Organizational Behavior and Human Decision Processes 110 (2009) 70–79

Contents lists available at ScienceDirect

Organizational Behavior and Human Decision Processes journal homepage: www.elsevier.com/locate/obhdp

Testosterone–status mismatch lowers collective efficacy in groups: Evidence from a slope-as-predictor multilevel structural equation model Michael J. Zyphur a,1, Jayanth Narayanan b,1,*, Gerald Koh c, David Koh c a

Department of Management and Marketing, Level 10, 198 Berkeley Street, The University of Melbourne, Victoria 3010, Australia Department of Management & Organization, NUS Business School, National University of Singapore, 1 Business Link, Singapore 117592, Singapore c Department of Epidemiology and Public Health, Yong Loo Lin School of Medicine, National University of Singapore, National University Health System, Block MD3, 16, Medical Drive, Singapore 117597, Singapore b

a r t i c l e

i n f o

Article history: Received 21 July 2008 Accepted 4 May 2009 Available online 21 June 2009 Keywords: Status Collective efficacy Testosterone Groups

a b s t r a c t The study of the biological underpinnings of behavior is in its nascent stages in the field of management. We study how the hormone testosterone (T) is related to status and collective efficacy in a group. We assessed salivary testosterone of 579 individuals in 92 teams. We find that T does not predict status within the group. We also tested the effects of a mismatch between T and status in the group on the collective efficacy of the group. Using a novel slope-as-predictor multilevel structural equation model, we find that the greater the mismatch between T and status in the group (i.e., the more negative the within-group correlation among T and status), the lower is the collective efficacy of the group. We discuss the implications of our findings for the study of the biological underpinnings of group behavior in organizations. Ó 2009 Elsevier Inc. All rights reserved.

The study of human behavior is in the midst of a revolution. This revolution has its impetus in the integration of the biological sciences—classically considered outside the purview of the ‘‘soft sciences”—into fields such as economics, sociology, and social psychology. Although fields such as economics are embracing these developments (e.g., Camerer, Loewenstein, & Prelec, 2005; Glimcher, 2003; Kosfeld, Heinrichs, Zak, Fischbacher, & Fehr, 2005), management scholars have largely ignored the role of biology in the pursuit of their study (see exceptions: Arvey, Bouchard, Segal, & Abraham, 1989; Fox, Dwyer, & Ganster, 1993; Nicolaou, Shane, Cherkas, Hunkin, & Spector, 2008; White, Thornhill, & Hampson, 2006). In order to remain on the cutting edge of social science scholarship, the field of management and organizational studies must now catch up with related disciplines that are pioneering the integration of their study with biology. We believe that understanding the relationship between hormones and organizational behavior is an ideal starting point for this purpose. This is because many hormones can be thought of as individual differences much like personality, and they are related to attitudes and behavior (e.g., Fox et al., 1993). Hormones are chemicals that are released from an endocrine gland or from other tissues into the blood stream and organs such * Corresponding author. E-mail address: [email protected] (J. Narayanan). 1 The first two authors contributed equally to this paper. This paper was conceived and executed when the first author was at the National University of Singapore. 0749-5978/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.obhdp.2009.05.004

as the brain. Hormones provide a slower means of control over the functioning of biologically processes compared to the nervous system (Brown, 1994). Hormones fall mainly into four types: peptides, amino acids, fatty acids, and steroids. Here, our focus is on the steroid testosterone (T), which affects social behavior in groups (Mazur & Booth, 1998). While hormones such as T can change in response to feedback from the environment, basal levels of hormones are predictive of a number of social behaviors and are generally stable over time. For example, levels of T change following status-relevant contests (e.g., Booth, Shelley, Mazur, Tharp, & Kittock, 1989; Rejeski, Gagne, Parker, & Koritnik, 1989; Schultheiss, Wirth, & Stanton, 2004a, 2004b; Schultheiss et al., 2005), but such effects are transitory—as noted by Mazur and Booth, ‘‘changes in T associated with winning and losing tend to be short-lived and modest in size” (1998, p. 371). In the current paper, we address the relationship between T and status within groups, and how this relationship influences a group’s collective efficacy. We focus on T within groups for multiple reasons. First, T is related to social behavior in humans and many other species (Nieschlag, Nieschlag, & Behre, 2004). Further, T heritability coefficients have been estimated at around .40 (see Meilke, Stringham, Bishop, & West, 1987), meaning that 40% of the variation in levels of T can be accounted for by genetic factors. This lends credence to the idea that T is a relatively stable individual characteristic, a notion confirmed by test–retest correlations of around .70—commensurate with traditional individual-difference measures, such as personality (see Newman, Sellers, & Josephs, 2005; Sellers, Mehl, & Josephs, 2007). Also, T is one of the most well studied hormones

M.J. Zyphur et al. / Organizational Behavior and Human Decision Processes 110 (2009) 70–79

in relation to human behavior, allowing strong theoretical postulations regarding its possible role in organizational life (Kemper, 1990). Further, studies suggest that T’s effects on human psychology and behavior are stable across both males and females (e.g., Josephs, Sellers, Newman, & Mehta, 2006), allowing the study of T in organizations to occur in a gender-unbiased fashion—in fact, showing equivalent effects of T across genders, we believe, may serve to reduce gender biases justified through folk theories of T’s effects in the workplace. Our paper unfolds as follows. First, we explore basal levels of T as important individual-differences, and as predictors of phenomena relevant to intra-group behavior. Then, we discuss the dominant theoretical rationale relating T to status in groups, as well as how variance in this intra-group relationship should, in turn, influence collective efficacy. Then, we test our hypotheses with a novel ‘‘slope-as-predictor” multilevel structural equation model, where the effect of T on status within groups predicts collective efficacy between groups. We conclude by discussing the implications of our results for management scholars.

71

Further, across males and females, T is related to status in prison populations (Dabbs, Frady, Carr, & Besch, 1987; Dabbs & Hargrove, 1997; Dabbs, Ruback, Frady, & Besch,, 1988; Ehrenkranz, Bliss, & Sheard, 1974), a preference for more risky occupations and occupations requiring interpersonal confrontation (Dabbs, Alford, & Fielden, 1998; Dabbs, de La Rue, & Williams 1990), positions in occupations associated with higher status (Cristiansen & Knussman, 1987; Purifoy & Koopmans, 1979; Udry & Talbert, 1988; although see Dabbs, 1992), and higher levels of performance in such occupations (Fannin & Dabbs, 2003). Testosterone and dominance

Much of people’s lives at home, at work, or at play is spent in groups. A number of important social behaviors such as leadership, interpersonal attraction and aggression, and status attainment are enacted within the context of a social group. Not surprisingly, a large body of management research is devoted to the study of groups. Studies on group behavior point to a number of sociallydetermined factors such as group composition, function, and structure as important determinants of status conferral and attitudes within groups (for a summary, see Levine & Moreland, 1998). However, this research has hitherto paid little attention to how the biology of the human organism relates to group functioning (although see Woolley, Gerbasi, Chabris, Kosslyn, & Hackman, 2008). One of the most well studied biological factors that influences behavior and attitudes in groups is T. Influenced by signaling from the hypothalamus and pituitary, T is an androgen that is produced in the gonads of women and men, as well as the adrenal glands and brains of both genders (Brown, 1994)—because T is produced in large quantities by the testes, men have around seven times as much serum T as women (Mazur & Booth, 1998) and three times as much salivary T as women (Granger, Shirtcliff, Booth, Kivligham, & Schwartz, 2004). T has effects on all major organs and a host of neurological systems (Janowsky, 2006). To have its most proximal effects, T is broken down into estradiol (an estrogen) or other androgens, such as dihydrotestosterone. In the brain, these chemicals and their metabolites have been shown to affect neocortical, hippocampal, and amygdalic activity across both genders (Janowsky, 2006).

