Group Dynamics: Theory, Research, and Practice 2005, Vol. 9, No. 2, 93–103

Copyright 2005 by the Educational Publishing Foundation 1089-2699/05/$12.00 DOI: 10.1037/1089-2699.9.2.93

Power, Voice, and Hierarchy: Exploring the Antecedents of Speaking Up in Groups Gazi Islam and Michael J. Zyphur Tulane University This study examined the effects of interpersonal power and social dominance orientation on individuals’ propensities to voice opinions in a group task. It was proposed that individuals high in power would tend to voice their opinions to a greater extent than individuals with less power and that this effect would increase with higher levels of social dominance orientation, that is, the degree to which individuals believe that social hierarchies are justified. This relationship was tested using moderated multiple regression. Significant results were found for the interactive relationship between power and social dominance orientation on voice. Implications and future directions are discussed.

The implications of allowing group members to participate in group decision-making contexts have been a major topic in organizational behavior (e.g., Locke & Schweiger, 1979). Participation has been linked to satisfaction (Locke & Schweiger, 1979), goal commitment (Locke, Shaw, Saari, & Latham, 1981), superior customer service (Kizilos, Thomas, & Strickstein, 1994), and ultimately, job performance (Cotton, Vollrath, Froggatt, Lengnick-Hall, & Jennings, 1988; for a dissenting view, see Wagner, 1994). However, the formal right to participation is often not enough to secure effective, democratic group decisions. Several theories of group behavior, from Janis’s (1972) groupthink to threat rigidity (Staw, Sandelands, & Dutton, 1981), have shown that forces come into play that often limit the amount of input that is displayed in group situations. According to Bedeian (2002), leaders are often isolated from the day-to-day realities of their organization owing to employees’ unwillingness to speak candidly. What is at stake in these situations is the expression of

voice, the behavioral corollary of the formal right to participate. The driving question behind the expression (or nonexpression) of voice amounts to the following: When people have the ability to voice their opinions and have something to gain from voicing their opinions, why do we so often see that they refrain from doing so?

Predicting Voice Organizational voice is a phenomenon in which people express opinions or thoughts in order to make some sort of organizational change (Hirschman, 1970). LePine and Van Dyne (1998) suggested that this definition allows a distinction to be drawn between voice and other, related constructs. For example, complaining behavior differs from voice in that no change is promoted by the behavior, and affiliative behavior differs in that the focus of the behavior is on building relationships rather than promoting change. As opposed to complaining and affiliative behavior, voice behavior may or may not be critical and may be either helpful or taxing on relationships (LePine & Van Dyne, 1998). Voice is verbal activity that is used to promote constructive change in the status quo (Rusbult, Farrell, Rogers, & Mainous, 1988). As it stands to reason that voicing constructive concerns and ideas is key to participating in organizational life (Bedeian, 2002), it would seem that the effectiveness of organizational participation interventions would hinge on whether employees actually participated in decision making. Since the 1970s, the usefulness

Gazi Islam, A. B. Freeman School of Business, Tulane University; Michael J. Zyphur, Department of Psychology, Tulane University. We thank Robert Folger for his help with the ideas contained in this study. We also thank Mathilda du Toit and Stephen du Toit from Scientific Software International for their gracious help with this project. Without their advice, the current work would not have been possible. Correspondence concerning this article should be addressed to Gazi Islam, 706 A. B. Freeman School of Business, Tulane University, New Orleans, LA 70118. E-mail: [email protected] 93

94

ISLAM AND ZYPHUR

of participation interventions has been debated (e.g., Dachler & Wilpert, 1978; Locke & Schweiger, 1979). Even with advances in metaanalytic procedures, the evidence is still mixed as to the performance benefits of participation (cf. Cotton et al., 1988; Wagner, 1994). One reason for the often incongruent results may be that studies tend to focus on whether an intervention has been implemented, not whether the members of the organization are actually voicing their opinions more as a result of the intervention. Similarly, studies of voice have often conceptualized voice as the existence of formal grievance systems, without examining the informal processes that occur in groups to assess the degree to which individuals are engaging in voice or not (Olson-Buchanan & Boswell, 2002). However, many have argued that the opportunity to speak may not necessarily imply the presence of an open dialogue within the organization but may instead lead to hiding true opinions (Liden & Mitchell, 1988). Given the inconsistent effectiveness of participation interventions and systems, it may be worth looking more closely at the specific interpersonal variables that make up group participation activities, such as voice behavior. Much research has addressed an individual’s voice in organizational settings, describing the various means of voicing concerns with the ultimate goal of improving organizational outcomes (e.g., Hirschman, 1970; Withey & Cooper, 1989). The majority of research that has examined voice in the last several decades, however, has focused on the consequences of voice. For example, voice has been linked with creativity (Zhou & George, 2001) and harmonious group functioning (Dyck & Starke, 1999). This focus on consequences, however, leaves a large gap in our knowledge of the antecedents of voicing opinions and ideas (LePine & Van Dyne, 1998). The model of voice we propose in the current study attempts to fill this empirical lacuna by examining the processes that lead to voice. We begin with the assumption that voice behavior is motivated by an individual’s desire to constructively change a situation (LePine & Van Dyne, 1998). For a verbal expression to count as voice, it must be perceived by the individual as promoting such a change and therefore as desirable. This conception of voice clarifies the theoretical problem of why people would not voice

