The Downside of Looking Like a Leader 1
Running Head: LEADER CONFIDENCE ANDFOLLOWER PARTICIPATION
The Downside of Looking Like a Leader:
How a Leader’s Confident Nonverbal Demeanor Can Stifle Follower Participation
Connson C. Locke
University of California, Berkeley
Cameron Anderson
University of California, Berkeley
Author Note
Connson C. Lockeand Cameron Anderson, HaasSchool of Business, University of California, Berkeley.
Connson C. Locke is now at the Department of Management, London School of Economics, United Kingdom.
This research was supported in part by the Schwabacher Dissertation Fellowship, UC Berkeley Institute for Research on Labor and Employment, and Haas X-Lab. We gratefully acknowledge the invaluable feedback received from James R. Detert, Dacher Keltner, Laura J. Kray, Daan Stam, and Barry Staw.
This article is based on the first author’s doctoral dissertation, submitted to theUniversity of California, Berkeley. An earlier version was presented at the 70th annual meeting of the Academy of Management in Montreal, Canada, August 2010.
Correspondence concerning this article should be addressed to Connson C. Locke, Department of Management, London School of Economics, Houghton Street, London,WC2A 2AE, United Kingdom. E-mail: .
Abstract
Leaders can appear more competentby exhibiting a confident nonverbal demeanor. However, the current investigation reveals that a leader’s confident demeanor can also have a negative consequence: the stifling of follower participation. In three laboratory studies, we found that the more confidence the leader exhibited nonverbally, the less followersparticipated in joint discussions. Furthermore, followers deferred to leaders who exhibited a confident demeanor even when leaders made incorrect decisions. The mechanism for this stifling effect depended on the context. When follower participation was not invited by the leader, the leader’s confidentdemeanor stifled participationby makingthe leader appear more threatening. In contrast, when follower participation was invited,the leader’s confident demeanor stifled participation by making the leader appear more competent.
The Downside of Looking Like a Leader:
How a Leader’s Confident Nonverbal Demeanor Can Stifle Follower Participation
Leaders reap many benefits from displaying confidentnonverbal behaviors such as expanded posture, frequent and direct eye contact, and confident tone of voice. For example, confident leaders are afforded more influence andjudged to be more competent, charismatic, and effective (Awamleh Gardner, 1999; Driskell, Olmstead, Salas, 1993; Ridgeway, 1987). Indeed, leaders are encouraged to adopt a confident nonverbal demeanor in order to be more effective and enhance their image (e.g., Bates, 2005; Fritz, Brown, Lunde, & Banset, 2005; Howell Costley, 2006).
While displaying confident nonverbal behavior might make a leader appear more competent, we argue that there may also be an unexpected downside. Specifically, we propose that leaders who display a confident nonverbal demeanor can stifle their followers’ participation, leading to poorer communication and suboptimal decisions. We tested this hypothesis in three laboratory studies thatexamineddyadic interactions between leaders and followers. Specifically, we usedvideo recordings of the interactions to code leaders’ nonverbal behavior, followers’ reactions and participation, and examine the mechanisms underlying this phenomenon.
Nonverbal Behaviorand Leadership
Two sets of findings have linked leadership to nonverbal displays of confidence. First, research has shown that confident nonverbal behaviors help individuals emerge as a leader in groups (Driskell et al., 1993; Ridgeway, 1987). Individuals who exhibit a confident demeanor tend to beconsidered more skilled, have better leadership ability, and are given more influence and leadership responsibility in joint decisions than those who do not display confidence (Anderson Kilduff, 2009; Carli, LaFleur, & Loeber, 1995; Driskell et al., 1993; Ridgeway, 1987).
Second, research has shown that for individuals who occupy positions of leadership, confident nonverbal behaviors help them appear more competent to followers. Confident nonverbal behavior enhances perceivers’ judgments of leaders’ charisma and effectiveness – sometimes even more than the content of leaders’ speech or their actual performance (Awamleh Gardner, 1999; Holladay Coombs, 1993, 1994). Therefore, confidence is widely viewed as a core ingredient of effective leadership (Hannah, Avolio, Luthans, & Harms, 2008; Luthans, Luthans, Hodgetts, & Luthans, 2001; Peterson, Balthazard, Waldman, & Thatcher, 2008).
