. ** p < .01. *** p < .001.

conict was marginally negatively related to leader-rated performance (b = .41, t(16) = 1.79, p < .10) and signicantly negatively related to member-rated performance (b = .66, t(20) = 3.94, p < .001).

Last, in Hypothesis 3, we proposed that conict would mediate the relationship between team power and team performance. To test this hypothesis, we followed the four-step procedure Baron and Kenny (1986) suggest. First, we showed in our test of Hypoth-

120

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128

esis 1 that the independent variable is related to the mediator team power is positively signicantly related to both relationship and process conict. Second, we showed in our test of Hypothesis 2 that the mediator is related to the dependent variableboth relationship and process conict are negatively signicantly related to both leader-rated team performance and member-rated team performance. Third, we showed that the independent variable is directly signicantly related to the dependent variableteam power is signicantly negatively related to both leader-rated team performance (b = .49, t(17) = 2.24, p < .05) and member-rated team performance (b = .42, t(21) = 2.86, p < .05). Last, as required for full mediation, we showed that the effect of the independent variable (team power) on the dependent variable (team performance, leader-rated and member-rated) becomes nonsignicant when we control for the mediator (conict) (for leader-rated performance: b = .26, t(17) = .41, n.s. for memberrated performance: b = .24, t(21) = 1.15, n.s.). Sobel tests conrmed these analyses: process and relationship conict mediate the relationship between team power and leader-rated team performance (process conict: z = 2.10, p < .05; relationship conict: z = 1.99, p < .05) and member-rated team performance (process conict: z = 2.50, p < .05; relationship conict: z = 2.40, p < .05).2 Study 1 discussion In this rst study of the effects of team power, we nd that team power has a predominantly negative effect on teams. Teams with high levels of power had lower levels of team performance. The level of process and relationship conict in high-power teams fully explains this effect; that is, high-power teams had higher levels of process and relationship conict, and process and relationship conict were signicantly negatively related to team performance. We did not nd a signicant effect of team power on task conict, suggesting the effects of team power may be felt predominantly in the interpersonal and process issues of teams. Issues of power are central to the allocation of resources at the heart of process conicts (Greer & Jehn, 2007) or the way in which members interpersonally relate to one another, which is central to relationship conicts (Jehn, 1995). Although this study provides an important rst test of the notion that the power of a team may have an impact on its processes and performance, it has several limitations. First, the sample size is small. Although tests for the inequality of variances were not violated, we would still like to see replication of our ndings in a larger sample. Second, the teams in this organization performed different tasks and had potentially different expectations for performance. Perhaps the tasks of the high-power teams were more difcult, which is why they performed worse, or perhaps leaders
2 We selected our measure of congruence to ensure we tapped into a specic kind of social perception within the teamself-other agreement and specically the correspondence between targets self-views and perceivers appraisals. In line with past research on social perception (Cronbach, 1955; Funder & West, 1993), we note that power congruence is distinct from other measures of perceptual correspondence. For example, it is distinct from the more general construct of consensus (general agreement among perceivers in appraisals of a target), because high levels of such consensus could still reect low levels of power congruence if the consensus view differed greatly from the targets self-view. Because of our focus on correspondence with self-view, our priority in measuring power congruence is to compare actual self-views and actual peer appraisals, rather than to examine, for example, the variance created by unique targets, perceivers, or relationships in average perceptions (which is the focus of models such as SRM). Our analytical priorities are therefore similar to those of Anderson and colleagues (Anderson et al., 2006, p. 1099) in their analysis of elevation accuracy. Because we are likewise focused on a construct that is dened in terms of the difference between actual self-ratings and actual peer ratings, we nd that differences in raw ratings aggregated to the group level allow us to capture our construct appropriately while still eliminating any problems with the dependence of raw ratings on group membership.

expected more from high-power teams than low-power teams. To remove extraneous factors, we would like to see an investigation of the performance differences between low- and high-power teams on a common task with an objective, task-based performance outcome. The cross-sectional nature of the data in this study is also limiting insofar as it prevents us from drawing causal conclusions. Perhaps poor performance by high-compared to lowpower teams leads to higher levels of conict in high-power teams, rather than the other way around as we proposed. A controlled task provides a distinct advantage in addressing this possibility, as we can show that conict on a specic task can impede the eventual quality of task performance, strengthening our investigation of our proposed causal path. Last, our ndings paint a bleak picture for organizationsnamely, that the teams with power in an organization are often ineffective. Both theoretical and applied value exists in investigating factors that can brighten this picture. By identifying conditions that may moderate the relationship between team power and conict and performance, researchers may be able to provide organizations with ways to help their high-power teams avoid the pitfalls of being in power. Study 2: the moderating role of power congruence In Study 2, we address the limitations of Study 1 as well as incorporate an important and theoretically relevant moderator of the relationship between team power and team conict. We examine the basic hypotheses proposed in Study 1 in a larger, more diverse sample. We test the differences between high- and lowpower teams during a controlled information-sharing task, which simulates a task common to the interactions of both low- and high-power teams. Additionally, we suggest in this study that the relationship between team power and intragroup conict may be ameliorated when team members perceptions of the power they hold within the team are in alignment with others perceptions of their power, thereby integrating notions of team-held power and individually held power within the team. When one investigates power in organizations, perceptions of individual power are of critical importance (Fast, Gruenfeld, Sivanathan, & Galinsky, 2009; cf. Fiol, OConnor, & Aguinis, 2001; House, Rousseau, & Thomas-Hunt, 1995). For example, a CFO may have considerable control over an organization, but within the management team, her (presumably lesser) power relative to the CEO may dominate perceptions of her ultimate behavior within the team and her capacity to inuence others. Theory on congruent team perceptions (the degree to which team members see others in the team as they see themselves; Polzer, Milton, & Swann, 2002) suggests congruency of these perceptions of power within the team are central to understanding team interactions. Therefore, in this study, we look specifically at power congruence, or the degree to which members perceptions of their individually held power within the team are in alignment with other members perceptions of their power within the team. We integrate these two sides of team power (the teams power in the organization and members perceptions of individual power within the team) by looking at how power congruence might moderate the effects of team-level power in the organization on intragroup conict and team performance. We test Hypotheses 13, as outlined in Study 1, and introduce a new hypothesis focusing on the moderating role of power congruence, which we will outline below. Fig. 1 illustrates the model proposed in this second study. The moderating role of power congruence In explaining the moderating role of power congruence, we draw on research that has suggested that although hierarchies

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128

121

Power Congruence

Team Power

Intragroup Conflict

Team Performance

Fig. 1. Study 2 model.