To provide a foundation for these findings described above, and the hypotheses we propose below, it is important to explore the psychological effects of varying levels of T. Researchers have shown that T is related to implicit and explicit motivations for power and social dominance (Gray, Jackson, & McKinlay, 1991; Schindler, 1979; Sellers et al., 2007) as well as attention to status cues and status threats (Schultheiss & Brunstein, 2001). Further, T is associated with reductions in fear responses (e.g., Hermans, Putman, Baas, Koppeschaar, & Van Honk, 2006) and reduced attention to facial threat display (van Honk & Schutter, 2007). Because T causes status striving and awareness, T is most often linked to behavior only when status may be gained or is threatened (Kouri, Lukas, Pope, & Oliva, 1995; Mazur & Booth, 1998). Psychologically, levels of T are related to implicit motivations for power (for a review see Schultheiss, 2008), defined as the degree to which individuals desire to be socially visible, influential, and dominating (Winter, 1973). Such motivations for power are distinct from overt aggression that is often associated with T in folk understandings of the hormone (Schultheiss, 2008)—as noted by Rose (1975, p. 467), ‘‘testosterone is correlated more with assertiveness or an action orientation than with aggression in the usual context if attacking others.” Further, it is notable that all of the effects of T described here have been shown to hold across genders (see early work by Baucom, Besch, & Callahan, 1985; Schindler, 1979). This is sensible when considering that previous research has shown equivalent power motivations across men and women (Stewart & Chester, 1982) as well as equivalent responses to power cues (Pang & Schultheiss, 2005; Schultheiss & Brunstein, 2001), irrespective of the average difference between the sexes in T. Additionally, recent findings indicate that controlling for mean differences of T across the sexes (i.e., ‘‘centering” T levels within sex) allows equating the effects of T on psychological and behavioral outcomes (see Josephs et al., 2006). Finally, neurocognitive research shows that areas of the brain affected by testosterone are not those that tend to vary with gender (for a review see Janowsky, 2006).

Testosterone and social behavior

The mismatch effect

Considerable research evidence suggests that T is related to social behavior in humans (Nieschlag et al., 2004). Notably, T levels are associated with social dominance and status attainment in human groups (Archer, 2006; Klinesmith, Kasser, & McAndrew, 2006; Mazur & Booth, 1998; Tremblay et al., 1998)—in fact, the first studies linking T to status in animals helped coin the term ‘‘pecking order” when observing an increase in status among chickens as T increased (see Allee, Collias, & Lutherman, 1939). Although T is present in greater quantities in males, it has been shown to have an equivalent impact on social dominance in both males and females after controlling for the difference in T associated with gender (for discussions see Grant, 1998; Kemper & Collins, 1990; Mazur & Booth, 1998).

Importantly, other research shows negative effects of a ‘‘mismatch” in T and status, such that individuals high (low) in T and low (high) in status experience elevated heart rate and blood pressure, reductions in positive affect, and decrements in cognitive resources (Josephs et al., 2006). This is to say that when T and status are negatively related, individuals become emotionally disturbed and cognitively distracted, but the same is not true when T and status ‘‘match” (i.e., are positively correlated). This ‘‘mismatch hypothesis,” proposed and support by Josephs et al. (2006), states that when T and social status are at odds, there are a host of negative outcomes for individuals; low T people are uncomfortable in high-status positions, supposedly because of the implied dominance of their role and their having to engaging

Testosterone

72

M.J. Zyphur et al. / Organizational Behavior and Human Decision Processes 110 (2009) 70–79

in status-maintenance behaviors; high T individuals are uncomfortable being in low-status positions, likely because of the lack of dominance implied by their role and their having to engage in deferential behaviors. In both cases, individuals become psychologically stressed and show increased heart rates and blood pressure (Josephs et al., 2006) and reduced levels of cognitive functioning (Josephs, Newman, Brown, & Beer, 2003), supposedly because of the cognitive occupation associated with their social standing in relation to their desired social standing, which is related to levels of T. Summary and contribution of the current study Through the effects described above, T may be understood as having a status-seeking and status-maintaining function. Theoretically, T drives individuals to attain high status, focuses attention on threats to their status, and may have an antagonizing effect on fear responses to such threats. These three phenomena should increase individuals’ propensities to engage and persist in status contests and, through such contests, attain positions of social dominance—these effects are stable across the sexes (Dabbs, Alford, & Fielden, 1998; Grant & France, 2001; Purifoy & Koopmans, 1979). Additionally, it has been shown that when a mismatch between T and status exists, individuals become uncomfortable, which should further motivate them to attain levels of status commensurate with their levels of T within a social hierarchy (Josephs et al., 2006). With this in mind, and although the above research provides a strong foundation for the current study, it is notable that nearly all of the research described above suffers from one (or more) of three shortcomings: (a) it has either occurred at the level of individual behavior without being nested within an actual group or otherwise interpersonal context (e.g., Klinesmith et al., 2006), (b) it has been largely uncontrolled experimentally, without random assignment and without relying on individuals drawn from normal populations (e.g., Dabbs et al., 1987, 1988), or (c) it has not been over time, relying on effects observed across very short periods of time (e.g., Josephs et al., 2006). This has the result that there has been no systematic investigation of how levels of T influence status and attitudes over time within groups drawn from normal populations. In the current study we attempt to rectify this shortcoming in the T literature. Theory and hypotheses The most theoretically potent description of how T should influence status in groups is found in Mazur’s (1985) biosocial, face-toface interaction model of status contests (see also Kemper, 1990; Mazur & Booth, 1998). According to this model, simple and often unconscious signs are used to indicate and attain status. Individuals can sign their social standing through a variety of behaviors, such as posture or gesticulation, a ‘‘stare-down” or gaze aversion, or assertive voice behavior. These behaviors help to organize status hierarchies when individuals acknowledge the higher status of others and defer in some recognizable way, or challenge others’ status. Such challenges are dominance contests, wherein individuals compete for status. During status contests, psychological stress is likely to be the driving force behind winning or losing the contest (Mazur, 1985; Mazur & Booth, 1998). Consider the example of a ‘‘stare-down,” where two individuals hold each other’s gaze. Through this action, stress is created in each staring individual, until one individual becomes too uncomfortable to continue the status contest and averts his or her gaze, while the other party continues the stare. Such a contest can occur in an instant (i.e., it may be unconscious) and the winner of this contest may repeat the behavior to reify the