changes they see as desirable. In other words, the problem to be addressed by this study concerns why people who see changes as desirable would refrain from providing a voice for such changes. We approach this problem by noting that voice behavior, in addition to benefits, may incur hidden or overt costs and risks for the agent or agents involved (Cortina & Magley, 2003; Rusbult et al., 1988). The hesitance of a person to voice his or her opinion is seen here as a consequence of the “riskiness” of voicing a potentially dissenting opinion (cf. LePine & Van Dyne, 1998). Thus, it stands to reason that voice behavior could be predicted by elucidating variables that increase or decrease the riskiness of such behavior. We begin by examining the variable of power.

The Consequences of Power Power is an element in virtually all social interactions (Fiske, 1993). Researchers in group behavior have shown how group members tend to position each other in terms of status and how position in groups is intricately linked to the perception of psychological wellness (e.g., Goodwin, Operario, & Fiske, 1998; Tajfel & Turner, 1986). Power has been defined in various ways in the literature (Keltner, Gruenfeld, & Anderson, 2000), but many treatments tend to single out either control of resources, outcomes, and behavior (e.g., Keltner et al., 2000; Salancik & Pfeffer, 1977) or interpersonal–social status and dominance (e.g., Pratto, Sidanius, Stallworth, & Malle, 1994; Tiedens, Ellsworth, & Mesquita, 2000). Although the question of whether to regard power as control or as status is an important one, studies of power often use reputational measures, thus making it difficult to separate the two (see Keltner et al., 2000, for a review). In this study, we manipulate positional power within an organization, which is assumed to tap both resource control and status dimensions. We hold that power relations may affect voice by changing the interpersonal processes that result in voice behaviors. For instance, empirical research has shown that people’s propensity to freely interact in a workplace setting is heavily influenced by power relations. Keltner et al. (2000) described a variety of findings linking power to approach-related processes. For example, Anderson and Berdahl (2002)

POWER, VOICE, AND HIERARCHY

found that power correlated with extraversion, and Coats and Feldman (1996) linked power with social skills. An explanation given by Keltner et al. is that people in power have rewardrich environments that are not hindered by external consequences; such environments provide repercussion-free opportunities to voice opinions. If holding power increases the likelihood of activating a person’s approach system, then it would stand to reason that the powerful person would be more likely to engage in voice. In addition, a large amount of psychological research has examined what Kipnis (1976) referred to as the “metamorphic effects of power,” a phrase that describes how power holders tend to “become puffed up with their own importance” (Kipnis, 1984, p. 30), pushing their opinions over those of others. Not surprisingly, power has been negatively correlated with interpersonal perceptiveness (Anderson & Berdahl, 2002). Keltner et al. (2000) concluded that individuals with more power tend to pay less attention to others’ views and psychological states. In addition, Fiske (1993) showed that high-power individuals tend to stereotype others to a greater extent than low-power individuals, using more heuristics and ignoring individuating information about people. Less attentiveness to the states of others means that a person might be less aware of opposing viewpoints when making a statement. Thus, the perceived riskiness of voicing one’s opinion should be low for people with higher power, because of both the lower objective riskiness (i.e., high power means less external consequences) and the lack of subjective ability to perceive opposition (Fiske, 1993). Finally, the propensity toward greater voice behavior has been empirically linked to higher levels of satisfaction in organizations (Rusbult et al., 1988). Rusbult and associates found that people who felt overall satisfaction with their organization were more likely to engage in voice, and theorized that general satisfaction may be a key predictor of voice. In addition to satisfaction judgments, of which affect is a component, high and low power have been correlated with positive and negative emotions, respectively, in a variety of situations (Hecht & LaFrance, 1998; Watson & Clark, 1997). Also, as mentioned above, high-power individuals may have a more resource-rich environment and thus have more reason to be satisfied with their

95

organization. This provides a rationale for the view that power may have some effect on voice via the positive affect associated with power. Hypothesis 1: Power will be positively correlated with voice such that high-power individuals will express more voice than low-power individuals.