are inevitable in groups (Magee & Galinsky, 2008; Overbeck, Correll, & Park, 2005; Sidanius, 1993; Wegener, 1992), they are also dynamic, and perceptions of relative individual power within the team may conict (Owens & Sutton, 2001); that is, how an individual views his or her power may not be in agreement with how teammates view his or her power in the team (i.e., low power congruence). In contrast, the alignment of an individuals selfperception of power with how others view the amount of power he or she holds (i.e., high power congruence) may reduce uncertainty over ones place in the teams hierarchy and provide common internal scripts for expectations and coordinated action (Fiol, OConnor, & Aguinis, 2001), alleviating the competitive behaviors and power struggles among high-power team members. For low-power teams, which are not necessarily as plagued by such problems, we expect power congruence will not be of as much benet. Specically, we suggest that, especially in high-power teams, high power congruence will ameliorate the effects of team power on conict, whereas low power congruence will exacerbate such effects, primarily because low power congruence implies members might not know their place within the team, which might exacerbate the likelihood of conict within high-power teams. Sensitivity to the internal hierarchy runs high in these teams, and if some members overstep the bounds of their intragroup power, other members may readily reject and even punish them for such a misstep (e.g., Anderson, Ames, & Gosling, 2008; Anderson et al., 2006). The power literature has also shown that a sense of power will make people more approach-oriented and more likely to engage in conictual behavior (for reviews, see Keltner, Van Kleef, Chen, & Kraus, 2008; Keltner et al., 2003; Magee & Galinsky, 2008; Smith-Lovin & Brody, 1989). These tendencies imply that members of high-power teams will be more likely than members of lowpower teams to punish individuals who make a misstep by incorrectly estimating how much power they have (e.g., Anderson et al., 2008), given the higher likelihood for those with high power to take action in such situations (Keltner et al., 2003). The person making the misstep may then also take aggressive action within the team as when those with power come to feel incompetent, they have been shown to lash out against others (e.g., Fast & Chen, 2009). Therefore, low power congruence is particularly likely to exacerbate conicts in high-power teams. In contrast, when high power congruence exists, members may better know their place in the team, as clear power roles provide information about the appropriate degree of inuence to wield within the team (Bales, 1950; Berger, Rosenholtz, & Zelditch, 1980). By reducing uncertainty about appropriate behavior and providing common internal scripts for expectations and coordinated action (Fiol, OConnor, & Aguinis, 2001), power congruence can thus temper tendencies to-

ward competitive behaviors and power struggles in high-power teams. Power congruence may also benet high-power teams by reducing the threat members experience in interacting with their teammates. The very nature of being in a high-power team can be a threatening experience, as all members are likely to be interested in maintaining and advancing their power positions within the team (Mulder, 1977), and this quest for power may come at the cost of the power of other team members. When individual team members feel threatened, they may become defensive and dominant toward their teammates (e.g., Georgeson & Harris, 2006; Morrison et al., 2009). However, if these anxieties or feelings of threat could be reduced, members of high-power teams might be less likely to lash out against each other (Baumeister et al., 1996; Martorana, Galinsky, & Rao, 2005) and become embroiled in team conicts. We propose that power congruence may provide such a means to potentially reduce the threat experienced in such high-power teams. When power congruence exists, high-power team members perception of their individual power is in alignment with how their teammates view their power within the team (Mead, 1934; Swann, 1987). This congruence allows high-power team members to form condent expectations of one anothers behavior (cf. Swann, 1987) and may reduce uncertainty about their own positions in the teams hierarchy (cf. Berger, Ridgeway, Fisek, & Norman, 1998; De Cremer & Sedikides, 2005; Fiol, OConnor, & Aguinis, 2001; Ridgeway & Berger, 1986; Van den Bos & Lind, 2002). In high-power teams, this congruence may signicantly reduce the threat individuals feel when they interact with their other high-power team members. In addition, high power congruence allows team members to avoid directing excessive amounts of attention and energy to defending and advancing their self-view through impression-management behaviors (e.g., Swann & Ely, 1984), which high-power team members may already be overly disposed to do, especially when interacting with high-power others (e.g., Chattopadhyay et al., 2010; Greer & van Kleef, 2010). Those personal resources are thus made more available to focus on facilitating smooth and effective achievement of work tasks. In such situations, when high power congruence exists in highpower teams, the potential benets associated with being in a high-power team (e.g., greater resource control, more seasoned, politically savvy team members, or pride in being part of a highpower team) may more readily come into play and allow highpower teams to perform well. Therefore, we propose the following: Hypothesis 4. Power congruence will weaken the positive relationship between team power and conict, such that high-power teams will have lower levels of conict (task, process, relationship) when power congruence exists.

122

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128

Study 2 methods We tested our hypotheses using multi-source data collected from 42 pre-existing work teams (322 employees) of a multinational nancial corporation with ofces in the Netherlands. The average team size was 7.67 (SD = 3.04), with team sizes ranging between 3 and 13. The average team tenure was 3.65 (SD = 2.01) years. Twenty-ve percent of the respondents were non-Dutch, and 51% were female. The average age of employees was 32 (SD = 5.19). The teams participating in our study came from a variety of departments across the company, including internal audit, investment banking, and human resources.

1976). Not only must high-power teams in the organization, such as TMTs, rely on their collective decision-making abilities (e.g., Henderson & Fredickson, 1996), but low-level teams responsibilities (e.g., customer service, production) also increasingly include making collective decisions about operating procedures, member roles, scheduling, and other issues (e.g., LaFollette, Hornsby, Smith, & Novak, 1996). A nal advantage of this task is that its objective performance outcome offers a clear and reliable measure of performance (McGrath, 1984). Measures We gathered our data in this study from multiple sources: team power was based on organizational records, team conict was based on student-assistant ratings of videotapes of the team interaction as well as team member self-reports, power congruence was calculated from scores given team members gave to themselves and others, and team performance was based on an objective measure of task performance. All survey questions were answered on a 7-point scale, with 1 being low agreement with the item and 7 being high agreement. Team power We based our distinction between high- and low-power teams on the organizational level of the teams. Fitting our denition of team power, and similar to Study 1, teams that had control over other teams and/or had the ability to make decisions that would impact the whole company were classied as high-power teams. Two company experts (one internal who worked in human resources and had in depth-knowledge of the company structure, and one external from an outside consulting company who had helped to design the current company hierarchical structure) determined this classication using organizational charts and company knowledge. Initial inter-rater agreement was 98%. The experts discussed and resolved classication for the one team about which they initially disagreed. They then identied 12 teams as high power and 30 teams as low power. Our team power variable is thus a dichotomous variable where 0 represents low-power teams and 1 represents high-power teams.3 In addition, as a measure check, we also asked team members whether their teams occupied a low- or high-power position within the company (specically, we asked members to respond from 1 to 7 to the question How much inuence does your team have within your company?). This measure check was successful, as high-power teams reported occupying higher-power positions than low-power teams (t = 4.43, p < .001). Of the 12 high-power teams, ve were management teams, such as supervisory teams of business units (e.g., the management team of the internal auditing department), and seven were policysetting teams that did not have teams reporting directly to them but still controlled enormous resources in the organization and had a large impact on company policy (i.e., the advisory teams for organizational communications). The low-power teams included, for example, secretarial teams and other lower-level teams, such as junior auditors. Power congruence We measured individual-level power on the basis of a roundrobin question wherein team members had to respond to the question How much inuence does each member of your team have within the team? for each of the members on their team, includ3 We operationalized team power in the same manner as in Study 1. Similarly to Study 1, we found that the team power was linearly (and not quadratically) related to conict and performance, using both continuous and dichotomous measures of team power.