outcome (Chase, Bartolomeo, & Dugatkin, 1994; Kemper & Collins, 1990)—once ambiguity in status has been reduced such contests will become less and less frequent (for non-human discussions see Pagel & Dawkins, 1997). Because T is associated with status striving, increased likelihood of perceiving status threats, and reduced fear responses, it is reasonable to conclude that high T levels should allow an individual to seek and persist in such stress-causing status contests (see Viau & Meaney, 1996). Through iterative, dyadic wins and losses of status contests across group members, status hierarchies will form in groups as people find (or create) their social position (Chase, 1974; for computational models see Hogeweg, 1988). Because T should increase status contest participation and wins in such contests, the position of each individual in the status hierarchy should conform to their level of T as individuals iteratively win and lose status contests within a group. Adding credence to such an idea is the research described above, where T was described as being positively related to motivations for power, a focus on status and status cues, reduced fear responses, and discomfort with T–status mismatch. Hypothesis 1. Testosterone will predict status within a group. Notwithstanding the relationship between T and status within a group, it is possible to postulate a between-group effect of the T–status relationship based on the ‘‘mismatch” hypothesis described earlier. As noted, when T and status are negatively correlated, reductions in positive affect and cognitive occupation result (Josephs et al., 2006), both of which will affect group process. For example, positive affect has been linked with helping behaviors and other antecedents to positive intra-group dynamics (see George & Brief, 1992). Additionally, it is likely that a focus on one’s status rank will occur at the expense of a focus on group goals. Such a focus, and any focus on the individual-centered goal of increasing status, may negatively impact group performance (DeShon, Kozlowski, Schmidt, Milner, & Wiechmann, 2004). With such reductions in affect and cognitive resources in a group, actual group performance is likely to fall below possible group performance—termed ‘‘process loss” (Thompson, 1967). When this occurs, group members are likely to become aware of such a deficit, which in turn should influence their collective beliefs regarding their group’s ability to complete tasks successfully. Thus, one likely result of reductions in positive affect and cognitive resources in a group is a reduction in collective efficacy—defined as a group’s shared perception of its ability to succeed at a task (Bandura, 1986). Social cognitive theory proposes that self-efficacy, which is an evaluation of one’s own ability to complete a task, has a strong influence on human affect, motivation, and action (Bandura, 1988). Bandura (1986, 2000) notes that people in groups hold beliefs about the ability of their group to perform a task, and such beliefs have consequences for the group’s motivation, resilience, and performance. A number of studies on collective efficacy point to its importance in group functioning (e.g., Campion, Medsker, & Higgs, 1993; Durham, Knight, & Locke, 1997). The idea of collective efficacy (CE) is an extension of Bandura’s initial work on self-efficacy (see Bandura, 1977). Bandura (1990) notes that self-efficacy beliefs are the product of a complex process of self-persuasion that relies on cognitive processing of performance accomplishments, vicarious experiences, verbal persuasion, and physiological states. Feelings of CE are at the core of a group’s functioning and such feelings have been associated with superior group outcomes, such as setting more challenging goals (Durham et al., 1997) and persisting in the face of challenges (Bandura, 2000), with the result that groups high in CE are more likely to produce higher levels of performance on group tasks (Campion et al., 1993).

M.J. Zyphur et al. / Organizational Behavior and Human Decision Processes 110 (2009) 70–79

Beliefs of collective efficacy develop over time and are a result of group members’ observations of other members within the group (Gibson, 1999; Marks, Mathieu, & Zaccaro, 2001). One of the important features of a group is the nature of the hierarchy in the group. Leaders and high status members exert an important influence on the functioning of groups. Indeed, one primary function of a leader is to induce a belief that the group possesses the ability to attain its goals. Therefore, the behavior of leaders and high status members of a group exert a particularly important influence on the development of collective efficacy of the group (Chen & Bliese, 2002; Zaccaro, Blair, Peterson, & Zazanis, 1995). Although the relationship between T and status within a group is an empirical question, as we articulated above, there is likely to be variance across groups in this relationship. In other words, there are likely to be groups where high T members have high status, as well as groups where low T members have high status. We propose that variance in the T–status relationship within a group is likely to have an important effect on how efficacious the group feels as a whole. Clues to the possible nature of this relationship can be inferred from the mismatch effect proposed by Josephs and colleagues (2006). In the words of Josephs et al. (2006, p. 1001), ‘‘if a less assertive, low testosterone individual lands a high status position, the resulting mismatch between a desire to ‘fly below the radar’ and the position of high status might generate a strong reaction, ranging from fear to confusion to arousal to a motivation to return to a safer level of status”. Therefore, when low T individuals in a group assume positions of status (i.e., a mismatch), they may be less likely to transmit a sense of efficacy about the group (and vice versa). In contrast, when high T individuals assume these positions (i.e., a match), they may be more likely to instill a sense of efficacy in their group members. We hypothesized that to the extent that T levels and status perceptions in the group are matched, groups would feel a greater sense of collective efficacy. Hypothesis 2. The effect of T on status within a group will predict collective efficacy in a positive direction, such that as the relationship between T and status becomes more positive (negative), collective efficacy will be increased (reduced). Method Participants Participants were 579 students, 259 males and 320 females, enrolled in an introductory organizational behavior course. The average age of the sample was 20.97 (SD = 1.68) years. Procedure Participants were randomly assigned to 92 groups ranging in size between 4 and 7. These groups met twice weekly for the duration of the semester (12 weeks) and were required to work on a variety of tasks in-class, as well as produce a professional management-training video as a final class project. There were no formally-assigned roles or positions within a status hierarchy at the start of the task. Therefore, groups were allowed to naturally form status hierarchies and adopt roles—in groups, this is a natural process over time (Moreland & Levine, 1988). Measures Status. During the sixth week of the semester, participants rated the status of each member of their group in a 360° fashion along five items using a 7-point scale—example items are ‘‘(This individ-

73

ual) influences group goals and decisions” and ‘‘(This individual) leads conversation in the group”. Responses from every group member were averaged to compute an individual’s standing along each item, and individuals’ ratings of themselves were excluded to avoid bias (see Atwater & Yammarino, 1992). To assess the measure’s psychometric properties, an exploratory factor analysis was conducted and a single factor was extracted accounting for 84.65% of the variance in the measure. Additionally, Cronbach’s alpha was computed at .96. Because status in a group is a within-groups construct (i.e., the only variance of concern is that among individuals within a group), the amount of between- versus within-groups variance in the measure was assessed with an ANOVA. Results showed a non-significant amount of between-groups variance in the measure (F = .62, p = .95), indicating the items functioned appropriately at the within-group level of analysis (Dansereau, Alutto, & Yammarino, 1984). Testosterone. Testosterone was measured via saliva collected from all participants in late September in Singapore. Saliva collection was performed during the 6th week of the semester during a fixed time period in the afternoon in order to control for the circadian rhythm of testosterone levels. Participants were advised in advance not to eat or drink except for water 1 h before saliva collection, in order to minimize possible food debris and stimulation of salivation. They were also not allowed to brush or floss their teeth an hour before saliva collection to minimize oral micro-injury, which may spuriously increase salivary testosterone levels (Kivligham et al., 2004). For saliva collection, approximately 2 ml of passive drool was collected in a 15 ml polypropylene tube by gently spitting through a sterile funnel. Materials to stimulate saliva were not used. Saliva collected was systematically inspected for visible blood contamination. A pilot test on saliva sample collection procedure was performed to minimize artificial inflation of testosterone measured, as recommended by various researchers (e.g., Dabbs, 1991; Shirtcliff, Granger, & Likos, 2002). Cotton-based saliva collection methods were not used because of possible interference with salivary testosterone (e.g., Granger et al., 2004; Shirtcliff, Granger, Schwartz, & Curran, 2001). After saliva collection, samples were immediately kept in an icebox and transported to the laboratory within 4 h. The samples were centrifuged at 1000 g at 4 °C for 15 min, and the supernatant was aliquoted into vials and stored at 80 °C until analysis. Salivary testosterone remains stable for at least 2 years when stored at this temperature (Granger et al., 2004). Salivary testosterone concentration (qg/ml) was determined using the ER Testosterone EIA Kit, produced by Salimetrics. Average testosterone for females was 57.44 (SD = 24.06) and for males it was 151.38 qg/ ml (SD = 48.86). Because hypotheses center on the T–status relationship within groups, the group-mean along T was removed from each participant’s T score (i.e., T was group-mean centered; see Raudenbush & Bryk, 2002). Additionally, to remove any effect of gender on status, the mean along T for each gender was removed from each participant’s T score. In other words, any effect of group or gender in the T–status relationship was removed.1 Collective efficacy. Collective efficacy was measured at the end of the semester, 6 weeks after saliva and status data were collected. A two-item measure was adapted from Bandura (2005); the items

1 To assure an equivalent effect of T on status across males and females, in an exploratory model we tested for a testosterone X gender interaction when predicting status within groups. In line with the majority of literature on the effects of T on social outcomes across genders, this revealed no statistically significant interaction. Additionally, in an exploratory model, we examined for an effect of gender composition on the T–status effect within groups. We found no statistically significant effect, indicating that gender composition does not moderate the within-group effect of T on status.

