The interactive influence of conscientiousness and openness to experience on dissent

Abstract

Recent research suggests that two psychological factors are necessary for group members to dissent: a capacity to form alternate perspectives on group issues, and a motivation to publicly express them. With respect to the influence of personality variables, Openness to Experience may predispose individuals toward the formation of alternate opinions, while Conscientiousness may motivate the articulation of deviant perspectives once they have been formed. Data across three studies evidenced this interactive pattern such that only among individuals high on Openness did Conscientiousness predict willingness to influence one's group, in particular via expression of dissenting opinions. Effects were not moderated by collective identification, indicating that these personality and social identity level variables can operate independently of one another.

Keywords:

• Dissent
• Personality
• Openness
• Conscientiousness
• Social identity
• Normative conflict model

Collective decision-making failures ranging from the abortive Bay of Pigs invasion to losses of NASA space shuttles to the current economic crisis are often characterized as resulting from a lack of dissent (Janis, 1972; Shiller, 2008; Sunstein, 2006). In each case, problems that seem obvious in hindsight went largely unarticulated among decision makers until it was too late. A great deal of research has sought to understand why non-normative perspectives often go unexpressed within groups. Influential psychological explanations cite processes of groupthink (Janis, 1972) or a spiral-of-silence (Noelle-Neumann, 1974), and posit that while informational pressures reduce the likelihood that group members will generate deviant perspectives, normative pressures prevent those who do hold different views from expressing them for fear of social costs, including ridicule, rejection, and ejection (e.g., see Sunstein, 2006).

A Social Identity Approach to Dissent

Recent research has, however, highlighted an alternate perspective: that there are likely to be circumstances in which group members consider and express non-normative opinions (e.g., Crane & Platow, in press; Hornsey, 2006; Hornsey, Smith, & Begg, 2007; Iyer, Schmader, & Lickel, 2007; Morton, Postmes, & Jetten, 2007; Packer, 2008; Roccas, Klar, & Liviatan, 2006; Sani & Todman, 2002). Inverting the groupthink logic suggests that two factors are likely to be necessary for group members to dissent: (a) a capacity to form and hold alternate perspectives on group-related issues, and (b) a motivation to express said perspectives. My colleagues and I have examined these factors within a social identity framework termed the Normative Conflict Model, which posits that high levels of identification with a group are sufficient to motivate expression of deviant perspectives if a member believes that doing so will benefit the group (Packer, 2008, 2009; Packer & Chasteen, 2010). As such, although strongly identified members tend to conform more to group norms than weak identifiers (e.g., Johnson & White, 2003; Spears, Doosje, & Ellemers, 1997; Terry & Hogg, 1996), recent studies have shown that when norms are perceived to be collectively harmful, this pattern reverses and strong identifiers dissent more than weakly identified members (Packer, 2009; Packer & Chasteen, 2010; see also Crane & Platow, in press). However, consistent with the notion that dissent is a collectively oriented act for strong identifiers, perceptions of personal harm do not increase expressions of dissent among these group members.

The Potential Role of Personality Variables

Importantly, although collective identification provides one source of motivation for expressions of dissent and may also increase members’ vigilance with respect to group-relevant problems (Packer, 2009), this social identity variable is not the only possible predictor of the two factors posited above as necessary for deviant opinion expression. There may well be other psychological variables that influence the ability to form and maintain alternate perspectives, and that motivate opinion expression (e.g., Hirschman, 1970; Santee & Maslach, 1982). In the current paper I examine the influence of two personality variables, Conscientiousness and Openness to Experience, which may act independently of or potentially in interaction with collective identification to predict dissent within groups. Both are components of the widely used “Big Five” model, which has identified five over-arching dimensions of personality on the basis of factor-analytic studies of trait descriptions (Digman, 1990; McCrae & Costa, 1987). Openness to Experience captures “the extent to which individuals are imaginative, sensitive to aesthetics, curious, independent thinkers, and amenable to new ideas, experiences and unconventional perspectives” (George & Zhou, 2001, p. 514). Unsurprisingly, Openness has been linked to innovation and creativity in a variety of domains, including scientific and artistic endeavors (Feist, 1998; George & Zhou, 2001; McCrae, 1996). As such, this variable is a plausible predictor of the capacity to form and maintain alternate perspectives on group behavior, and we might well expect Openness to bear some relationship with dissent.