Procedure We recruited teams for our study by contacting departmental heads within the nancial corporation where the study took place. Departmental heads invited all teams within their department to attend the sessions in which we conducted our study. We told teams that attended the sessions that participation was voluntary and they could leave at any time. In return for their participation, teams received a training and individualized team-effectiveness report. On the day of our study, team members entered a conference room and received an initial survey assessing demographic and other baseline team characteristics, including perceptions of power within the team. Following survey completion, team members learned they were going to be doing a task together to assess team dynamics. We then informed them that performance on the task was important for the quality of feedback they would get in their post-training report, in which they would be benchmarked against other participating teams. Immediately following the task, we administered a post-task survey regarding group dynamics during the decision-making task, such as the level of conict team members experienced. Note that when we administered the surveys, at least one empty seat separated team members from each other. We assured team members both orally and in writing that their responses would remain anonymous to everyone at their company. After the completion of the post-task survey, we debriefed the teams. For the task, teams completed a logic puzzle together, similar to traditional hidden-prole tasks (e.g., Stasser & Titus, 1985). In this logic puzzle, teams received information about ve different managers at their company who worked in ve different business units in ve different locations and were managing ve different projects with ve different budgets. We asked teams to determine which manager worked in which business unit in which location on which project and with which budget, on the basis of clues distributed to each of the team members (e.g., a member would receive ve clues, one of which would read Manager D does not work in Asia). Each member received the same amount of clues and the same amount of shared versus unshared clues as other team members. Because all clues were critical to full completion of the puzzle, the presence of the unshared clues meant no one member could solve the puzzle on his or her ownthey had to share information to reach a joint solution. We then informed teams that their performance would be based on what they were able to complete on the team solution sheet before time (15 min) was up. We chose this task because it resembles a common decisionmaking procedure for both high- and low-power teams, with members coming to the table to solve a dilemma based on their own unique information. Indeed, effective teams are able to identify a problem, gather relevant information, and evaluate and select alternatives to the problem that emerge from the relevant information (e.g., Janis, 1982; Kahneman, Slovic, & Tversky, 1982; Simon,

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128

123

ing themselves. In other words, members rated on a scale of 17 the capacity of each person in their team (including themselves) to exert inuence. We based self-perceptions of power on participants own ratings, and other-perceptions of power on the average of other members ratings for each other. In rating the power of their teammates, members showed high agreement with each other (ICC[1] = .49, ICC[2] = .88, F [1, 274] = 5.71, p < .001; rwg = .90). To calculate the degree to which members power self-views were congruent with how others viewed them, we followed the procedure employed by Polzer et al. (2002).2 We rst arrived at an individual-level incongruence score for each participant by calculating the absolute difference between self-perception and others-perception of ones power. We then took the average of these difference scores across all team members who rated the focal individual, resulting in an individual-level incongruence score for that person. Signicant F-tests, ICCs, and rwg scores conrmed the appropriateness of aggregating power congruence (ICC[1] = .19, ICC[2] = .44, F = 106.71, p < .001; rwg = .78) to the team level (Klein & Kozlowski, 2000; LeBreton & Senter, 2008). To obtain a team-level incongruence score, we averaged the individual-level incongruence score of all team members. We then reverse-coded this team-level incongruence score to reect congruence, to ease interpretation of results. Intragroup conict Coders assessed intragroup conict by rating the videos of the team task, and team members also provided self-report ratings on intragroup conict on the post-task survey. Such triangulation is especially important in studies of intragroup conict, as De Dreu and Weingart (2003) note that too much conict research has relied purely on self-report survey measures, which may not be as insightful as behavioral ratings. For the video-coder ratings, two coders, blind to the hypotheses, rated videos of team interactions during the 15-min decisionmaking task (due to logistical difculties, videotaping was not possible for 5 of the teams in our sample, so we excluded them from analyses). For the ratings of these videos, coders lled out the Jehn (1995) measure of task and relationship conict and the Jehn and Mannix (2001) measure of process conict on a scale of 17 (7 indicating high agreement). The coders exhibited high reliability in their rankings (for task conict, rwg = .98, for relationship conict, rwg = .78, for process conict, rwg = .99). Additionally, the internal reliability of the scales for each conict type was sufcient (task conict a = .75, relationship conict a = .91, process conict a = .95), and a factor analysis showed high discriminant validity for the conict types (three distinct factors, with all loadings above .79). For the self-report ratings, we used the same measuresthe Jehn (1995) scale for task and relationship conict and the Jehn and Mannix (2001) scale for process conict. These measures again showed sufcient internal reliability (task conict a = .92, relationship conict a = .84, process conict a = .91) and discriminant validity (a conrmatory factor analysis revealed that a 3-factor structure was the best t for the data (v2(322) = 400.53, p < .001, CFI = .96, NNFI = .95, RMSEA = .09). We then triangulated these two sourcesself-report survey ratings and video-coder ratings of actual behaviorto form our measure of conict. We found high agreement between these two sources (task conict: rwg = .83; relationship conict: rwg = .89; process conict: rwg = .91), so we averaged them together for each conict type to form the nal measures of conict. Task performance We assessed task performance by the number of correct answers the teams had for the logic puzzle. For each correct match