74

M.J. Zyphur et al. / Organizational Behavior and Human Decision Processes 110 (2009) 70–79

read, ‘‘I am confident in my group’s ability to complete its responsibilities effectively,” and, ‘‘I am confident that my group is capable of good performance”. As collective efficacy is a between-groups construct, the amount of between vs. within-groups variance in the measure was assessed using an ANOVA, and the proportion of between-groups variance to total variance (i.e., ICC[1]) was computed at .10. Results showed a significant amount of between-groups variation in the measure (F = 1.48 and p = .007), justifying the use of the items at the between-group level of analysis (Dansereau et al., 1984). Additionally, the correlation between the two items in the measure was .99 at the between-group level, so the average of the items was computed and used in all analyses. T–status mismatch. Mismatch in the T–status relationship within a group was operationalized as the effect of T on status in that group. If the effect of T on status within a group is negative and strong, then T can be thought of as being mismatched with status, because those high in T will tend to have low status. Alternatively, if the effect of T on status within a group is positive and strong, then T can be thought of as being aligned with status, because those high in T will tend to have high status. In other words, as the relationship among T and status within a group becomes more negative, T and status become more mismatched; as the relationship among T and status within a group becomes more positive, T and status become more matched. In effect, then, T–status mismatch is a variable that exists between groups—it is the T–status relationship for each group—but each group’s standing along this variable describes the group’s T–status relationship. We describe below how we operationalized this variable as a latent ‘‘random slope” variable indicated by the within-group T–status relationship. As a cursory check to assure adequate levels of difference between groups in the T–status relationship, we computed the correlation among T and status for each group. These correlations are plotted in Fig. 1. As shown in this figure, there is substantial variation in the within-group T–status relationship, with correlations ranging from around .75 to .75. This broad range of values indicates the suitability of the T–status relationship as a differentiator of groups in the extent to which T and status are mismatched. Control variables. Although the mean difference between males and females along T was removed, it is still possible that differ-

ences in the gender composition of each group could impact a group’s CE, so we controlled for any such compositional differences. Additionally, we sought to control for differences in the number of individuals within each group. Therefore, control variables at the between-group level were both group size and gender composition. ‘‘Slope-as-predictor” multilevel structural equation model To test our hypotheses, we specified a novel ‘‘slope-as-predictor” multilevel structural equation model (see Fig. 2) using a fullinformation maximum likelihood estimator. This model allows us to simultaneously estimate effects at multiple levels of analysis and model latent variables dynamically across these levels (for the logic of multilevel structural equation models, see Mehta & Neale, 2005; Rabe-Hesketh, Skrondal, & Pickles, 2004). At the within-group level of analysis, we specified a latent variable to account for covariation among the observed status variables—this is a traditional latent variable (i.e., a latent factor). Also at this level of analysis, we regressed the latent status variable onto T to test Hypotheses 1 and 2. At the between-group level of analysis we specified CE as a latent variable indicated by the scores of individual group members—in multilevel-modeling parlance such a latent variable is termed a ‘‘random intercept” (see Raudenbush & Bryk, 2002). Such a latent variable specification is more appropriate than using average values along CE for each group. This is because the individuals in each group serve as reflective indicators of the group-level CE variable, making the group-level CE variable latent (see Lüdtke et al., 2008). To justify the use of this latent CE variable between groups, we examined the model-estimated variance of CE for statistical significance before regressing this variable onto any other variables. There was a statistically-significant amount of variation around the latent CE variable (r2 = .075 and p < .01). Also, we allowed the effect of T on status to randomly vary between groups. The latent variable capturing variation between groups along the T–status relationship is termed a ‘‘random slope” in multilevel literature (see Raudenbush & Bryk, 2002). This random slope, effectively, allows each group to have a unique T–status relationship, and differences in this relationship between the groups is captured by the latent slope variable. This latent slope variable is, then, an operationalization of T–status mismatch. When a group’s standing along the latent slope variable is negative and strong, this means that there is a strong negative relationship between T and status within the group and, thus, there is a T–status mismatch. Alternatively, when a group’s standing along the latent slope variable is positive and strong, this means that there

Sizej .023

.10 Gender_Cj

-.21 28.02*

<.01 .00006*

CEj

Between -group

RSj

1.04*

Within-group Testosterone ij

Statusij

1.0 y1i j

Fig. 1. Histogram of the testosterone–status relationship, as a correlation for each group.

y2i j

.95 .76 y3 ij

.86 y4 ij

.68 y5i j

Fig. 2. Results from a multilevel structural equation model, showing unstandardized effects; y1ij y5ij = the five status items for each individual i in a group j; RS = random slope of the within-group T–status effect; gender_C = gender composition; CE = collective efficacy and size = group size.

75

M.J. Zyphur et al. / Organizational Behavior and Human Decision Processes 110 (2009) 70–79 Table 1 Descriptive statistics and correlations among study variables. Variable

M

r

1

1. Testosterone 2. Status1 3. Status2 4. Status3 5. Status4 6. Status5 7. Collective_E 8. RS 9. Group size 10. Gender_C

99.46 4.31 4.42 4.44 4.02 4.41 5.88 <.01 6.39 1.55

59.80 1.17 1.10 0.92 1.02 .86 .28 <.00 .61 .23

– .01 .02 .01 .03 .02 – – – –

2

3

4

– .91 .88 .88 .88 – – – –

– .87 .92 .85 – – – –

– .88 .88 – – – –

5

6

7

– – – – –

– .65 .14 .19

8

9

10

– .03

–

– .84 – – – –

– .48 .07

Note: status1–status5 is the items measuring status; RS is random slope of the T–status effect; collective_E is collective efficacy; and gender_C is the gender composition; correlations among testosterone and status1–status6 are within-group correlations (N = 586), while correlations among collective_E, RS, group size, and gender_C are between-group correlations (N = 92); all correlations associated with collective_E and RS are model-estimated.