Social Dominance Orientation In the last section, objective correlates of power, such as lack of consequences and less sophisticated perceptions of others’ psychological states, were identified. However, the presence of power in a group can also create group pressures associated with dominance. In particular, studies of power and authority have identified people’s hesitance to contradict figures of authority (e.g., Kramer, 1998; Milgram, 1983), as well as retaliatory behaviors that are directed at individuals who do voice their contrary opinions (e.g., Cortina & Magley, 2003). Independent of the objective rewards that await highversus low-power individuals, there may be an apprehension on the part of low-power individuals to voice their own opinions when confronted by dominant individuals. Literature on authority (e.g., Milgram, 1983) has highlighted people’s reluctance to voice opinions that run contrary to those in authority positions; at the same time, willingness to “take a stand” is often necessary in order to contribute in a work situation (Staw & Boettger, 1990). As such, the extent to which people feel that other people or the group as a whole are imbued with authority should hinder their likelihood of coming forth with views of dissent. This effect may be interpreted in the light of an individual’s social dominance orientation (SDO). Both Sidanius’s and Pratto’s explorations of social dominance concern a cognitive state in which one places people or groups as hierarchically dominant or subordinate with respect to each other (Pratto et al., 1994; Sidanius, 1993). A person who is high in SDO feels that such social hierarchies are just and that some people deserve to be in positions of authority over others. In contrast, a person low in SDO sees the world through a more egalitarian lens, tending not to believe in the legitimacy of hierarchies. Thus, the higher amount of SDO a person holds, the more influence power should exert in decid-

96

ISLAM AND ZYPHUR

ing whether to dissent from a group. People who are high in SDO, when put into a power position, should feel the power of their authority to a greater extent than people who are low in SDO; conversely, those who lack power should feel the oppressive force of authority much more if they are high in SDO. It thus stands to reason that the effects of power should be amplified in those who are high in SDO: Hypothesis 2: SDO will moderate the relationship between power and voice, such that the correlation between power and voice will be attenuated as a person’s SDO decreases.

Summary Our model postulates that SDO and power will have an interactive effect on voice behavior. From previous studies of power’s effect on voice, as well as treatments of SDO that show dominance attitudes to be central to how people react to positions of authority, we conclude that voice behavior will be greatest when highpower individuals show high levels of SDO and lowest when low-power individuals show high SDO. By contrast, power position should matter less for individuals who are low in SDO (i.e., individuals who have more egalitarian attitudes). In the next section, we empirically test this model in an experimental group setting.

Method Participants The participants for this study consisted of 121 university undergraduates recruited from psychology courses; they received extra credit for their participation in the study. Participants were randomly assigned to groups ranging in size from 3 to 6 (specifically, there were 12 groups of 3 participants, 6 groups of 4 participants, 5 groups of 5 participants, and 6 groups of 6 participants), so that the groups would be small enough to allow participation by individual members while allowing the possibility of social pressure and power dynamics to emerge.

Procedure Participants were presented with a business simulation involving a hypothetical corporation,

GazCorp. According to the organizational chart presented to participants, there were six possible positions, arranged hierarchically, in the GazCorp system. These positions continuously varied from CEO, the highest position, to line worker, the lowest position. Participants were told that each group member would be randomly assigned to one of the six positions. In actuality, participants were randomly assigned to either the highest or lowest positions, with equal numbers of high and low power assignment across all groups and within all groups of a given size (in odd-numbered groups, numbers of high- and low-power individuals did not differ by more than 1). All group members were asked not to divulge their own position, and in no case did anyone do so. Thus, each person believed that he or she was either the highest or lowest ranking member of the simulated organization (i.e., high or low power). The group then engaged in a discussion task, which involved coming to a decision about the 5-year plan for GazCorp. The task was semistructured, with several open-ended questions that the group had to answer. Participants were also told that, on the basis of their participation in the task, they would be able to receive and give rewards in the form of raffle tickets, for a $50 drawing to be held at a later date. The individuals in the high-power position would be able to allocate six tickets to others in their group according to performance; those in the low-power position were told they would not be able to allocate any tickets for the raffle. In reality, all participants received a raffle ticket at the end of the study. After the group task, the participants completed both voice and SDO measures. They were then debriefed and received their raffle tickets

Measures Power was operationalized through the assigned position of participants in the business simulation (CEO ⫽ high power; line worker ⫽ low power). The manipulation was structured to measure power according to the previously explained definition, which intertwines both resource control and status components. Control over resources was manipulated by the ability to distribute rewards, in the form of raffle tickets. Status–positional power was manipulated by assigning and explaining, through the organiza-