Conscientiousness, on the other hand, captures the extent to which people “actively plan, organize and carry out tasks. Conscientious individuals are goal driven … and exhibit high commitment to goal achievement” (Stewart, 1999, p. 960). On the surface, Conscientiousness may seem an unlikely contributor to expressions of dissent—for instance, whereas Openness positively predicts creativity and innovation, Conscientiousness is negatively associated with these outcomes (Feist, 1998; George & Zhou, 2001). That said, however, Conscientiousness may still have an important role to play in expressions of dissent; specifically, in motivating the articulation of alternate perspectives once they have been formed. Critically, characteristics subsumed within the Conscientiousness factor (e.g., organizational ability, commitment to goal achievement) are theoretically independent of individuals’ goals (Stewart, 1999), which in some circumstances or among some individuals may be determined by external rules and social norms, but in other circumstances or among other individuals may be internally/self-determined. Consistent with the latter possibility, Barrick, Mount, and Strauss (1993) found that Conscientiousness was positively associated with autonomous goal setting (i.e., determining one's own goals; see also Judge & Ilies, 2002); further, a meta-analysis by Barrick and Mount (1991) found that Conscientiousness was positively associated with job performance across five highly varied occupational groups, including occupations in which work-related goals are likely to be determined by rules and norms (e.g., police, sales) and occupations in which goals are likely to be self-determined (e.g., managers, professionals).

Although Conscientiousness has sometimes been equated with conformity (see Barrick & Mount, 1991, for discussion), the relationship between this personality trait and conformity to group norms is in actuality likely to be moderated by other variables that shape individuals’ attitudes toward and goals with respect to the norms in question. Thus, Conscientiousness may interact with another variable such as Openness that predisposes individuals toward deviant perspectives on group-related issues, such that dissent is most likely among individuals high in both traits. To the extent that an individual forms a deviant opinion that he/she believes to be correct, helpful, and/or moral, the more Conscientious (that is to say, organized and goal-oriented) he/she is, the more likely he/she may be to attempt to persuade others of its virtue. Expressed slightly differently, although individuals high in Openness may be more likely than others to encounter or think up deviant perspectives, their opinions may go unarticulated without the requisite level of drive and discipline to do so.

I will report evidence of an interactive relationship between Conscientiousness and Openness in three datasets. In each case these personality variables were measured as part of studies designed primarily for other purposes; nevertheless, a clear and theoretically meaningful pattern emerges across analyses. Importantly, these data also allow for investigation of relationships between collective identification (a social identity variable) and personality traits. There are at least three possible hypotheses:

1. Functional antagonism: Self-categorization theory posits that when individuals identify with a group, they undergo a process of depersonalization in which the sense of self transitions from a personal to a collective level of identity (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Depersonalization involves a shift in the self-concept from “my” idiosyncratic traits to “our” shared characteristics, and thus accounts for the standard positive association between group identification and conformity to group norms (e.g., Abrams, Wetherell, Cochrane, Hogg, & Turner, 1990; Terry & Hogg, 1996). Consideration of the self-categorization process might lead to the conjecture that (in the relevant group context) identified group members abandon idiosyncratic distinctions and adopt a uniform set of characteristics consistent with an ingroup prototype. In this sense, personal and collective levels of identity have been described as functionally antagonistic to one another (Turner et al., 1987). One might thus predict an interaction between personality variables and identification, such that personal-level variables predict behavior among weakly identified group members (those who have not fully adopted the group's identity), but not among strongly identified group members, who operate instead in terms of collectively shared characteristics (e.g., Verkuyten & Hagendoorn, 1998).

2. Independence: Applied broadly, however, the above hypothesis may reflect a too literal interpretation of the self-categorization process. Groups do indeed have prototypes that prescribe common patterns of behavior among identified members; however, it is unlikely that these prototypes stipulate all aspects of behavior. As such, there may still be room for systematic individual-level variation among identified group members as a function of their idiosyncratic personality traits. A group prototype may, for instance, have a norm of cooperation that influences responses of identified members, but have little to say about the extent to which typical group members are Open to Experience. Thus, although we would expect identified group members to behave cooperatively, other aspects of their behavior might continue to vary as function of trait level Openness. According to this hypothesis, collective identification and at least some personality variables may operate independently of one another.