is a strong positive relationship between T and status within the group and, thus, there is a T–status match. To justify the use of this random slope between groups, we examined its model-estimated variance for statistical significance. There was a statistically-significant amount of variation around the random slope (r2 = .0000604, p < .01).2 Based on there being adequate amounts of variation along CE and the random slope between group, we regressed CE onto the latent random slope to test Hypothesis 2. At the between-group level we also included groups’ gender composition and the number of individuals in each group. To control for their effects we regressed CE onto these variables. Results The intercorrelations and descriptive statistics associated with all variables are presented in Table 1. As shown in Table 1, T has an extremely small relationship with all status variables, and this is also true of T’s effect on the latent status variable (see Table 2 and Fig. 2), as evidence by the statistically non-significant mean of the random slope variable (i.e., the statistically non-significant average effect of T on status across all groups; a = .001 and p = .74). Another illustration of this overall effect is shown in Fig. 1, where the distribution of T–status correlations across groups is centered at zero (M = .01). This lack of an effect of T on status supports Hypothesis 1. Turning to the effect of T–status mismatch (i.e., the random slope) on CE, we find a very strong correlation among these variables (see Table 1). To test the effect of T–status mismatch on CE we first regressed CE on the control variables group size and gender composition. This reduced variance in CE across groups marginally (r2 = .071, p < .01; pseudo-R2 = .05). We then regressed CE onto the random slope—a ‘‘slope-as-predictor” multilevel model. The effect of the random slope on CE was statistically-significant with a one-tailed test (see Table 2 and Fig. 2; c = 28.02 and p = .04). Additionally, this reduced the variance of CE significantly after accounting for the effect of group size and gender composition (r2 = .023 and p < .63; Dpseudo-R2 = .64). This result provides 2 It is notable that although the amount of variation in the random slope is small in absolute terms, this is to be expected. The variance term in this case reflects the squared standard deviation of differences between groups in the effect of T on status within the groups. The scaling of T and status is such that the unstandardized effect of T on status will be small in all cases. This is because the effect of T on status is a value indicating the amount of change in status (which ranges between 1 and 7) as a function of a change of 1.0 along T (which has a range from around 10 to over 250). Because of the difference in variation among the two variables (see Table 1), the magnitude of the unstandardized effect will be small even when there is a large standardized effect (because r = B*SDx/SDy). This produces the appearance of very little variation along the random slope, but a different picture emerges when the T– status relationships are standardized, as shown in Fig. 1.

Table 2 Results of ‘‘slope-as-predictor” multilevel structural equation model. PE Within-group Status ? Y1 Status ? Y2 Status ? Y3 Status ? Y4 Status ? Y5 Statusr2 Between-group CEr2 RSr2 RSa (T ? status) RS ? CE Group_S ? CE Gender_C ? CE

PEstd

SE

PE/SE

p – <.01 <.01 <.01 <.01 <.01

1.00 .95 .76 .86 .68 1.04

.95 .96 .94 .94 .92 1.00

– .02 .02 .02 .03 .08

– 47.12 37.11 35.10 26.81 13.03

.02 <.01 <.01 28.03 .10 .21

1.00 1.00 – .80 .22 .18

.05 <.01 <.01 16.14 .10 .21

.49 2.01 .32 1.74 1.04 .98

.63 .04 .75 .04a .30 .33

Note: PE is the parameter estimate; PEstd is standardized parameter estimate; SE is standard error; PE/SE is t-value associated with parameter estimate; statusr2 is variance of latent status variable; CEr2 is variance of latent CE (collective efficacy) variable; RSr2 is variance of the random slope of the T–status effect; and RSa is the mean of the random slope. a One-tailed test of statistical significance.

support for Hypothesis 2. Finally, because of the interpretational difficulties associated with the random slope’s effect on CE, we sought to illustrate the change in CE as a function of standardized changes in the T–status relationship. We did this by estimating CE values with a zero-order standardized relationship among T and status, and then plus and minus one standard deviation around this zero-order T–status relationship, as well as at the limits of the distribution shown in Fig. 1. Estimated CE values are shown in Fig. 3, where higher (lower) standardized effects of T on status are associated with higher (lower) CE values (see Fig. 4). Discussion The study of the biological underpinnings of organizational behavior is important for many reasons. First, it serves to bridge the divide between the ‘‘hard” and ‘‘soft” sciences. Second, rather than simply adding more complexity to empirical studies, this integration has the potential to greatly aid a basic scientific understanding of organizational phenomena. Third, a biological focus has the potential to add relevancy to management science on the larger academic stage, both through the use of well known hard-science methods as well as by appealing to a wider audience of scholars. By taking a biological approach in the current study, we advance social scientific research in multiple ways. First, the utility of physiological measures—and T specifically—in the pursuit of

76

M.J. Zyphur et al. / Organizational Behavior and Human Decision Processes 110 (2009) 70–79

Fig. 3. Graph depicting changes in collective efficacy as a function of changes in the correlation among testosterone and status within a group, where collective efficacy is plotted along the y-axis and values above each bar are correlations among testosterone and status. CE = collective efficacy.

understanding group behavior is shown. Second, the finding that T and status are not necessarily linked in groups from normal populations addresses an important divide in the sociological and psychological literature, where the determinants of status in human groups have been debated extensively. While one view contends that biological factors outweigh other social and cognitive processes of status allocation (Rosa & Mazur, 1979), an alternative view posits that humans differ from primates in that these social and cognitive factors play a more important role in status allocation (Berger, Cohen, & Zelditch, 1972; Ridgeway, 1993). To the best of our knowledge, there has been no large scale study of the nature that we have undertaken to verify the biological model of status allocation in groups. We show that T in itself is insufficient to explain status in attainment in groups. Our findings provide preliminary evidence that factors other than T might be leading to status allocation in groups. The exact nature of what these factors are and how they interact with T merits further investigation.

Fig. 4. Scatterplot showing the relationship between collective efficacy and the testosterone–status slope. Because both variables are latent, scores along both variables were estimated using the factor-score estimation function in Mplus version 5.1.

Our findings suggest that the view of status in the theoretical formulation discussed by Mazur (1985) is restricted to contexts where status positions may be gained through dominance-oriented behaviors. Humans are unique in that they exhibit relative ranking on characteristics that are not merely dominance oriented. The characteristics associated with status in human groups can be any nominal characteristic on which people differ such as age, gender, race, and the like. Unlike in the animal kingdom, the process of creating a status hierarchy in human populations may not be a result of status allocation through dominance contests (Ridgeway, 1987). The nominal characteristic(s) that determines status in human groups depends on the context of the social interaction. In the context of interdependent groups, a long tradition of research has examined the performance expectations that people hold of individuals who belong to the groups characterized by any given nominal characteristic. The importance of performance expectations in the allocation of status is evident in this comment from Blau (1964, p. 47): Men who make essential contributions to a group as a whole, or to its members individually, have an undeniable claim to superior status. . . The obligations of group members to those who make such benefits possible are discharged by according them superior status. They command respect and compliance, which are rewards for having made contributions in the past and as incentives for making them in the future. Status characteristics theory, formulated by Berger et al. (1972) provides a precise formulation of this intuition that performance expectations matter in groups. The theory posits that status hierarchies are driven by these expectations. The expectations about one’s own and other group members’ ability to contribute to accomplishing tasks determine the deference that members show each other. Members defer to individuals of whom they expect high performance. Thus, an individual who commands high performance expectations will be given greater opportunities to influence decisions and outcomes in such interactions. Based on status characteristics, expectation states theory argues that the essential foundation for status within a group lies in the member’s expectations of one another’s performance with regard to the group task (Berger et al., 1972). The emphasis here is on the cooperative allocation of status based on inferred ability to contribute to the group (Ridgeway, 1984; see also Driskell, Olmstead, & Salas, 1993). The theory is specific to groups whose members depend on each other and work together to achieve a common group goal— much like groups in our empirical setting. In such groups, status hierarchies are determined by both specific cues—cues that provide information about a member’s competence or expertise with respect to a clearly defined—and specifiable task and diffuse cues—cues that provide information about a member’s general ability, which can be presumed to affect their performance on a variety of different tasks within some domain. Specific status cues will tend to vary across settings; indicators of specific experience and/or task-relevant training and education are most likely to be seen as specific status cues. Obvious physical characteristics and social category differences such as gender, ethnicity, age, or attractiveness are most likely to be viewed as diffuses status cues (see Berger, Rosenholtz, & Zelditch, 1980). Both types of status cues are distinct and useful in determining performance expectations in a group (Bunderson, 2003). Considering the relationship between T and performance expectations, there is no reason to assume a relationship between T and performance expectations or expertise within a group populated by individuals drawn from a normal population. Therefore, it is reasonable to assume no relationship between T and status in groups. In other words, instead of dominance behaviors (which may be related to T) allowing individuals to acquire status in groups, it is more likely that task-relevant knowledge, skills, and abilities (which are likely unrelated to T) will allow individuals