POWER, VOICE, AND HIERARCHY

tional structure chart, the relative positions of the different workers. A manipulation check, in the form of the question “How much power does your position carry in the company?” with a response scale ranging from none at all to a great deal (on a scale from 1 to 7), yielded a mean value of 6.58 (SD ⫽ 1.06) for high-power individuals, whereas those in low-power positions had a mean of 1.91 (SD ⫽ 1.06). This difference was statistically significant, F(1, 119) ⫽ 575.16, p ⬍ .001. SDO was measured using Pratto et al.’s (1994) scale. This scale consists of 14 items, assessing belief in hierarchy. Examples of items on this scale are “Some groups of people are simply not the equals of others” and “Some people are just more worthy than others.” The scale was determined to have adequate reliability (␣ ⫽ .81; Cronbach, Gleser, Nanda, & Rajaratnam, 1972). Further, the results of an exploratory factor analysis, using maximum-likelihood estimation, extracted a single factor with an eigenvalue of 4.2. Voice was measured by each participant, who rated both his or her own and all other group members’ voice behavior. Voice measures were adopted from Van Dyne and LePine’s (1998) voice scale. This scale consists of six items, anchored from 1 (strongly disagree) to 7 (strongly agree). The six items on this scale concern the degree to which a person pronounces his or her opinion in the group. Slight modifications were made to the scale in order to accommodate it to an experimental setting, resulting in a four-item scale. Examples of the items included in the voice scale are “This particular coworker spoke up in ways that affected the quality of group interaction” and “This particular coworker developed and made recommendations concerning relevant issues.” Each individual’s voice score was computed by averaging their scores as rated by all other partic-

97

ipants (self-ratings were excluded because of problems with the validity of self-ratings; cf. Atwater & Yammarino, 1992). Thus, the reliability of this measure is a function of both the scale reliability across items (␣ ⫽ .88) and interrater reliability after collapsing across items (␣ ⫽ .72, an acceptable level of reliability, as this value is a function of both scale and interrater reliabilities; LeBreton, Burgess, Kaiser, Atchley, & James, 2003; Shavelson, Webb, & Rowley, 1989).

Results Table 1 displays the means, standard deviations, and intercorrelations of the study variables. Although there was a significant negative effect of group size on voice behavior (discussed below), there were no significant correlations between any of the study-relevant variables, indicating an inability to reject the null associated with Hypothesis 1. To assess the interaction between power and SDO on voice, we conducted a hierarchical moderated regression analysis (Cohen & Cohen, 1983). To allow a proper interpretation of the data and also make the scales of the dependent and independent variables comparable (Aiken & West, 1991), we centered power, SDO, and voice around the grand mean before entering them into the regression analysis. Although our data were nested within group, which is at times justification for group-mean centering, we used the technique of grand-mean centering owing to our small intraclass correlation (as discussed below; Hofmann & Gavin, 1998). Voice served as the dependent variable in all steps of the regression, with group size disaggregated to the individual level of analysis and entered into the first step (to control for varying group sizes), power and SDO entered into the second step, and in the final step a

Table 1 Means, Standard Deviations, and Intercorrelations Among Study Variables Variable 1. 2. 3. 4.

Group size Power Social dominance orientation Voice behavior

* p ⬍ .05.

M

SD

1

2

3

4

4.17 1.53 5.35 4.91

1.20 0.50 0.82 1.01

— ⫺.09 ⫺.08 ⫺.23*

— .09 .06

— .02

—

98

ISLAM AND ZYPHUR

composite term, Power ⫻ SDO, entered. Significance of results was ascertained by observing the significance of changes in R2 from the second to the third step. Table 2 displays the beta weights and changes in R2 associated with this regression analysis. As can be seen, neither power nor SDO independently predicted voice behavior. (Note that as an examination of the effects of controlling for group size, we also tested the effects of power and SDO before controlling for group size, and they were not significant, p ⬎ .05.) However, consistent with Hypothesis 2, the interaction between power and SDO was significantly related to voice behavior (␤ ⫽ .259, p ⬍ .01). The nature of this interaction is depicted in Figure 1. Position of power (high vs. low) was predictive of voice behavior only for individuals high in SDO. We tested this in a subsequent analysis, using a median split, in which we conducted separate regression analyses for those high and low in SDO. Our results indicate a differential relationship between power and voice across high and low SDO groups. More specifically, for those high in SDO, power predicted voice (␤ ⴝ .293, p ⫽ .023), whereas for those low in SDO, power did not (␤ ⴝ ⫺.199, p ⫽ .124). These relationships are evident in Figure 1, depicted as voice means at high and low SDO (median split) by high and low power. The results of our individual-level results for the moderating effect of SDO on the relation-

Table 2 Hierarchical Multiple Regression Analysis Predicting Voice Behavior Voice Predictor Step 1: Group size Step 2 Group size Power Social dominance orientation Step 3 Group size Power Social dominance orientation Power ⫻ Social Dominance Orientation

␤

a

⌬R2

⫺.23*

.05*

⫺.23* .05 ⬎⫺.01

⬎.01

⫺.23** .05 ⫺.05 .26**

.07**

Note. The beta weights are reported after each step. * p ⬍ .05. ** p ⬍ .01.