3. Functional agonism: Finally, an interaction opposite to the functionally antagonistic pattern described above is not inconceivable. If identification is understood in terms of engagement with a group (e.g., Tyler & Blader, 2003), we might expect strong identifiers to be the most likely of all members to exhibit certain idiosyncratic characteristics, at least to the degree that doing so is perceived to benefit the group. In a social exchange type formulation in which membership rewards (e.g., status) are presumed to be based, at least in part, on contributions (Anderson & Kilduff, 2009; Blau, 1964; Homans, 1961; Kelley & Thibaut, 1978), individuals for whom membership in the group is important may be particularly interested in proving themselves to be valuable group members and distinguishing themselves in ways that benefit the collective. As a result, the manner in which strongly identified members act within the group may be influenced by their beliefs about what unique characteristics they can contribute (see also Harkins & Petty, 1982; Postmes, Haslam, & Swaab, 2005; Postmes, Spears, Lee, & Novak, 2005).

Study 1

As an initial investigation, this study examined the interactive influence of Openness and Conscientiousness on group members’ beliefs that they had a responsibility to influence their group regarding important issues. These measures were administered as part of a larger study investigating attitudes toward gay marriage in Canada. I predicted that Conscientiousness would positively predict members’ feelings of responsibility to influence their group to the extent that they were high (but not low) on Openness.

Methodological note

As noted above, this and all subsequent studies were designed primarily for other purposes; for this reason, personality measures were administered at the end of each session. Limitations of this approach are outlined in the General Discussion.

Methods

Participants

Participants were 106 introductory psychology students at the University of Toronto who had lived in Canada for at least 5 years (M _age = 20.19, 70% female). They received course credit for participation.

Procedure and measures

In this and all subsequent studies, participants were run in groups of up to six, and measures were administered (individually) on computers using MediaLab presentation software. In each case participants were told that the studies were examining attitudes about issues related to groups to which they belonged. In the current study participants rated their identification with Canada at the beginning of the session. In all studies, collective identification was assessed using six items commonly employed in the literature (e.g., “I am proud to live in Canada”, “Being part of Canada is an important part of my identity”, “I feel an attachment to other people in Canada; 1 = strongly disagree, 6 = strongly agree; Cronbach's α =.93; see Ashmore, Deaux, & McLaughlin-Volpe, 2004).

Subsequently, as part of a larger questionnaire on attitudes toward gay marriage, participants rated the extent to which they felt a responsibility to influence their group. There were four such items (e.g., “I believe that I have some responsibility for influencing the way that other people in Canada think about important issues”) to which participants responded on a 6-point scale from 1 (strongly disagree) to 6 (strongly agree). A mean score was taken, with higher values indexing greater felt responsibility to influence the group (α =.89).

In this and all subsequent studies, Openness to Experience and Conscientiousness (as well as the other three traits) were measured at the end of each session using the Big Five Inventory (John, Donahue, & Kentle, 1991). Participants rated the extent to which behavioral characteristics related to these personality dimensions were self-descriptive (e.g., “is curious about many things”, “does a thorough job”; 1 = not at all descriptive of me, 5 = very descriptive of me). Negatively worded items were reverse-scored and means were calculated, such that higher scores indicated higher levels of each construct (Openness α =.78; Conscientiousness α =.82).

Results

Descriptive statistics and bivariate correlations among predictors for this and all subsequent studies are presented in Table 1. Hierarchical linear regression was used to test hypotheses. Mean-centered main effect terms were entered in Step 1, and their two and three-way interaction terms were entered in Steps 2 and 3 respectively. Unstandardized regression coefficients for all critical tests and any other significant effects are reported in the text. There were significant main effects of identification (b = .39, t = 4.98, p <.0001), Openness (b = .50, t = 3.79, p <.001), and Conscientiousness (b = .46, t = 3.79, p <.001), such that all three variables positively predicted the extent to which individuals felt a responsibility to influence their group. Critically, however, the main effects of the personality variables were moderated by a significant interaction between them (b = .37, t = 1.95, p =.05). The interaction was decomposed by examining the relationship between Conscientiousness and felt responsibility to influence the group at levels one standard deviation below and above the mean of Openness. Consistent with predictions, Conscientiousness was positively related to responsibility to exert influence among individuals high in Openness (b = .62, t = 4.19, p <.0001), but was not among individuals low in Openness (b = .21, t = 1.15, p =.25). Thus, Conscientiousness was predictive of beliefs about responsibility to exert influence only among those individuals whose level of Openness would likely predispose them toward alternate perspectives—those, in other words, who would likely have non-normative views to present.