M.J. Zyphur et al. / Organizational Behavior and Human Decision Processes 110 (2009) 70–79

to acquire status in groups. Future studies should examine this proposition further. Although there is no relationship between T and status at the individual level, we show that the T–status relationship in a group can impact group attitudes. Specifically, we show that as T and status become more negatively related (i.e., as T and status become more ‘‘mismatched”) there are decrements in group collective efficacy. Although previous research has addressed the factors that precede the development of collective efficacy (e.g., Bandura, 1986), none of this research has explored the physiological basis of this important group phenomenon. The current findings demonstrate the utility of aligning T and status in groups. Future research should determine the psychological and group-process mechanisms through which the T–status relationship impacts collective efficacy. Although the effect of collective efficacy on group performance has been demonstrated extensively in the literature, it is an important omission from our study design given the constraints of the data that we were in possession. Although we would expect a similar relationship with performance, the effect of the status–T mismatch on the performance of the group is worthy of further investigation. From a methodological perspective, we illustrate a novel slopeas-predictor multilevel structural equation model, and show how random slopes may be used as operationalizations of interesting constructs in multilevel—and more specifically group—research. Additionally, this multilevel structural equation model illustrates how traditional structural equation models and multilevel models may be integrated for the betterment of management science, allowing the dynamic use of latent factors and slopes in a single complex structural equation. Limitations There are a number of limitations to the current study that should be kept in mind when interpreting our results. First, although individuals were randomly assigned to groups and all groups were given similar instructions for their group task, there was very little control over group process in the current study. Although we see this as possibly beneficial to the extent that this allowed groups to enact any T-consistent behavior patterns that may have arisen, this also means that we were unable to control for any confounding circumstance or variables that may have moderated or otherwise altered the T–status relationship. Next, although we show a very large effect of T–status mismatch on CE, this is not the same as showing this same effect on the important outcome variables that CE has been shown to affect. Specifically, future studies should attempt to uncover the effect of T–status mismatch on group performance and other criteria. Additionally, given the current findings, such studies could focus on the mediated effect of T–status mismatch on group performance through CE. With regards to the measurement of T, serum measurements are superior to salivary measurements. Shirtcliff et al. (2002) have found that when salivary levels of T are used to estimate for serum free T, the former underestimates free T and behavior relationships by as much as 65% in females and 20% in males. Hence, the effect size of any association between salivary T and behavior is probably an under-estimation of the true effect size if serum free testosterone was used. Another factor to consider with the measurement of T is that we sampled T at a single point in time for each participant. Previous research has shown that T has a state-dependent nature due to factors such as winning and losing status contests (for a review see Mazur & Booth, 1998). Future studies should confirm our results while measuring T at multiple times across the day and week. Our measurement choice was made on account of resource and logistical constraints. Nevertheless, our measurement

77

of T should be kept in mind when making inferences about our findings. Next, it is important to note that our inference of a causal relationship between the within-group T–status relationship and CE is based on the fact that T and status were measured at week 6 and CE at week 12. However, our analyses do not control for levels of CE at week 6, limiting our ability of rule out the possibility that pre-existing levels of CE may have steered the ordering of status within the groups, and therefore acted to cause the T–status relationship in some fashion. Future studies could be used to make stronger causal inferences. Finally, it is notable that our sample age range is both young on average and relatively homogeneous. Additionally, individuals in our sample were obtained from a university in Singapore, and, although we did not explicitly measure race, the majority of them were from an Asian background. Both of these factors limit the generalizability of our findings to other populations. Future studies should confirm our findings with other samples. Future research The avenues for future research in the relationship between T and status, as well as this within-group relationship on CE, are many. Importantly, the current study was not designed to investigate the process that led to both a T–status mismatch or match within a group, as well as how the T–status relationship came to influence CE. Looking at Fig. 1 indicates how very substantial variance in the T–status relationship is between groups. Future studies should investigate why certain groups show such a strong positive relationship and why others show such a negative relationship. Additionally, these same studies could be targeted towards understanding exactly how T–status mismatch influences CE. Additionally, although CE has been shown to be an important group attitude for predicting group outcomes (e.g., Campion et al., 1993), CE is often considered important only to the extent that it explains such outcomes. In the current study we were unable to gain access to reliable performance outcome data, and thus future studies should be directed towards investigating how T–status mismatch influences important group outcomes such as performance. However, it is notable that in such studies, our finding that T–status mismatch has a strong effect on CE can serve as a basis for positing a mediating role of CE in any impact of T–status mismatch on performance or other equally important group-level outcome variables. Acknowledgments This project was made possible through research funding from the National University of Singapore (R-317-000-067-112/133). We thank the editor and the reviewers for helping us refine the paper. We thank Dan Mcallister for helping us with collecting this data. References Allee, W., Collias, N., & Lutherman, C. (1939). Modification of the social order in flocks of hens by the injection of testosterone propionate. Physiological Zoology, 12, 412–440. Archer, J. (2006). Testosterone and human aggression: An evaluation of the challenge hypothesis. Neuroscience and Biobehavioral Reviews, 30, 319–345. Arvey, R. D., Bouchard, T. J., Segal, N. L., & Abraham, L. M. (1989). Job satisfaction: Environmental and genetic components. Journal of Applied Psychology, 74, 187–192. 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. Bandura, A. (1977). Social learning theory. New York: General Learning Press.

78

M.J. Zyphur et al. / Organizational Behavior and Human Decision Processes 110 (2009) 70–79