In the current research, we build from the latter set of findings, focusing on the nonverbal display of confidence shown by individuals in leadership positions. Specifically, we extend prior work by proposing that there may be an unexpected downside to the leader’s display of a confident nonverbal demeanor – namely, that it might stifle follower participation and hamper joint decision making.
The term “leader” has been defined in a multitude of ways (Bass, 2008). We use the term “leader” to refer to individuals who occupy supervisory positions and “follower” to refer to the persons they supervise (e.g., Hollander, 1992; Kellerman, 2008; Yukl, 2006). Moreover, our focus is on nonverbal displays of confidence, such as those that imply task ability (e.g., confident tone of voice, straight posture, direct eye contact, confident gestures;Driskell, et al., 1993; Ridgeway, 1987; Ridgeway, Berger, & Smith, 1985). These have been differentiated from dominance cues (e.g., commanding tone of voice, forward looming posture, intrusive gestures), which imply “control through threat”(Driskell et al., 1993,p. 51). While the display of confidence has been shown to produce perceptions of competence, dominance cues have not (Carli et al., 1995; Driskell et al., 1993; Ridgeway, 1987).
Leader Confident Nonverbal Demeanor and Follower Participation
We propose that leaders who display a more confident nonverbal demeanor will stifle follower participation for at least two reasons. First, a leader’s confident demeanor mayincrease feelings of threat among followers. Prior work has shown that fear and feelings of threat are primary reasons why followers inhibit themselves and fail to express their ideas or opinions (Edmondson, 1999; Kish-Gephart, Detert, Trevino, & Edmondson, 2009). Followers most frequently cite the fear of negative consequences as the reason for not raising an important issue to their leaders (e.g., Milliken, Morrison, Hewlin, 2003; Ryan Oestreich, 1991).
Leaders who convey more confidence might signal to followers they are more steadfast in their own opinions and thus less open to follower input (Staw & Ross, 1980). Being in a position of followership increases sensitivity to threat and punishment, in particular the threat of losing favor with leaders (Keltner, Gruenfeld, Anderson, 2003). Therefore, followers might be less likely to voice their opinions and ideas when the leader expresses more confidence because they are threatened by the possibility of damaging their relationship with the leader.
Second, a leader’s confident demeanor might stifle voice precisely because it makes the leader appear more competent. Research on status characteristics theory has found that in task groups, members grant higher status and influence to those individuals whom they believe to have superior capabilities (Berger, Cohen, Zelditch, 1972). Individuals who believe they possess relatively lower competence inhibit their own contributions and afford others greater influence to increase the group’s chances of success (Anderson, Srivastava, Beer, Spataro, & Chatman, 2006; Berger et al., 1972; Driskell & Mullen, 1990).
As mentioned earlier, when leaders display a confident nonverbal demeanor they appear more competent (e.g., Awamleh & Gardner, 1999; Carli et al., 1995). By appearing more competent, therefore, leaders might unintentionally lead followers to contribute less. Followers might participate less and afford more control to the leader so that the group has the best chance of succeeding. Indeed, such deference would be seen as functional because it would seem to place more weight on the judgments of those with the greatest perceived expertise (Anderson & Brown, 2010).
Prior work on interpersonal complementarity (Horowitz et al., 2006; Kiesler, 1983) provides indirect evidence that leader confidence stifles follower participation. This work primarily examines behavioral units comprised of verbal and nonverbal elements (e.g., Strong et al., 1988; Dryer & Horowitz, 1997), but some research has focused specifically on nonverbal behavior(TiedensFragale, 2003). In that research, individuals who displayed stronger nonverbal behavior (e.g., expanding arms and legs away from the body) causedtheir discussion partners to adopt weaker nonverbal behavior (i.e., pulling arms and legs closer to the body; TiedensFragale, 2003). We extend this research in part by focusing on how leaders’ nonverbal behavior can impact followers’ participation in joint decisions, and in turn, the quality of those decisions.
We conducted three laboratory studies. In Study 1, pairs of participants engaged in a joint decision-making task where one person was randomly assigned to be the discussion leader. The leader’s nonverbal behavior was measured through video coding and was expected to affecthow much the follower participated. Studies 2 and 3 established causality by using a confederate as the leader and manipulating his nonverbal style. Study 3 tested the effects of the leader’s demeanor on follower deference, or concession in the joint decision. Both Studies 2 and 3 also examined the mechanisms underlying the relationship between leader demeanor and follower participation.