(e.g., identifying a manager with the correct project he managed), teams received a point. Scores could and did range from 0 to 20. Controls We initially controlled for gender diversity, national diversity, educational heterogeneity, job department, average team and organizational tenure, team size, general team interdependence outside of the training task (four items, e.g., I have to work closely with my colleagues to do our work properly), team communication frequency outside of the task (How frequently does this team meet?), a video-rating of task focus (How focused was this team on the task?, rwg = .89) and of task effort (How hard did this team work on the task?, rwg = .88), self-report of task effort (To what extent did you actively participate in this task?), and the routineness of the task for the team (How routine was the task in this exercise for your team?). Initial regression tests showed that only national diversity, team tenure, team communication frequency, task routineness, and task focus affected our model, so to conserve power, they were the only variables we included in subsequent analyses. Treatment of multi-level nature of dataset As this dataset involves individuals nested within the teams, we rst investigated the appropriateness of running our analyses at the team level (as specied by our theory). We found support for averaging individual team member responses to the team level with signicant F-tests and inter-class correlation coefcients (ICCs) (task conict: rwg = .82; ICC[1] = .33, ICC[2] = .79; relationship conict: rwg = .70; ICC[1] = .26, ICC[2] = .73, process conict: rwg = .76; ICC[1] = .27, ICC[2] = .74; power congruence (rwg = .78; ICC[1] = .19, ICC[2] = .44). Analysis We tested our hypotheses using hierarchical regression analysis. All variables were centered, according to the procedure of Aiken and West (1991). Study 2 results Table 4 presents means, standard deviations, and correlations. As seen in Table 4, team power has a signicant positive relationship with all three conict types and a signicant negative relationship with task performance.3 Means and standard deviations of these variables at low and high levels of team power are in Table 5. To test our rst hypothesis, we used hierarchical regression analysis to investigate if high-power teams differed from lowpower teams in their amount of intragroup conict. As seen in Table 6, high-power teams indeed had a signicantly higher amount of conict than low-power teams: high-power teams had higher levels of task conict (b = .53, t(36) = 3.18, p < .01), relationship conict (b = .49, t(36) = 2.90, p < .01), and process conict (b = .63, t(36) = 4.14, p < .001). In our second hypothesis, we proposed that intragroup conict would be negatively related to team performance. This hypothesis was supported for process conict (b = .45, t(36) = 2.26, p < .05). We did not nd a signicant effect of task or relationship conict on team task performance. In our third hypothesis, we proposed that conict would mediate the relationship between team power and task performance. This hypothesis is supported for process conict. First, our tests of Hypothesis 1 show that team power is signicantly related to process conict. Second, tests of Hypothesis 2 show that process conict is signicantly, negatively related to task performance.

124

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128

Table 4 Study 2 means, standard deviations, and correlations.a Variable 1. National diversity 2. Team tenure 3. Communication 4. Task routineness 5. Task focusb 6. Team powerc 7. Power congruence 8. Process conictb 9. Task conictb 10. Relationship conictb 11. Task performance
* ** *** a b c

Mean .35 3.65 5.17 3.67 3.22 .74 2.19 1.98 1.49 7.05

SD .25 2.01 .55 1.12 1.41 .37 .94 .58 .38 4.16

1 .02 .08 .23 .02 .24 .21 .43** .28 .06 .40*

10

.06 .18 .13 .03 .02 .05 .10 .15 .16

.01 .21 .06 .21 .06 .14 .09 .09

.02 .22 .14 .07 .05 .04 .04

.06 .14 .09 .14 .25 .27

.19 .62*** .44** .44** .39*

.04 .23 .17 .09

.65*** .48** .55**

.64*** .36*

.36*

p < .05. p < .01. p < .001. n = 42. n = 37. Categorical variable.

Table 5 Study 2 mean comparisons.a Dependent variable F Low-power teams M Team size Team tenure Organizational tenure Task interdependence Gender diversity Nationality diversity Functional diversity Communication Task routineness Task effort self-report Task effort video-rated Task focus video-rated Task conictb Relationship conictb Process conictb Team power congruence Team task performance
* ** a b

High-power teams SD 2.31 .39 .54 .77 .03 .26 .25 .50 1.03 .16 .30 1.33 .10 .06 .44 .39 .60 M 6.50 1.61 2.95 6.16 .45 .28 .44 4.59 3.14 5.22 5.38 3.58 2.46 1.81 2.85 .55 3.00 SD 3.07 .44 .49 .55 .05 .33 .32 .53 .89 .35 .67 1.07 .19 .12 .77 .21 1.15

.31 .11 .48 1.23 2.55 1.19 .10 .19 .10 .72 .08 .01 8.30** 8.40** 6.29* 3.21 6.09*

7.79 1.46 2.42 6.09 .36 .38 .48 5.43 2.91 5.11 4.75 2.94 1.85 1.41 1.87 .79 6.21

p < .05. p < .01. n = 42. n = 37.

Third, team power is directly signicantly, negatively related to task performance (b = .36, t(36) = 2.33, p < .05), and this signicant relationship disappears (b = .29, t(36) = 1.85, n.s.) when we add process conict into the equation. We also tested for mediation with a Sobel test, which yielded a signicant result as well (z = 2.08, p < .05). In our fourth hypothesis, we proposed that power congruence would moderate the relationship between team power and conict, such that in teams with high power congruence, team power would be less positively related to conict. This hypothesis was supported for process conict (b = .48, t(36) = 2.13, p < .05). A contrast analysis showed that compared with all other teams, teams with high power and low power congruence had the most process conict (t(1, 34) = 9.18, p = .004, MeanHigh power-Low congruence = 3.47, SDHigh power-low congruence = .40, MeanAll other cells = 2.18, SDAll other cells = .15). The interaction plot, as seen in Fig. 2, revealed an ordinal interaction: power congruence did not appear to affect low-power teams, but for high-power teams, when power congruence was high, team power was less positively related to process conict. We did not nd effects of power congruence on the relationships between team power and task conict, relationship conict, or

team performance, preventing us from nding moderated mediation. A Sobel test of an indirect moderated mediation path also did not reach signicance. Study 2 discussion In this multi-source eld study, we nd that high-power teams perform signicantly worse than low-power teams on a decisionmaking task. Higher levels of performance-detracting process conict in high-power teams fully explain this effect. However, we do nd that team power is less negative when high-power team members perceptions of how much power they have within their team is in alignment with other members perceptions. This congruence allows high-power teams to have lower levels of process conict, and therefore better team performance. General discussion Across two studies, we have shown that high-power teams perform worse than low-power teams. In both studies, we nd that

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128 Table 6 Study 2 results of hierarchical regression analysis. Intragroup task conict Step 1 National diversity Team tenure Communication Task routineness Task focus F R2/Adj. R2 Step 2 Task conict Relationship conict Process conict F R2/Adj. R2 DR2 Step 3 Team power Power congruence F R2/Adj. R2 DR2 Step 4 Team power power congruence F R2/Adj. R2 DR2 n = 37. * p < .05. ** p < .01. *** p < .001. .53** .16 2.45* .37/.22 .22 .12 2.11 .38/.20 .00 .49** .02 2.17 .34/.19 .19 .29 2.03 .37/.19 .00 .63*** .02 3.71** .47/.35 .34 .48* 4.20** .55/.42 .07 .36* .16 2.80* .40/.26 .10 .15 2.47* .41/.26 .00 .14 .24 .12 .05 .03 .61 .09/.00 Intragroup relationship conict .10 .26 .12 .13 .26 1.03 .14/.00 Intragroup process conict .30 .18 .09 .05 .08 1.09 .15/.01 Task performance .43* .13 .12 .11 .27 2.41 .28/.16