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall. Bandura, A. (1988). Perceived self-efficacy: Exercise of control through self-belief. In J. P. Dauwalder, M. Perrez, & V. Hobi (Eds.). Annual series of European research in behavior therapy (Vol. 2, pp. z27–z59). Amsterdam/Lisse, Netherlands: Swets & Zeitlinger. Bandura, A. (1990). Perceived self-efficacy in the exercise of personal agency. Journal of Applied Sport Psychology, 2, 128–163. Bandura, A. (2000). Exercise of human agency through collective efficacy. Current Directions in Psychological Science, 9, 75–78. Bandura, A. (2005). Self-efficacy beliefs of adolescents. Information Age Publishing. Baucom, D. H., Besch, P. K., & Callahan, S. (1985). Relation between testosterone concentration, sex role identity, and personality among females. Journal of Personality and Social Psychology, 48, 1218–1226. Berger, J., Cohen, B. P., & Zelditch, M. Jr., (1972). Status characteristics and social interaction. American Sociological Review, 37, 241–255. Berger, J., Rosenholtz, S. J., & Zelditch, M. (1980). Status organizing processes. Annual Review of Sociology, 6, 479–508. Blau, P. (1964). Exchange and power in social life. New York: Wiley. Booth, A., Shelley, G., Mazur, A., Tharp, G., & Kittock, R. (1989). Testosterone and winning and losing in human competition. Hormones and Behavior, 23, 556–571. Brown, R. (1994). An introduction to neuroendocrinology. Cambridge: Cambridge University Press. Bunderson, J. S. (2003). Recognizing and utilizing expertise in work groups: A status characteristics perspective. Administrative Science Quarterly, 48, 557–591. Camerer, C., Loewenstein, G., & Prelec, D. (2005). Neuroeconomics: How neuroscience can inform economics. Journal of Economic Literature, 43, 5–60. Campion, M. A., Medsker, G. J., & Higgs, A. C. (1993). Relations between work group characteristics and effectiveness: Implications for designing effective work groups. Personnel Psychology, 46, 823–850. Chase, I. D. (1974). Models of hierarchy formation in animal societies. Behavioral Science, 19, 374–382. Chase, I. D., Bartolomeo, C., & Dugatkin, L. (1994). Aggressive interactions and intercontest interval: How long do winners keep winning? Animal Behaviour, 48, 393–400. Chen, G., & Bliese, P. D. (2002). The role of different levels of leadership in predicting self and collective efficacy: Evidence for discontinuity. Journal of Applied Psychology, 87, 549–556. Cristiansen, K., & Knussman, R. (1987). Androgen levels and components of aggressive behavior in men. Hormones and Behavior, 21, 170–180. Dabbs, J. M. (1991). Salivary testosterone measurements: Collecting, storing, and mailing saliva samples. Physiology and Behavior, 49, 815–817. Dabbs, J. M. (1992). Testosterone and occupational achievement. Social Forces, 70, 813–824. Dabbs, J. M., Alford, E. C., & Fielden, J. A. (1998). Trial lawyers and testosterone: Blue-collar talent in a white-collar world. Journal of Applied Social Psychology, 28, 84–94. Dabbs, J. M., de La Rue, D., & Williams, P. M. (1990). Testosterone and occupational choice: Actors, ministers, and other men. Journal of Personality and Social Psychology, 59, 1261–1265. Dabbs, J. M., Frady, R. L., Carr, T. S., & Besch, N. F. (1987). Saliva testosterone and criminal violence in young adult prison inmates. Psychosomatic Medicine, 49, 174–182. Dabbs, J. M., & Hargrove, M. F. (1997). Age, testosterone, and behavior among female prison inmates. Psychosomatic Medicine, 59, 477–480. Dabbs, J. M., Ruback, R. B., Frady, R. L., & Besch (1988). Saliva testosterone and criminal violence among women. Personality and Individual Differences, 9, 269–275. Dansereau, F., Alutto, J., & Yammarino, F. (1984). Theory testing in organizational behavior. Englewood Cliffs NJ: Prentice Hall. DeShon, R. P., Kozlowski, S. W. J., Schmidt, A. M., Milner, K. R., & Wiechmann, D. (2004). A multiple-goal, multilevel model of feedback effects on the regulation of individual and team performance. Journal of Applied Psychology, 89, 1035–1056. Driskell, J. E., Olmstead, B., & Salas, E. (1993). Task cues, dominance cues, and influence in task groups. Journal of Applied Psychology, 78, 51–60. Durham, C., Knight, D., & Locke, E. (1997). Effects of leader role, team-set goal difficulty, efficacy, and tactics on team effectiveness. Organizational Behavior and Human Decision Processes, 72, 203–231. Ehrenkranz, J., Bliss, E., & Sheard, M. (1974). Plasma testosterone: Correlation with aggressive behavior and social dominance in man. Psychosomatic Medicine, 36, 469–475. Fannin, N., & Dabbs, J. M. (2003). Testosterone and the work of firefighters: Fighting fires and delivering medical care. Journal of Research in Personality, 37, 107–115. Fox, M., Dwyer, D., & Ganster, D. C. (1993). Effects of stressful job demands and control on physiological and attitudinal outcomes in a hospital setting. Academy of Management Journal, 36, 289–318. George, J., & Brief, A. P. (1992). Feeling good–doing good: A conceptual analysis of the mood at work-organizational spontaneity relationship. Psychological Bulletin, 112, 310–329. Gibson, C. B. (1999). Do they do what they believe they can? Group efficacy and group performance across tasks and cultures. Academy of Management Journal, 42, 138–152. Glimcher, P. W. (2003). Decisions, uncertainty, and the brain: The science of neuroeconomics. Cambridge, MA: The MIT Press.