Study 1
This study was designed to test our central hypothesis – that a confident demeanor displayed by leaders will stifle follower participation. Participants were randomly paired into dyads to work on a decision-making task; within the dyadthey were randomly assigned to the role of leader (supervisor) or follower (subordinate). We videotaped dyads’ deliberations to allow for behavior coding.
Method
Participants. Participants were 86 undergraduates (57 female, 29 male) who were paid $15. Participants formed 43 dyads and were 20 years old on average (SD = 1.61).
Procedures. Two participants were scheduled for each laboratory session. Upon arrival they were told the study involved decision making using a business simulation task. The participants were assigned to the roles of supervisor and subordinate. According to previous research and theoretical conceptions, individuals in positions of leadership tend to have legitimacy, control over resources, and evaluative power (Bass, 2008; French & Raven, 1959; Yukl, 2006).Therefore, similar to previous research (Johnson, 1993, 1994), we gave the supervisor legitimacy by telling both participants their role assignments were based on their work history, which had been measured several days earlier in an online survey they both completed; in fact they were randomly assigned. We also gave the supervisor control over resources and evaluative power(Herold, 1977) by telling both participants they would be entered into a lottery for a $50 cash prize and the supervisor would determine how much of the prize the subordinate would receive based on the supervisor’s evaluation at the end of the session. In contrast, the subordinate would not evaluate the supervisor or determine how much of the prize the supervisor would receive.
To prepare for the decision-making task (adapted from Johnson, 1993, 1994), the supervisor read two memos from the “Head Office” that described an organization facing problems and outlined possible solutions to those problems. At the same time, to further establish the supervisor and subordinate roles, the subordinate performed a clerical task that was briefly checked by the supervisor to ensure it was completed correctly. The supervisor then engaged the subordinate in a videotaped 15-minute discussion with the goal of reaching consensus on the solutions. After the discussion, as a leader/follower role manipulation check, participants rated the degree to which they assumed the role of the leader in the discussion. Participants were then fully debriefed and thanked.
Leader/follower role manipulation checks. After the dyad’s discussion, participants privately rated two statements measuring the degree to which they assumed the role of leader: “I led our discussion,” and “I was in control during the discussion.” Each item was rated on a scale of 1 (strongly disagree) to 7 (strongly agree). These items correlated together (α = .84) and were thus combined. Participants also indicated which role they had played, supervisor or subordinate.
Supervisor confident demeanor. The confidence of supervisors’ demeanor was coded from videotape of the discussions. Based on previous research on nonverbal behavior, we wanted to sample from three different nonverbal channels (eyes, body, vocal qualities) and chose the following task cues: frequency of eye contact, upright posture, and vocal loudness (Awamleh& Gardner, 1999; Driskell et al., 1993; Ridgeway, 1987). Eye contact reflected the amount of time participants looked directly at their partner. Consistent with previous research, it was measured in seconds for the entire discussion and then divided by the total time the dyad spent working together (Edinger Patterson, 1983;Murphy, 2007). As in previous research (Murphy, 2007), posture was rated on a scale from 1 (slumped) to 7 (straight) and loudness from 1 (quiet) to 7 (loud).In addition, participants were rated on how confident their nonverbal style was overall from 1 (uncertain, weak) to 7 (confident, strong). This molar measure was included to capture any other potentially relevant nonverbal behaviors in addition to eye contact, posture, and loudness.
The “thin slices” literature has shown that short portions of nonverbal behavior have predictive utility comparable to lengthier slices (Ambady Rosenthal, 1992; Murphy, 2005). Furthermore, leaders’ behavior in the beginning of joint deliberations has more impact on the interpersonal dynamic than their later behavior (Bass, 2008; Shaw, 1961). Therefore, ratings for posture, loudness, and overall nonverbal style were based on a one-minute slice taken from thebeginning of each interaction. Because the rate of eye contact was low and sometimes did not occur in the first minute (participants gazed at the memos while talking), eye contact was measured for the entire discussion.
To establish inter-coder reliabilitya second coder watched 20% (10) of the dyads. The two coders agreed in their judgments of posture (r = .77), loudness (r = .70), nonverbal power (r = .80), and eye contact (r = .92). These four items also correlated with each other (α = .70). That is, some supervisors tended to give more eye contact, sit up straighter, speak louder, and generally convey more confident nonverbal behaviors than others. Therefore, we standardized and combined these four measures to create one aggregate index of supervisor confident demeanor.