125

Task performance .43* .13 .12 .11 .27 2.41 .28/.16 .11 .15 .45* 2.98* .46/.31 .15 .29 .11 2.94* .53/.35 .04 .14 2.67* .54/.35 .00

5 4 3 2 1 0 Low High

Team Power
Power Congruence Low Power Congruence High
Fig. 2. Moderating effect of power congruence on the relationship between team power and process conict.

higher levels of process conict in high-power teams can fully explain these effects. One of this papers strengths is showing these ndings in different organizational contexts, including the nancial and telecommunications industries, and on different outcome variables, including daily organizational team performance as well as team performance on a standardized, controlled team decisionmaking task. Our ndings offer hope to managers by showing that when team members have an accurate understanding of how

much power they have within the team, the negative effects of team power may be avoided. These ndings make several contributions to existing research. First, our ndings extend the literature on individual differences in power and on team effectiveness. Previous research has found, for example, that individuals in high-power positions are more responsive in setting priorities and using information than less powerful organizational members (Overbeck & Park, 2006). However, our ndings suggest that when these powerful individuals must work together as a team, they may actually be less effective than low-power teams because they experience high levels of process conict. Such internal team conicts associated with team power may override the cognitive benets of individually held power. We also extend psychological research showing that individually held power decreases perspective taking and understanding of others (e.g., Galinsky, Magee, Inesi, & Gruenfeld, 2006; Keltner & Robinson, 1997), by exploring additional ways in which power may impair interactions among team members. Our ndings also extend research on team effectiveness (for a review, see Kozlowski & Ilgen, 2006) by nding signicant differences in process conict and task performance based on differences in power composition between and within teams. Our paper thus shows the importance of accounting for power as a determinant of team composition when studying team dynamics. Second, our ndings contribute to research on top management teams (TMTs). Though some variance exists in how past researchers have dened TMTs (e.g., Carpenter, Geletkanycz, & Sanders, 2004; Hambrick & Mason, 1984), we believe most conceptualizations can be classied as a form of a high-power team. Indeed, despite many variations in composition, TMTs have remarkably consistent effects on rm outcomes, suggesting they operate mainly through a significant and shared characteristic such as team power (Carpenter et al., 2004). However, the specic role of power in shaping TMT dynamics has not received much investigation (cf. Certo, Lester, Dalton, &

Intragroup Process Conflict

126

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128

Dalton, 2006), nor have the underlying team dynamics or processes of TMTs (cf. Certo et al., 2006; Vyakaranam & Handelberg, 2005). Our ndings suggest TMTs may actually have poorer performance than other organizational teams when TMT members perceptions of their power within the team are not in alignment with their teammates perceptions. These ndings indicate the importance of clarifying individual power positions within TMTs to help them achieve better team performance. Third, our ndings extend research on interpersonal congruence in organizations by showing the importance of interpersonal congruence processes can vary across team contexts. Specically, our ndings suggest power congruence has a larger impact on members of high-compared to low-power teams, perhaps because the addictive nature of power (Mulder, 1977) makes individual intragroup power more salient and more important in high-power team interactions. Past literature has demonstrated that when individuals consider a characteristic important, verication (or the lack thereof) becomes a greater and more inuential concern (Swann & Pelham, 2002). Our results add nuance to this nding by suggesting the team context in which an individual operates, as well as his or her personal feelings, determines a characteristics importance. In addition, our results suggest a characteristics importance moderates consequences not only for individuals but also for teams. Last, our ndings contribute to the intragroup conict literature (e.g., De Dreu, 2007; De Dreu & Weingart, 2003; Jehn & Bendersky, 2003) by looking at an important new antecedent of conict: team power. Our ndings across both studies suggest a teams power in the organization may have substantial impact on the process conict that occurs in the team, and our ndings in Study 2 suggest accurate self-perceptions of power within the team may alleviate this effect. This nding extends past research on the antecedents of intragroup conict, which has primarily focused on demographic characteristics (e.g., Jehn, Northcraft, & Neale, 1999; Pelled, Eisenhardt, & Xin, 1999), by showing that power is a new and important factor of team composition that may inuence intragroup conict. Additionally, we found, in line with a growing amount of evidence (e.g., Greer & Jehn, 2007; Jehn et al., 1999), that process conict negatively affects team performance. Although past research has suggested process conict may benet performance through improving task t (cf. Jehn & Bendersky, 2003), other research has suggested process conict may hurt performance because of the high emotionality associated with issues such as task delegation and its associated personal connotations (Greer & Jehn, 2007). Our ndings inform this debate by showing support for the proposition that process conict negatively affects team performance. We do not nd a strong effect of task conict on objective team performance measures in our studies, though this result is not surprising, as a meta-analysis by De Dreu and Weingart (2003) also did not nd a strong effect of task conict on team performance. Future research would benet from further investigation into the conditions under which task conict is good or bad for team performance. Future directions and limitations The ndings in these studies open up the possibility for several new research directions. The rst derives from our suggestion that one of the primary theoretical mechanisms underlying our effects is that being a member of a high-power team instills individuals with a high sense of power, which sets in play the processes described above and is compounded by the fact that in high-power teams, multiple individuals with similarly high senses of power are interacting. This should occur for all forms of high-power teams, regardless of the type of high-power team (management team, policy-setting team, etc.). We believe this provides a more parsimonious explanation for our effects than structural or func-