Figure 1. Interaction between power and social dominance orientation (SDO) on voice behavior.

ship between power and voice are promising. However, to test for the robustness of this effect across groups (i.e., the variance in the regression coefficient of the interaction term) and to allow us to correct for possible nonindependence related to individuals nested within groups, we also conducted hierarchical linear modeling (HLM; Bryk & Raudenbush, 1992; Pollack, 1998) with a slopes- and intercepts-asoutcomes model, first standardizing our data to ease the interpretation of HLM (Pollack, 1998). To examine the intraclass correlation (ICC) of our data (i.e., the degree of nonindependence), we followed procedures outlined in Kenny, Mannetti, Pierro, Levi, and Kashy (2002), adjusting our n for unequal group size in accord with recommendations by Kenny and La Voie (1985). The computed ICC was small and negative (␳ ⫽ ⫺.052), indicating negative nonindependence in our data. As discussed by Kenny et al. (2002), negative nonindependence arises when there is a fixed resource within a group (e.g., time, power, money; also discussed by Sadler & Judd, 2001, as associated with any zero-sum situation). In our case, this may be conceptualized simply as the fact that individuals had a limited amount of time for discussion, and because voice scores represent the amount of talking done by an individual, the more an individual talked in a group, the less able other group members were to talk. Support for this conceptualization of our ICC also comes in the form of the negative relationship between group size and voice score (see Tables 2 and 3), indicating that when more individuals were in a

POWER, VOICE, AND HIERARCHY

99

Table 3 HLM of Multilevel Effects With Intercepts and Slopes as Outcomes Effect Fixed Model for group means Intercept, ␥00 Group size, ␥01 Model for PWR-VC Intercept, ␥10 Group size, ␥11 Model for SDO-VC Intercept, ␥20 Group size, ␥21 Model for PWR/SDO-VC Intercept, ␥30 Group size, ␥31

Gamma

SE

t

p

.12 ⫺.26

.08 .08

1.49 ⫺3.28

.15 ⬎.01

.03 .06

.11 .10

0.23 0.57

.82 .58

⫺.05 .05

.13 .14

⫺0.36 0.35

.72 .73

.23 .13

.08 .09

2.99 1.47

.01 .15

Parameter variance Random Group mean, u0j PWR-VC slope, u1j SDO-VC slope, u2j PWR/SDO-VC slope, u3j Level 1 effect, rij

.04 .16 .23 .01 .62

␹2

p

31.30 36.83 40.66 17.88

.01 ⬎.01 ⬎.01 .27

Note. Degrees of freedom for random effects (but not fixed effects) are based on only 17 groups, as 12 groups contained only 3 members, disallowing computation. HLM ⫽ hierarchical linear modeling; PWR ⫽ power; VC ⫽ voice; SDO ⫽ social dominance orientation; PWR/SDO ⫽ power by social dominance orientation interaction term.

group, there was less average talking, due to all members having to share the limited amount of time (i.e., the ratio of time to people decreased). Problematically, HLM is unable to handle data that are negatively nonindependent, as HLM computes ICCs with variances, and variances may not be negative (Kenny et al., 2002). Kenny et al. (2002) recommended a variety of possible procedures to allow HLM to compute the ICC; however, we discussed this matter with the publishers of HLM and were advised on a course of action different from that outlined in Kenny et al. We were told that instead of producing a negative variance component to correctly estimate a negative ICC, which would cause the program to malfunction, HLM will automatically keep variance positive and small (M. du Toit & S. du Toit, personal communication, December 24, 2004). As HLM corrects the degrees of freedom associated with statistical significance in accord with an absolute ICC value (i.e., an ICC of .5 and one of ⫺.5 would require the same adjustment to degrees of free-

dom, as they indicate the same amount of nonindependence), we computed effects using HLM without a setup able to model a negative ICC. (We would like to note that failing to model a negative ICC will not affect coefficient estimates, only degrees of freedom [M. du Toit & S. du Toit, personal communication, December 24, 2004; see Kenny et al., 2002; Nezlek & Zyzniewski, 1998].) This produced a model with a small and positive ICC (␳ ⫽ .063), allowing a slightly more conservative correction to degrees of freedom than would have been allowed with the original, negative ICC estimate. Results of this model appear in Table 3. These results are consistent with our original analysis and indicate that although group size significantly predicted average voice (␥01 ⫽ ⫺.260, p ⫽ .003), it was not a significant predictor of the effects of power, SDO, and their interaction on voice. Further, by examining the variance component related to the effect of the interaction of power and SDO on voice (i.e.,

100

ISLAM AND ZYPHUR

u3j), one finds that the stability of the effect across groups is indicated, ␹2(16, N ⫽ 85) ⫽ 17.88, p ⫽ .268, providing support for the consistency of Hypothesis 2 across groups.