Granger, D. A., Shirtcliff, E. A., Booth, A., Kivligham, K. T., & Schwartz, E. B. (2004). The ‘‘trouble” with salivary testosterone. Psychoneuroendocrinology, 29, 1229–1240. Grant, V. J. (1998). Maternal personality: Evolution and the sex ratio. Routledge. Grant, V. J., & France, J. T. (2001). Dominance and testosterone in women. Biological Psychology, 58, 41–47. Gray, A., Jackson, D. N., & McKinlay, J. B. (1991). The relation between dominance, anger, and hormones in normally aging men: Results from the Massachusetts male aging study. Psychosomatic Medicine, 53, 375–385. Hermans, E. J., Putman, P., Baas, J. M. P., Koppeschaar, H., & Van Honk, J. (2006). A single administration of testosterone reduces fear potentiated startle in humans. Biological Psychiatry, 59, 872–874. Hogeweg, P. (1988). Mirror beyond mirror, puddles of life. In C. Langton (Ed.), Artificial life, SFI studies in the sciences of complexity (pp. 297–316). Redwood City, California: Adisson-Wesley Publishing Company. Janowsky, J. S. (2006). Thinking with your gonads: Testosterone and cognition. Trends in Cognitive Sciences, 10, 77–82. Josephs, R. A., Newman, M. L., Brown, R. P., & Beer, J. M. (2003). Status, testosterone, and human intellectual performance: Stereotype threat as status concern. Psychological Science, 14, 158–163. Josephs, R. A., Sellers, J. G., Newman, M. L., & Mehta, P. H. (2006). The mismatch effect: When testosterone and status are at odds. Journal of Personality and Social Psychology, 90, 999–1013. Kemper, T. D. (1990). Social structure and testosterone: Explorations of the socio-biosocial chain. New Jersey: Rutgers Univ. Press. Kemper, T. D., & Collins, R. (1990). Dimensions of microinteraction. American Journal of Sociology, 96, 32–68. Kivligham, K. T., Granger, D. A., Schwartz, E. B., Nelson, V., Curran, M., & Shirtcliff, E. A. (2004). Quantifying blood leakage into the oral mucosa and its effects on the measurement of cortisol, dehydroepiandrosterone and testosterone in saliva. Hormones and Behavior, 46, 39–46. Klinesmith, J., Kasser, T., & McAndrew, F. T. (2006). Guns, testosterone, and aggression: An experimental test of a mediational hypothesis. Psychological Science, 17, 568–571. Kosfeld, M., Heinrichs, M., Zak, P., Fischbacher, U., & Fehr, E. (2005). Oxytocin increases trust in humans. Nature, 435, 673–676. Kouri, E. M., Lukas, S. E., Pope, G. G., & Oliva, P. S. (1995). Increased aggressive responding in male volunteers following the administration of gradually increasing doses of testosterone cypionate. Drug and Alcohol Dependence, 40, 73–79. Levine, J. M., & Moreland, R. L. (1998). Small groups. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), Handbook of social psychology (4th. ed., Vol. 2, pp. 415–469). Boston: McGraw-Hill. Lüdtke, O., Marsh, H. W., Robitzsch, A., Trautwein, U., Asparouhov, T., & Muthén, B. (2008). The multilevel latent covariate model: A new, more reliable approach to group-level effects in contextual studies. Psychological Methods, 13(3), 203–229. Marks, M. A., Mathieu, J. E., & Zaccaro, S. J. (2001). A temporally based framework and taxonomy of team processes. Academy of Management Review, 26, 355–376. Mazur, A. (1985). A biosocial model of status in face-to-face primate groups. Social Forces, 64, 377–402. Mazur, A., & Booth, A. (1998). Testosterone and dominance in men. Behavioral and Brain Sciences, 21, 353–363. Mehta, P. D., & Neale, M. C. (2005). People are variables too: Multilevel structural equation modeling. Psychological Methods, 10, 259–284. Meilke, A. W., Stringham, J. D., Bishop, D. T., & West, D. W. (1987). Quantitating genetic and nongenetic factors influencing androgen production and clearance rates in men. Journal of Clinical Endocrinology and Metabolism, 67, 104–119. Moreland, R. L., & Levine, J. M. (1988). Group dynamics over time: Development and socialization in small groups. In J. McGrath (Ed.), The social psychology of time: New perspectives (pp. 151–181). Newbury Park, CA: Sage. Newman, M. L., Sellers, J. G., & Josephs, R. A. (2005). Testosterone, cognition, and social status. Hormones and Behavior, 47, 205–211. Nicolaou, N., Shane, S., Cherkas, L., Hunkin, J., & Spector, T. D. (2008). Is the tendency to engage in entrepreneurship genetic? Management Science, 54, 167–179. Nieschlag, S., Nieschlag, E., & Behre, H. M. (2004). Testosterone: Action, deficiency, substitution. Cambridge: Cambridge University Press. Pagel, M., & Dawkins, M. S. (1997). Peck orders and group size in laying hens: ‘Future contracts’ for non-aggression. Behavioural Processes, 40, 13–25. Pang, J. S., & Schultheiss, O. C. (2005). Assessing implicit motives in US. college students: Effects of picture type and position, gender and ethnicity, and crosscultural comparisons. Journal of Personality Assessment, 85(3), 280–294. Purifoy, E. E., & Koopmans, L. H. (1979). Androstenedione, testosterone, and free testosterone concentration in women in various occupations. Social Biology, 26, 179–188. Rabe-Hesketh, S., Skrondal, A., & Pickles, A. (2004). Generalized multilevel structural equation modeling. Psychometrika, 69, 167–190. Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods. Sage. Rejeski, W. I., Gagne, M., Parker, P. E., & Koritnik, D. R. (1989). Acute stress reactivity from contested dominance in dominant and submissive males. Behavioral Medicine, 15, 118–124. Ridgeway, C. (1984). Dominance, performance, and status in groups. A theoretical analysis. In E. Lawler (Ed.). Advances in group processes: Theory and research (Vol. 1, pp. 59–93). Greenwich, CT: JAI Press. Ridgeway, C. L. (1987). Nonverbal behavior, dominance, and the basis of status in task groups. American Sociological Review, 52, 683–694.

M.J. Zyphur et al. / Organizational Behavior and Human Decision Processes 110 (2009) 70–79 Ridgeway, C. L. (1993). Gender, status and the social psychology of expectations. In P. England (Ed.), Theory on gender/feminism on theory (pp. 175–197). New York: Aldine. Rosa, E., & Mazur, A. (1979). Incipient status in small groups. Social Forces, 58, 18–37. Rose, R. M. (1975). Testosterone and aggression in man. Psychosomatic Medicine, 36, 467–468. Schindler, G. L. (1979). Testosterone concentration, personality patterns, and occupational choice in women. Unpublished dissertation, University of Houston, Texas. Schultheiss, O. C. (2008). Implicit motives. In O. P. John, R. W. Robins, & L. A. Pervin (Eds.), Handbook of Personality: Theory and Research (3rd ed., pp. 603–633). New York: Guilford. Schultheiss, O. C., & Brunstein, J. C. (2001). Assessment of implicit motives with a research version of the TAT: Picture profiles, gender differences, and relations to other personality measures. Journal of Personality Assessment, 77, 71–86. Schultheiss, O. C., Wirth, M. M., & Stanton, S. J. (2004a). Effects of affiliation and power motivation arousal on salivary progesterone and testosterone. Hormones and Behavior, 46, 592–599. Schultheiss, O. C., Wirth, M. M., & Stanton, S. (2004b). Effects of affiliation and power motivation arousal on salivary progesterone and testosterone. Hormones and Behavior, 46, 592–599. Schultheiss, O. C., Wirth, M. M., Torges, C. M., Pang, J. S., Villacorta, M. A., & Welsh, K. M. (2005). Effects of implicit power motivation on men’s and women’s implicit learning and testosterone changes after social victory or defeat. Journal of Personality and Social Psychology, 88, 174–188. Sellers, J. G., Mehl, M. R., & Josephs, R. A. (2007). Hormones and personality: Testosterone as a marker of individual differences. Journal of Research in Personality, 41, 126–138. Shirtcliff, E. A., Granger, D. A., & Likos, A. (2002). Gender differences in the validity of testosterone measured in saliva by immunoassay. Hormones and Behavior, 42, 62–69.

79

Shirtcliff, E. A., Granger, D. A., Schwartz, E., & Curran, M. J. (2001). Use of salivary biomarkers in biobehavioral research: Cotton-based sample collection methods can interfere with salivary immunoassay results. Psychoneuroendocrinology, 26, 165–173. Stewart, A. J., & Chester, N. L. (1982). Sex differences in human social motives: Achievement, affiliation, and power. In A. J. Stewart (Ed.), Motivation and society (pp. 172–218). San Francisco: Jossey-Bass. Thompson, J. D. (1967). Organizations in action. New York: McGraw-Hill. Tremblay, R. E., Schaal, B., Boulerice, B., Arseneault, L., Soussignan, R. G., Paquette, D., et al. (1998). Testosterone, physical aggression, dominance, and physical development in early adolescence. International Journal of Behavioral Development, 22, 753–777. Udry, J. R., & Talbert, L. M. (1988). Sex hormone effects on personality at puberty. Journal of Personality and Social Psychology, 54, 291–295. van Honk, J., & Schutter, D. J. L. G. (2007). Testosterone reduces conscious detection of signals serving social correction. Psychological Science, 18, 663–667. Viau, V., & Meaney, M. J. (1996). The inhibitory effect of testosterone on hypothalamic–pituitary–adrenal responses to stress is mediated by the medial preoptic area. Journal of Neuroscience, 16, 1866–1876. White, R. E., Thornhill, S., & Hampson, E. (2006). Entrepreneurs and evolutionary biology: The relationship between testosterone and new venture creation. Organizational Behavior and Human Decision Processes, 100, 21–34. Winter, D. C. (1973). The power motive. New York: The Free Press. Woolley, A. W., Gerbasi, M. E., Chabris, C. F., Kosslyn, S. M., & Hackman, J. R. (2008). Bringing in the experts: How team composition and collaborative planning jointly shape analytic effectiveness. Small Group Research, 39, 352–371. Zaccaro, S., Blair, V., Peterson, C., & Zazanis, M. (1995). Collective efficacy. In J. E. Maddux (Ed.), Self-efficacy, adaptation, and adjustment (pp. 305–328). New York: Plenum Press.