tional factors, as high-power teams come in many forms with different internal power structures (e.g., power dispersion vs. shared leadership) and different functional tasks and responsibilities (e.g., policy-making teams vs. management teams). The one factor all forms of high-power teams have in common is their power, or inuence, within the organization. However, future research should further investigate the exact nature and experience of power in high-power teams (from structural effects to psychological effects). A related avenue for future research would be to extend our present focus on teams with high formal power, or hierarchical level in the organization, to teams with other forms of power. Although Mintzberg (1983) suggested formal power may be the most inuential form of power in organizations, examining the generalizability of our ndings to other bases of team power, such as expertise, would still be valuable (French & Raven, 1959). Furthermore, although we feel we can treat our results with condence, as we controlled for a number of variables that could potentially differ between high- and low-power teams (including the task [in Study 2], and variables such as effort, focus, and diversity), alternative explanations could still exist. For example, perhaps high-power teams have less motivation to participate in research studies. Or perhaps high-power teams are more likely to consist of individuals with certain personality types, such as social-dominance orientation, need for power, or Machiavellianism, who are more sensitive to and reactive on power issues than those in low-power teams. Future investigation of high- and low-power teams in a laboratory environment would be valuable, as would research that pulls apart potential covarying factors, such as motivation, status, and personality type. Also, given the organizations we studied, we limited our sample to the comparison of low- and high-power teams (as is common in power research; see Keltner et al. (2003), for a review). Investigation of team power in a laboratory setting where multiple levels of team power could be manipulated more easily would be interesting. One potential way to reduce process conicts in high-power teams is to identify and manage their root causes. One such root cause may be the presence of status contests, or status conicts, among high-power team members (e.g., Bendersky & Hayes, in press; Overbeck et al., 2005). Recent work suggests a struggle for status, or power, is a real and powerful phenomenon in teams (Bendersky & Hayes, in press; Greer & van Kleef, 2010), and one especially likely to occur in high-power teams, given the addictiveness of power (e.g., Mulder, 1977). This may explain why highpower team members, more than low-power team members, engage in process conicts (such as trying to get better assignments than a rival; Bendersky & Hayes, in press; Greer & van Kleef, 2010). Past research has noted that people tend to dress up conicts to hide the underlying issue (e.g., Harinck, De Dreu, & Van Vianen, 2000), and that attempts to gain status must occur within the constraints of group norms and collective task completion (e.g., Chizhik, Alexander, Chizhik, & Goodman, 2003; Ridgeway, 1982; Shelly & Troyer, 2001). For example, to gain status, one might try to express an innovative but divergent perspective on how to accomplish the task. Another member, who wants to also have higher status, might try to discredit that member by disagreeing with the proposal (not because he actually disagrees, but because he does not want others to buy into that members idea). In this way, a status conict can be dressed up as a conict about another issue, such as the process. Future research into the exact ways in which status contests and power struggles may manifest themselves in process conicts, and what this transformation of status and power issues to process conicts implies for team functioning, would be interesting. In addition, investigating the exact mechanisms by which conict impacts team performance would be valuable. Although research has often speculated that negative

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128

127

emotions and impaired task focus underlie the negative effects of conict (cf. De Dreu & Weingart, 2003; Jehn, 1995), scant research has empirically examined these relationships. Relatedly, future research would benet from further examination of factors that can ameliorate the negative relationship between team power and task performance. Although we identied power congruence as one such factor, others are possible, such as cooperative norms within the team. Second, future research would benet from examining ways in which teams can ensure each members perceptions of his or her power within the team is congruent with other members perceptions. Work by Anderson et al. (2006, 2008) has shown that people often err on the side of caution in reporting their own status; our ndings urge caution in allowing incongruent perceptions of power to emerge, as they can cause havoc in high-power teams. Identifying exactly when power (in)congruence is likely to occur would thus be valuable. For example, when team members perceive their teams hierarchy as illegitimate (e.g., Lammers, Galinksky, Gordijn, & Otten, 2008), power incongruence may be more likely to exist. When members do not fully endorse the formal internal hierarchy, they may be more likely to challenge the existing power structure within the team and engage in power moves. Additionally, although power congruence ameliorated the process conicts present in high-power teams, we call for future investigation of other factors that could reduce these negative conicts and enable high-power teams to better capitalize on their potential and outperform low-power teams should be investigated. For example, high-power teams have more resources, and therefore more privileges and opportunities, than other teams. Because of this higher level of prestige, members are likely to value and be proud of their membership, which could lead to higher identication, loyalty, and cohesion within their high-power teamif their potential for destructive process conicts can be overcome. Conclusion High-power teams are not always high performers. High-power teams have high levels of process conict, which can interfere with effective team performance. As a result, low-power teams may outperform high-power teams. However, when high-power team members perceptions of their individual power within the team is in alignment with how other team members perceive their power, high-power teams can avoid process conicts and exhibit better team performance. Both managers and management researchers can use these ndings to better understand how high-power teams (such as TMTs) differ from low-power organizational teams and to identify ways in which high-power team performance can be improved. Acknowledgments We would like to thank Carsten De Dreu, Bart De Jong, Anneloes Raes, and three anonymous reviewers for their helpful comments on earlier drafts of this manuscript. References
Aiken, L., & West, S. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage. Anderson, C., Ames, D. R., & Gosling, S. D. (2008). Punishing hubris: The perils of overestimating ones status in a group. Personality and Social Psychology Bulletin, 34, 90101. Anderson, C., & Berdahl, J. (2002). The experience of power: Examining the effects of power on approach and inhibition tendencies. Journal of Personality and Social Psychology, 83, 13621377. Anderson, C., Srivastava, S., Beer, J. S., Spataro, S. E., & Chatman, J. A. (2006). Knowing your place: Self-perceptions of status in face-to-face groups. Journal of Personality and Social Psychology, 91, 10941110.