Discussion This study is an attempt at once to understand what influences people to voice their opinions and to unpack the importance of holding power in social groups. We present a theoretical model in which the correlates of holding power and perceiving status hierarchies are analyzed in terms of their tendencies to affect voice behavior. Our results partially validate this model. The two hypotheses previously presented constitute a partial moderation model, in which there is a main effect of power on voice behavior, and this relationship is heightened by SDO. Our results confirm this heightening effect but fail to find a main effect of power. Thus, the model implied by the experimental data is a full moderation model, with SDO constituting a necessary requisite for the effects of power on voice. The conclusion that SDO fully moderates the power–voice relationship is not inconsistent with our theoretical explanation about the role of SDO and provides support for Hypothesis 2. However, the failure to find a main effect of power does seem surprising given the proliferation of research showing a main effect of power on voice, irrespective of SDO (Keltner et al., 2000). Below, we present several possible explanations of why this surprising effect may have occurred. First, it is important to note that the experimental manipulation of power was by necessity weaker than the “real world” experience of power in actual organizations. For one, the rewards in the current task were smaller (raffle tickets vs. a paycheck), and the status markers were less legitimate (performance on a pretest vs. wealth, education, and many other status symbols of power). Thus, our power manipulation may not have had the power to evoke realistic social power dynamics. Also important may be the fact that participants did not know that they were assigned to dichotomous, high-, or low-power positions and thought that a continuum of power positions was reflected in the simulation. Thus, they may

not have believed their power (or lack of power) to be absolute; rather, they may have assumed there were people close to them in power. Because of this, they may have been reluctant to act out their power roles to the fullest. Further, the power manipulation that was used was based on a hierarchical organization chart, which may mean that the power roles taken on by participants were highly sensitive to SDO levels. That is, because SDO involves attitudes toward hierarchy, presenting power positions through a hierarchical scheme may have heightened the moderating effect of SDO on power, such that only people who believed in the justness of hierarchy “bought into” the power manipulation. There is also a possibility that past research has misconstrued the power–voice effect by ignoring the moderating variable of SDO (Pratto et al., 1994). As none of the past studies on the power–voice effect have examined or controlled for SDO, it could be that some power effects found in past research have contained hidden effects of SDO. For instance, field studies of those in high-power positions may inadvertently be confounding power effects with SDO, as power has been correlated with SDO in real-world settings (Pratto et al., 1994). If this is the case, then parsing apart the effects of power and SDO on voice would require random assignment to power positions. The results of this study, which suggest a power–SDO interaction, challenge the view of power as simply a control over resources or status. If power’s effect on voice is moderated by the ways people view hierarchy in social situations, then this effect is not simply the product of objective power positions and risks but also contains a social– cognitive component. This moderating effect points to the ways people categorize themselves with respect to their groups and explains how power dynamics may be altered on the basis of individual attitudes (Sidanius & Pratto, 1999). This has implications for any theory of the effects of power and calls for a view of power not only as a resource and status variable but as one that describes global views of how society is (or should be) structured. From our results, although we do not deny that power holders treat others differently than non-power-holders (Kipnis, 1976) as well as promote behavioral differences among their

POWER, VOICE, AND HIERARCHY

subordinates (Kets De Vries, 1991; Mossholder, Bennett, Kemery, & Wesolowski, 1998), we can assert that such effects are dependent on the cognitive frameworks of those involved. Consistent with this conception, our study indicates that the “objective” (i.e., risk-related, structural) elements of power had effects on voice only when viewed through the attitudinal lens of social dominance. The interaction effect also raises the question of the plasticity of dominance orientations. Sidanius and Pratto (1999) presented SDO as a traitlike, individual difference variable; however, they also stress the importance of socialization in forming attitudes toward hierarchy. If a manager were trying to promote equal participation in a group, our results suggest that part of his or her task would require promoting an egalitarian climate within the organization, one that provides not only a structural possibility for voice but also a group ethic that deemphasizes the dominance of some actors over others. The ability to do this would be contingent on the plasticity of SDO. Along this vein, applied research should examine the effectiveness of interventions in promoting egalitarian values in group settings. One implicit lesson from this study is that our understanding of how people voice their concerns in groups must move beyond simply observing participation programs and interventions. As we have seen, even when participation is allowed, and even encouraged through incentives, people may refrain from giving their opinions owing to group and cognitive factors that are extrinsic to the content of the group task (Bedeian, 2002; Keltner et al., 2000). Briefly put, speaking up is not an automatic result of removing barriers to expression but often requires the right kind of social and attitudinal framework to nurture expression. We have attempted to present an approach to voice that links people’s expression of opinions with the power and authority structures, both objective and psychological, that permeate groups. Such a linkage, it is argued, facilitates our understanding of voice in both theoretical and practical ways. For any field that concerns itself with organizations, the understanding of expression, power, and authority must be central to understanding life in the workplace.