Subordinate participation.Based on previous research (e.g., Johnson, 1993; Leffler, Gillespie, Conaty, 1982), we used speaking time as the measure of subordinate participation. Dyads varied in the total amount of time they worked together thus creating variation in total dyad speaking time (M = 10.55, SD = 3.80), making it necessary to convert subordinate speaking time to a percentage of the total speaking time (otherwise some subordinates would score higher on this measure simply because their dyad worked longer on the task). Analyses were run with both percentages of speaking time and arcsine transformation of percentages.1The arcsine transformation did not substantively change the results; therefore percentages of speaking time are reported.
Results
Tests of the role manipulation suggest that it was successful. First, all participants accurately recalled the role they had played. Further, using a repeated measures ANOVA with the dyad as the unit of analysis and role (supervisor, subordinate) as the within-dyad factor, we found that supervisors (M = 5.17, SD = 1.07) reported behaving as a leader more than did subordinates (M = 3.42, SD = 1.273), F(1, 42) = 34.76, p < .001, η2 = .45.
As expected, subordinates (M = 36.40, SD = 15.03) spoke a significantly smaller share of the time than supervisors (M = 63.60, SD = 15.03), F(1, 42) = 35.21, p < .001, η2 = .46. And supervisors (M = .17, SD = .73) exhibited a more confident demeanor than did subordinates2 (M = -.19, SD = .54), F(1, 39) = 11.36, p = .002, η2 = .23.
To test the hypothesis that the supervisor’s confident demeanor stifled subordinate participation, we regressed the subordinate’s participation on the confidence of the supervisor’s demeanor ( = -.41, B = -8.08, SE = 2.84, p = .007, R2 = .17). This analysis indicates that the more confident the supervisor’s demeanor, the less the subordinate spoke and participated.
To test the robustness of this finding, we examined the results by supervisor sex, as prior work has shown differences in the way subordinates respond to female and male leaders (see Eagly, Makhijani, Klonsky, 1992 for a meta-analysis). First, subordinate participation did not differ when the supervisor was female (M = 38.47, SD = 14.43) versus male (M = 32.11, SD = 15.89), t(41) = 1.31, p = .20. Further, we conducted a moderated multiple regression (Aiken & West, 1991). The coefficients were = -.11 (B = -2.16, SE = 8.69, p = .81) for supervisor confident demeanor, = -.11 (B = -3.39, SE = 4.95, p= .50) for supervisor sex, and = -.30 (B = -4.25, SE = 6.32, p = .51) for the interaction term. Because the interaction term was not significant, supervisor sex did not moderate the effect of supervisor demeanor on subordinate participation. This suggests the more confidence supervisors exhibited nonverbally, the less subordinates participated, no matter the supervisors’ sex.
Discussion
Consistent with previous research, leaders exhibited more confident nonverbal behaviors than followers, and followers spoke less than leaders. Furthermore, this study demonstrated that the more the leader used a confident demeanor, the less the follower participated in the decision-making discussion – thus supporting our central hypothesis. Finally, the findings held regardless of the sex of the leader. Thus, when both female and male leaders exhibited more confidence nonverbally, followersparticipated less.
Study 2
After testing the basic phenomenon of interest in Study 1, in Study 2 we sought to establish the causal priority of leaders’ confident demeanor and to examine the mediating mechanisms underlying the effect. To establish causality, we used a research confederate as the leader and systematically varied his demeanor across conditions. Based on prior research on confident nonverbal demeanor, we used a pre-recorded confederate as the leader (Awamleh & Gardner, 1999; Carli et. al., 1995; Driskell et. al., 1993; Ridgeway, 1987), which afforded control over the leader’s nonverbal behavior and consistency in that behavior across dyads.
In line with the arguments outlined in the Introduction, we examined two possible mediators: perceived supervisor threat and perceived supervisor competence. That is, do leaders who display more confidence nonverbally stifle follower voice because they instill perceived threat in their followers, because they appear more competent to followers, or both?
Method
Participants. Participants were 92 undergraduates who participated in partial fulfillment of an organizational behavior course requirement. Five participants were excluded from the analysis when they entered the decision-making discussion having chosen the same candidate as the confederate, making any discussion unnecessary. Seven other participants suspected the supervisor was pre-recorded and were also excluded. Excluded participants came from both conditions (5 from confident demeanor; 7 from less confident demeanor). This left80 participants (43 male, 37 female), who were 21 years old (SD = 3.11) on average.