Astley, W. G., & Sachdeva, P. S. (1984). Structural sources of intraorganizational power: A theoretical synthesis. Academy of Management Review, 9, 104113. Bales, R. F. (1950). Interaction process analysis: A method for the study of small groups. Chicago: University of Chicago Press. Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology, 51, 11731182. Baumeister, R. F., Smart, L., & Boden, J. M. (1996). Relation of threatened egotism to violence and aggression. Psychological Review, 103, 533. Bell, C., & Song, F. (2005). Emotions in the conict process: An application of the cognitive appraisal model of emotions to conict management. The International Journal of Conict Management, 16(1), 3054. Bendersky, C., & Hayes, N. (in press). Status conict in groups. Organization Science. Berger, J., Ridgeway, C. L., Fisek, M. H., & Norman, R. Z. (1998). The legitimation and delegitimation of power and prestige orders. American Sociological Review, 63, 379405. Berger, J., Rosenholtz, S. J., & Zelditch, M. Jr., (1980). Status organizing processes. Annual Review of Sociology, 6, 479508. Brief, A. P., & Weiss, H. M. (2002). Organizational behavior: Affect in the workplace. Annual Review of Psychology, 53, 279307. Bruins, J. J., & Wilke, H. A. M. (1992). Cognitions and behavior in a hierarchy: Mulders power theory revisited. European Journal of Social Psychology, 22, 2139. Carpenter, M. A., Geletkanycz, M. A., & Sanders, W. G. (2004). Upper echelons research revisited: Antecedents, elements, and consequences of top management team composition. Journal of Management, 30, 749778. Certo, S. T., Lester, R. H., Dalton, C. M., & Dalton, D. R. (2006). Top management teams, strategy, and nancial performance: A meta-analysis. Journal of Management Studies, 43, 813839. Chattopadhyay, P. J., Finn, C., & Ashkanasy, N. M. (2010). Affective responses to professional dissimilarity: A matter of status. Academy of Management Journal, 53, 808826. Chizhik, A. W., Alexander, M. G., Chizhik, E. W., & Goodman, J. A. (2003). The rise and fall of power and prestige orders: Inuence on task structure. Social Psychological Quarterly, 66, 303317. Cohen, S. G., & Bailey, D. E. (1997). What makes teams work: Group effectiveness research from the shop oor to the executive suite. Journal of Management, 23, 239290. Cronbach, L. J. (1955). Processes affecting scores on understanding others and assumed similarity. Psychological Bulletin, 52, 177193. De Cremer, D., & Sedikides, C. (2005). Self-uncertainty and responsiveness to procedural justice. Journal of Experimental Social Psychology, 41, 157173. De Dreu, C. K. W. (2007). The virtue and vice of workplace conict: Food for (pessimistic) thought. Journal of Organizational Behavior, 29, 518. De Dreu, C. K. W., & Weingart, L. R. (2003). Task versus relationship conict, team performance, and team member satisfaction: A meta-analysis. Journal of Applied Psychology, 88, 741749. Eisenhardt, K. M., & Bourgeois, L. J. (1988). Politics of strategic decision making in high-velocity environments: Toward a midrange theory. Academy of Management Journal, 31, 737770. Evan, W. (1965). Conict and performance in R&D organizations. Industrial Management Review, 7, 3746. Fast, N. J., & Chen, S. (2009). When the boss feels inadequate: Power, incompetence, and aggression. Psychological Science, 20, 14061413. Fast, N. J., Gruenfeld, D. H., Sivanathan, N., & Galinsky, A. D. (2009). Illusory control: A generative force behind powers far-reaching effects. Psychological Science, 20, 502508. Fiol, C. M., OConnor, E. J., & Aguinis, H. (2001). All for one and one for all? The development and transfer of power across organizational levels. Academy of Management Review, 28, 224242. Fiske, S. T. (1993). Controlling other people: The impact of power on stereotyping. American Psychologist, 48, 621628. Fombrun, C. J. (1983). Attributions of power across a social network. Human Relations, 36, 493508. French, J., & Raven, B. (1959). The bases of social power. In D. Cartwright (Ed.), Studies in social power (pp. 150167). Ann Arbor, MI: Institute for Social Research. Funder, D. C., & West, S. G. (1993). Viewpoints on personality: Consensus, self other agreement and accuracy in judgments of personality: An introduction. Journal of Personality, 61(4), 457476. Galinsky, A. D., Gruenfeld, D. H., & Magee, J. C. (2003). From power to action. Journal of Personality and Social Psychology, 85, 453466. Galinsky, A. D., Magee, J. C., Inesi, M. E., & Gruenfeld, D. H. (2006). Power and perspective not taken. Psychological Science, 17, 10681074. Georgeson, J., & Harris, M. J. (2006). Holding onto power: Effects of powerholders positional instability and expectancies on interactions with subordinates. European Journal of Social Psychology, 36, 451468. Greer, L. L., & van Kleef, G. A. (2010). Equality versus differentiation: The effects of power dispersion on social interaction. Journal of Applied Psychology, 95, 10321044. Greer, L. L., & Jehn, K. A. (2007). The pivotal role of emotion in intragroup process conict. In M. Neale, E. Mannix, & C. Anderson (Eds.). Research on managing groups and teams (Vol. 10, pp. 2345). Oxford, UK: Elsevier. Guetzkow, H., & Gyr, J. (1954). An analysis of conict in decision making groups. Human Relations, 7, 367381. Guinote, A., Brown, M., & Fiske, S. T. (2006). Minority status decreases sense of control and increases interpretive processing. Social Cognition, 24, 169186.

128

L.L. Greer et al. / Organizational Behavior and Human Decision Processes 116 (2011) 116128 (Eds.). Research on managing groups and teams (Vol. 7, pp. 169199). Oxford: Elsevier. Overbeck, J. R., & Park, B. (2006). Powerful perceivers, powerless objects: Flexibility of powerholders social attention. Organizational Behavior and Human Decision Processes, 99, 227243. Owens, D. A., & Sutton, R. I. (2001). Status contests in meetings: Negotiating the informal order. In M. Turner (Ed.), Groups at work: Advances in theory and research (pp. 299316). Hillsdale, NJ: Lawrence Erlbaum. Pelled, L. H., Eisenhardt, K. M., & Xin, K. R. (1999). Exploring the black box: An analysis of group diversity, conict, and performance. Administrative Science Quarterly, 44, 128. Polzer, J. T., Milton, L. P., & Swann, W. B. (2002). Capitalizing on diversity: Interpersonal congruence in small groups. Administrative Science Quarterly, 47, 296324. Ragins, B. R., & Sundstrom, E. (1989). Gender and power in organizations: A longitudinal perspective. Psychological Bulletin, 105, 5188. Ridgeway, C. (1982). Status in groups: The importance of motivation. American Sociological Review, 47, 7688. Ridgeway, C. L. (1991). The social construction of status value: Gender and other nominal characteristics. Social Forces, 70, 367386. Ridgeway, C. L. (1997). Where do status beliefs come from? New developments. In J. Szmatka, J. Skvoretz, & J. Berger (Eds.), Status, networks, and structure (pp. 137158). Stanford, CA: Stanford University Press. Ridgeway, C. L., & Berger, J. (1986). Expectations, legitimation, and dominance behavior in task groups. American Sociological Review, 51, 603617. Roseman, I. J., Antoniou, A. A., & Jose, P. E. (1996). Appraisal determinants of emotions: Constructing a more accurate and comprehensive theory. Cognition & Emotion, 10(3), 241277. Russell, J. (1978). Evidence of convergent validity on the dimensions of affect. Journal of Personality and Social Psychology, 37, 345356. Shelly, R. K., & Troyer, A. (2001). Emergence and completion of structure in initially undened and partially dened groups. Social Psychological Quarterly, 64, 318322. Sidanius, J. (1993). The psychology of group conict and the dynamics of oppression: A social dominance perspective. In S. Iyengar & W. J. McGuire (Eds.), Exploitation in political psychology (pp. 183219). Durham, NC: Duke University Press. Sidanius, J., & Pratto, F. (1999). Social dominance: An intergroup theory of social hierarchy and oppression. Cambridge, UK: Cambridge University Press. Siegel, P. A., & Hambrick, D. C. (2005). Pay disparities within top management groups: Evidence of harmful effects on performance of high-technology rms. Organization Science, 16, 259274. Simon, H. A. (1976). Administrative behavior: A study of decision-making processes in administrative organizations. New York: Free Press. Smith, P. J., Jost, J. T., & Vijay, R. (2008). Legitimacy crisis? Behavioral approach and inhibition when power differences are left unexplained. Social Justice Research, 21, 358376. Smith-Lovin, L., & Brody, C. (1989). Interruptions in group discussions: The effects of gender and group composition. American Sociological Review, 54, 424435. Stasser, G., & Titus, W. (1985). Pooling of unshared information in group decisionmaking Biased information sampling during discussion. Journal of Personality and Social Psychology, 48, 14671478. Stearns, F. (1972). Anger: Psychology, physiology, and pathology. Springeld, IL: Charles C. Thomas. Swann, W. B. Jr., (1987). Identity negotiation: Where two roads meet. Journal of Personality and Social Psychology, 53, 10381051. Swann, W. B., & Ely, R. J. (1984). A battle of wills: Self-verication versus behavioral conrmation. Journal of Personality and Social Psychology, 46, 12871302. Swann, W. B., Jr., & Pelham, B. W. (2002). Who wants out when the going gets good? Psychological investment and preference for self-verifying college roommates. Journal of Self and Identity, 1, 219233. Thomas, K. W. (1992). Conict and negotiation processes in organizations. In M. D. Dunnette & L. Hough (Eds.), Handbook of industrial and organizational psychology (pp. 889935). Palo Alto, CA: Consulting Psychologists Press. Van den Bos, K., & Lind, E. A. (2002). Uncertainty management by means of fairness judgments. In M. P. Zanna (Ed.). Advances in experimental social psychology (Vol. 34, pp. 160). San Diego, CA: Academic Press. Van Kleef, G. A., Oveis, C., Van der Lwe, I., LuoKogan, A., Goetz, J., & Keltner, D. (2008). Power, distress, and compassion: Turning a blind eye to the suffering of others. Psychological Science, 19, 13151322. Vyakaranam, S., & Handelberg, J. (2005). Four themes of the impact of management teams on organizational performance. International Small Business Journal, 23, 236256. Wall, J. A., & Callister, R. R. (1995). Conict and its management. Journal of Management, 21, 515558. Wegener, B. (1992). Concepts and measurement of prestige. Annual Review of Sociology, 18, 253280.