101

References Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage. 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, 1362–1377. Atwater, L. E., & Yammarino, F. J. (1992). Does self– other agreement on leadership perceptions moderate the validity of leadership and performance predictions? Personnel Psychology, 45, 141–164. Bedeian, A. G. (2002). The dean’s disease: How the darker side of power manifests itself in the office of dean. Academy of Management Learning and Education, 1, 164 –174. Bryk, A. S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data analysis methods. Thousand Oaks, CA: Sage. Coats, E. J., & Feldman, R. S. (1996). Gender differences in nonverbal correlates of social status. Personality and Social Psychology Bulletin, 22, 1014 –1022. Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences. Hillsdale, NJ: Erlbaum. Cortina, L. M., & Magley, V. J. (2003). Raising voice, risking retaliation: Events following interpersonal mistreatment in the workplace. Journal of Occupational Health Psychology, 8, 247–265. Cotton, J. L., Vollrath, D. A., Froggatt, K. L., Lengnick-Hall, M. L., & Jennings, K. R. (1988). Employee participation: Diverse forms and different outcomes. Academy of Management Review, 13, 8 –22. Cronbach, L. J., Gleser, G. C., Nanda, H., & Rajaratnam, N. (1972). The dependability of behavioral measurements: Theory of generalizability for scores and profiles. New York: Wiley. Dachler, H. P., & Wilpert, B. (1978). Conceptual dimensions and boundaries of participation in organizations: A critical evaluation. Administrative Science Quarterly, 23, 1–39. Dyck, B., & Starke, F. A. (1999). The formation of breakaway organizations: Observations and a process model. Administrative Science Quarterly, 44, 792– 822. Fiske, S. T. (1993). Controlling other people: The impact of power on stereotyping. American Psychologist, 48, 621– 628. Goodwin, S. A., Operario, D., & Fiske, S. T. (1998). Situational power and interpersonal dominance facilitate bias and inequality. Journal of Social Issues, 54, 677– 698. Hecht, M., & LaFrance, M. (1998). License or obligation to smile: The effect of power and sex on

102

ISLAM AND ZYPHUR

amount and type of smiling. Personality and Social Psychology Bulletin, 24, 1332–1342. Hirschman, A. O. (1970). Exit, voice, and loyalty: Responses to decline in firms, organizations, and states. Cambridge, MA: Harvard University Press. Hofmann, D. A., & Gavin, M. B. (1998). Centering decisions in hierarchical linear models: Implications for research in organizations. Journal of Management, 24, 623– 641. Janis, L. L. (1972). Groupthink: Psychological studies of policy decisions and fiascoes. Boston: Houghton Mifflin. Keltner, D., Gruenfeld, D. H., & Anderson, C. (2000). Power, approach and inhibition (Rep. No. 1669). Stanford, CA: Stanford University, Graduate School of Business. Kenny, D. A., & La Voie, L. (1985). Separating individual and group effects. Journal of Personality and Social Psychology, 48, 339 –348. Kenny, D. A., Mannetti, L., Pierro, A., Levi, S., & Kashy, D. A. (2002). The statistical analysis of data from small groups. Journal of Personality and Social Psychology, 83, 126 –137. Kets de Vries, M. F. R. (1991). On becoming a CEO: The transference and the addictiveness of power. In M. F. R. Kets de Vries (Ed.), Organizations on the couch: Clinical perspectives on organizational behavior and change (pp. 120 –139). San Francisco: Jossey-Bass. Kipnis, D. (1976). The powerholders. Chicago: University of Chicago Press. Kipnis, D. (1984). The view from the top. Psychology Today, 18, 30 –36. Kizilos, M. A. C., Thomas, G., & Strickstein, A. (1994). Achieving superior customer service through employee involvement: An empirical investigation. Academy of Management Proceedings, 197–205. Kramer, R. M. (1998). Revisiting the Bay of Pigs and Vietnam decisions 25 years later: How well has the groupthink hypothesis stood the test of time? Organizational Behavior and Human Decision Processes, 73, 236 –271. LeBreton, J. M., Burgess, J. R. D., Kaiser, R. B., Atchley, E. K., & James, L. R. (2003). The restriction of variance hypothesis and interrater reliability and agreement: Are ratings from multiple sources really dissimilar? Organizational Research Methods, 6, 80 –128. LePine, J. A., & Van Dyne, L. (1998). Predicting voice behavior in work groups. Journal of Applied Psychology, 83, 853– 868. Liden, R. C., & Mitchell, T. R. (1988). Ingratiatory behaviors in organizational settings. Academy of Management Review, 13, 572–587. Locke, E. A., & Schweiger, D. M. (1979). Participation in decision making: One more look. Greenwich, CT: JAI Press.