Table 1 Descriptive statistics and bivariate correlations among predictors in Studies 1–3

	Study 1			Study 2			Study 3
	ID	O	C	ID	O	C	ID	O	C
Mean	4.72	3.51	3.26	5.06	3.56	3.57	4.41	3.56	3.42
(SD)	(1.08)	(.62)	(.67)	(.89)	(.61)	(.73)	(1.04)	(.57)	(.71)
Identification	1			1			1
Openness	.17	1		.04	1		.19*	1
Conscientiousness	.24*	−.14	1	.12^+	.50**	1	.22**	.14	1
^+ p <.10, *p <.05, **p <.01.

The interactive effect of Conscientiousness and Openness to Experience was not moderated by participants’ levels of collective identification (b = .04, t =.25, p >.80),1 nor were any other interactions significant. The lack of a three-way interaction is contrary to the functional antagonism and functional agonism hypotheses outlined above, and is most consistent with the hypothesis that these personality and social identity variables operate independently.

Study 2

Study 2 examined dissent more directly by eliciting actual behavior in a computerized chat-room, testing the interactive influence of Openness and Conscientiousness on ostensibly public expressions of opinion regarding a group norm.

Method

Participants

Participants were 33 introductory psychology students at The Ohio State University (M _age = 18.81, 40% female) who participated for course credit.

Procedure and measures

Participants started the session by completing the six-item measure of identification described in Study 1, applied to their university (e.g., “Being a student at OSU is an important part of my identity”; α =.90).

Participants read a short passage about a pro-alcohol norm among their fellow students, which concluded with the statement “it seems fair to say that there is a culture of alcohol use on the OSU campus.” Participants then had two opportunities to express opinions regarding alcohol use. Following Packer (2009; Packer & Chasteen, 2010), participants were led to believe that they would join an online chat-room with three fellow students to discuss their opinions regarding alcohol use among students. Upon entering the chat-room, participants were first asked to provide a public rating of their attitude toward alcohol use on campus using a scale from 1 (I am not at all concerned this is a problem) to 6 (I am very concerned that this is a problem), with the belief that their ratings would be visible to their peers. Having rated their own attitudes, participants saw the ostensible attitudes of others, pre-programmed by the experimenter to be consistent with the pro-alcohol norm (1s and 2s on the scale). Participants then read three statements supposedly written by the others, all of which expressed relatively positive attitudes toward drinking. For instance, one statement read: “drinking rocks!!! i’d rather be at a school that knows how to party than somewhere lame where everyone takes themselves too seriously … events are lots more fun when there's some alcohol. i’m not saying everon should get drunk or that everyone has to drink, but its part of the OSU experience.” Participants were asked to choose one of the statements and write a public response to it that would be seen by the other students. No instructions were given as to the form their response should take, and participants were free to express agreement or disagreement.

Openness to Experience (α =.72) and Conscientiousness (α =.85) were measured at the end of the session.

Results

Public ratings of concern

Participants’ public attitude ratings were analyzed following the same procedure as in Study 1.2 There was a marginal main effect of Conscientiousness (b = .67, t = 1.87, p =.07), and no main effects of Openness or identification. The lack of an identification effect in this study is not surprising given that past research has not observed a main effect of identification on dissent, but rather an effect only among members who perceive a norm to be harmful to the group (see Study 3 for investigation of this factor). Critically, the main effect of Conscientiousness was qualified by a significant interaction with Openness (b = 1.82, t = 3.96, p <.001). As shown in Figure 1, panel A, Conscientiousness was positively associated with public expressions of concern regarding alcohol use among highly Open individuals (1 SD above the mean: b = 1.50, t = 3.57, p <.01), but was not related to public concern among individuals low in Openness (1 SD below the mean: b = −.01, t = −.03, p >.90). Consistent once again with the independence hypothesis, the interaction of the personality variables was not qualified by level of collective identification (b = .19, t =.27, p >.70), nor were any other interactions significant.3

Figure 1. (a) Public ratings of concern regarding pro-alcohol norm predicted as a function of the interaction between Openness and Conscientiousness. (b) Odds of expressing disagreement with chat-room interlocutors predicted as a function of the same interaction. Note that odds below 1 indicate a greater likelihood of agreeing than disagreeing; odds above 1 would indicate a greater likelihood of disagreeing than agreeing.