Guinote, A., Judd, C. M., & Brauer, M. (2002). Effects of power on perceived and objective group variability: Evidence that more powerful groups are more variable. Journal of Personality and Social Psychology, 82, 708721. Hambrick, D. C., & Mason, P. (1984). Upper echelons: The organization as a reection of its top managers. Academy of Management Review, 9, 193206. Harinck, F., De Dreu, C. K. W., & Van Vianen, A. E. M. (2000). The impact of conict issue on xed-pie perceptions, problem solving, and integrative outcomes in negotiation. Organizational Behavior and Human Decision Processes, 81, 329358. Henderson, A. D., & Fredickson, J. W. (1996). Information-processing demands as a determinant of CEO compensation. Academy of Management Journal, 39, 575606. House, R. J., Rousseau, R. M., & Thomas-Hunt, M. (1995). The meso paradigm: A framework for the integration of micro and macro organizational behavior. In L. L. Cummings & B. M. Staw (Eds.). Research in organizational behavior (Vol. 17, pp. 71114). Greenwich, CT: JAI Press. Janis, I. L. (1982). Victims of groupthink (2nd ed.). Boston: Houghton Mifin. Jehn, K. A. (1995). A multimethod examination of the benets and detriments of intragroup conict. Administrative Science Quarterly, 40, 256282. Jehn, K. A. (1997). Qualitative analysis of conict types and dimensions in organizational groups. Administrative Science Quarterly, 42, 530557. Jehn, K. A., & Bendersky, C. (2003). Intragroup conict in organizations: A contingency perspective. Research in Organizational Behavior, 25, 189244. Jehn, K. A., & Mannix, E. A. (2001). The dynamic nature of conict: A longitudinal study of intragroup conict and group performance. Academy of Management Journal, 44, 238251. Kahneman, D., Slovic, P., & Tversky, A. (Eds.). (1982). Judgment under uncertainty: Heuristics and biases. New York: Cambridge University Press. Jehn, K. A., Northcraft, G., & Neale, M. A. (1999). Why differences make a difference: A eld study of diversity, conict, and performance in workgroups. Administrative Science Quarterly, 44, 741763. Keltner, D., van Kleef, G. A., Chen, S., & Kraus, M. W. (2008). A reciprocal inuence model of social power: Emerging principles and lines of inquiry. Advances in Experimental Social Psychology, 40, 151192. Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach, and inhibition. Psychological Review, 110, 265284. Keltner, D., & Robinson, R. J. (1997). Defending the status quo: Power and bias in social conict. Personality and Social Psychology Bulletin, 23, 10661077. Keltner, D., Young, R. C., Heerey, E. A., Oemig, C., & Monarch, N. D. (1998). Teasing in hierarchical and intimate relationships. Journal of Personality and Social Psychology, 75, 12311247. Klein, K. J., & Kozlowski, S. W. J. (2000). From micro to meso: Critical steps in conceptualizing and conducting multilevel research. Organizational Research Methods, 3, 211236. Kozlowski, S. W. J., & Ilgen, D. R. (2006). Enhancing the effectiveness of work groups and teams. Psychological Science in the Public Interest, 7(3), 77124. LaFollette, W., Hornsby, J. S., Smith, B. N., & Novak, W. I. (1996). The use of work teams in organizations. Mid-American Journal of Business, 11, 5561. Lammers, J., Galinksky, A. D., Gordijn, E. H., & Otten, S. (2008). Illegitimacy moderates the effect of power on approach. Psychological Science, 19, 558564. Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford University Press. Lazear, E. P., & Rosen, S. (1981). Rank-order tournaments as optimum labor contracts. Journal of Political Economy, 89, 841864. LeBreton, J. M., & Senter, J. L. (2008). Answers to 20 questions about interrater reliability and interrater agreement. Organizational Research Methods, 11, 815852. Magee, J. C., & Galinsky, A. D. (2008). Social hierarchy: The self-reinforcing nature of power and status. Academy of Management Annals, 2, 351398. Mannix, E. A., & Sauer, S. J. (2006). Status and power in organizational group research: acknowledging the pervasiveness of hierarchy. In E. Lawler & S. Thye (Eds.). Advances in group processes: Social psychology of the workplace (Vol. 23, pp. 49182). New York: Elsevier. Martorana, P. V., Galinsky, A. D., & Rao, H. (2005). From system justication to system condemnation: Antecedents of attempts to change power hierarchies. In M. A. Neale, E. A. Mannix, & M. Thomas-Hunt (Eds.). Research on managing groups and teams (Vol. 7, pp. 283313). Oxford: Elsevier. McGrath, J. E. (1984). Groups: Interaction and performance. Englewood Cliffs, NJ: Prentice Hall. Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chicago Press. Mintzberg, H. (1983). Power in and around organizations. Englewood Cliffs, NJ: Prentice Hall. Morrison, K. R., Fast, N. J., & Ybarra, O. (2009). Group status, perceptions of threat, and support for social inequality. Journal of Experimental Social Psychology, 45, 204210. Mulder, M. (1977). The daily power game. Leiden, Netherlands: H.E. Stenfert Kroese. Overbeck, J. R., Correll, J., & Park, B. (2005). Internal status sorting in groups: The problem of too many stars. In M. A. Neale, E. A. Mannix, & M. Thomas-Hunt