Locke, E. A., Shaw, K. N., Saari, L. N., & Latham, G. P. (1981). Goal setting and task performance. Psychological Bulletin, 90, 125–152. Milgram, S. (1983). Obedience to authority: An experimental view. New York: HarperCollins. Mossholder, K. W., Bennett, N., Kemery, E. R., & Wesolowski, M. A. (1998). Relationships between bases of power and work reactions: The mediational role of procedural justice. Journal of Management, 24, 533–552. Nezlek, J. B., & Zyzniewski, L. E. (1998). Using hierarchical linear modeling to analyze grouped data. Group Dynamics, 2, 313–320. Olson-Buchanan, J. B., & Boswell, W. R. (2002). The role of employee loyalty and formality in voicing discontent. Journal of Applied Psychology, 87, 1167–1174. Pollack, B. N. (1998). Hierarchical linear modeling and the “unit of analysis” problem: A solution for analyzing responses of intact group members. Group Dynamics, 2, 299 –312. Pratto, F., Sidanius, J., Stallworth, L. M., & Malle, B. F. (1994). Social dominance orientation: A personality variable predicting social and political attitudes. Journal of Personality and Social Psychology, 67, 741–763. Rusbult, C. E., Farrell, D., Rogers, G., & Mainous, A. G., III. (1988). Impact of exchange variables on exit, voice, loyalty and neglect: An integrative model of responses to declining job satisfaction. Academy of Management Journal, 31, 599 – 627. Sadler, M. S., & Judd, C. M. (2001). Overcoming dependent data: A guide to the analysis of group data. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology: Group processes (pp. 497–524). Malden, MA: Blackwell. Salancik, G., & Pfeffer, J. (1977). Who gets power— and how they hold on to it: A strategic contingency model of power. Organizational Dynamics, 5, 2–21. Shavelson, R. J., Webb, N. M., & Rowley, G. L. (1989). Generalizability theory. American Psychologist, 44, 922–932. Sidanius, J. (1993). The psychology of group conflict and the dynamics of oppression: A social dominance perspective. In S. M. Iyengar & W. J. McGuire (Eds.), Explorations in political psychology (pp. 183–219). Durham, NC: Duke University Press. Sidanius, J., & Pratto, F. (1999). Social dominance: An intergroup theory of social hierarchy and oppression. New York: Cambridge University Press. Staw, B. M., & Boettger, R. D. (1990). Task revision: A neglected form of work performance. Academy of Management Journal, 33, 534 –559. Staw, B. M., Sandelands, L. E., & Dutton, J. E. (1981). Threat-rigidity effects in organizational

POWER, VOICE, AND HIERARCHY

behavior: A multilevel analysis. Administrative Science Quarterly, 26, 501–524. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. G. Austin (Eds.), Psychology of intergroup relations (2nd ed., pp. 7–24). Chicago: NelsonHall. Tiedens, L. Z., Ellsworth, P. C., & Mesquita, B. (2000). Stereotypes about sentiments and status: Emotional expectations for high- and low-status group members. Personality and Social Psychology Bulletin, 26, 560 –573. Van Dyne, L., & LePine, J. A. (1998). Helping and voice extra-role behaviors: Evidence of construct and predictive validity. Academy of Management Journal, 41, 108 –119. Wagner, J. A. I. (1994). Participation’s effects on performance and satisfaction: A reconsideration of

103

research evidence. Academy of Management Review, 19, 312–331. Watson, D., & Clark, L. A. (1997). Extraversion and its positive emotional core. In R. Hogan & J. A. Johnson (Eds.), Handbook of social psychology (pp. 767–793). San Diego, CA: Academic Press. Withey, M. J., & Cooper, W. H. (1989). Predicting exit, voice, loyalty, and neglect. Administrative Science Quarterly, 34, 521–539. Zhou, J., & George, J. M. (2001). When job dissatisfaction leads to creativity: Encouraging the expression of voice. Academy of Management Journal, 44, 682– 697. Received August 3, 2004 Revision received January 9, 2005 Accepted January 12, 2005 䡲