Expressions of disagreement in written responses

Participants’ written responses to others’ statements were rated by two independent coders for the presence vs absence of disagreement; in other words, whether or not they explicitly expressed disagreement with the pro-alcohol views of their interlocutors. Inter-rater reliability was high (agreement rate = 97%), and inconsistencies were resolved by a third coder. Overall, 32.25% of participants expressed disagreement. Logistic regression analyses were used to predict disagreement from the main and interactive effects of identification, Openness and Conscientiousness; predictors were entered in the same series of steps as above. There were no main effects, but the two-way interaction between Openness and Conscientiousness was again significant (b = 2.59, Wald = 4.09, p <.05). Figure 1, panel B, plots the odds of expressing disagreement as a function of Conscientiousness and Openness—as is readily apparent, expressions of disagreement were most likely among participants high in both Openness to Experience and Conscientiousness. Among these individuals, the odds of expressing disagreement (versus not) were close to even; among all others, the odds were much more likely that they would refrain from disagreement. There were no other significant interactions.

Study 3

Packer and Chasteen (2010) found that strongly identified group members expressed heightened dissent (relative to weak identifiers) in conditions that asked them to reflect on how a norm could harm their group, but not in conditions that asked them to reflect on how the norm could be harmful to individuals. This finding is important for a social identity account of dissent because it suggests that dissent among strong identifiers is a psychologically loyal act, motivated by concern for collective rather than individual interests. Importantly, however, we would not necessarily expect this dissociation to occur when predicting dissent from individual-level personality variables. To the extent that collective identification and personality factors operate independently, it is possible that whereas a positive effect of identification on dissent would be manifest only in conditions that highlight collective harms, Openness and Conscientiousness would exert an interactive influence in response to both collective and individualistic forms of harm. That is, unlike strongly identified members, individuals high in Openness should be receptive to ideas regarding both collective and individualistic forms of normative harm and may be motivated to express contrarian opinions regarding both to the extent that they are also high in Conscientiousness.

In order to test these predictions Study 3 combined a subset of data from Packer and Chasteen (2010, Study 2a) with data from another unpublished experiment (run concurrently) which employed the same design but a different group norm. Whereas our previously reported analyses investigated the relationship between collective identification and dissent, the current analyses focus on the role of personality variables and their potential interaction with identification.

Methods

Participants

Participants were 145 introductory psychology students at the University of Toronto (M _age = 19.38, 74% female) who participated for course credit.

Procedure and measures

Using the six-item scale from Studies 1 and 2, participants rated their level of identification with the University of Toronto (e.g., “I am proud to be a student at the University of Toronto”; α =.89).

Participants were randomly assigned to receive information regarding one of two group norms. They were told that a majority (84%) of students at their university either held relatively positive attitudes toward plagiarism or were in favor of placing limits on freedom of speech on campus. Participants in both cases were then randomly assigned to conditions that asked them to generate arguments as to how the normative opinion could either harm the group as a whole or harm individuals within the group. Afterwards, participants were asked to indicate their attitudes toward the social norm, as well as their willingness to publicly express dissent regarding the norm in a variety of ways, including giving a speech and writing to a student newspaper. Seven such items were rated on a 1 to 6 scale (e.g., 1 = not at all willing, 6 = very willing) and were combined to create a composite measure of non-conformity, such that higher scores represented greater willingness to express dissenting views regarding a group norm (α =.76).

Openness to Experience (α =.70) and Conscientiousness (α =.82) were measured at the end of the session.

Results

Hierarchical linear regression was used to test predictions.4 Condition was added as a predictor, and was dummy-coded (1 = collectively harmful, 0 = individually harmful). Two, three, and four-way interaction terms were entered sequentially. Effects were not moderated by the group norm employed and this variable is not reported further. Conscientiousness exerted the only significant main effect, such that individuals higher in Conscientiousness expressed greater willingness to dissent (b = .27, t = 2.64, p <.01).

The interaction between condition and identification reported by Packer and Chasteen (2010) in a subset of this data was maintained in the current larger sample. In the current dataset the condition-by-identification interaction was significant at p <.01, such that collective identification positively predicted willingness to dissent among participants who had thought about how a norm might harm their group (p <.001), but did not among those who had thought about how a norm might be individualistically harmful (p >.20).5

The two-way interaction between Conscientiousness and Openness was also significant (b = .49, t = 2.70, p <.01). Consistent with the pattern described throughout, whereas Conscientiousness was not associated with dissent among individuals low in Openness (1 SD below the mean: b = .00, t = −.02, p >.90), Conscientiousness and reported willingness to dissent were positively associated among individuals high in Openness (1 SD above the mean: b = .46, t = 3.85, p <.001). Critically, and as hypothesized, the interaction between Openness and Conscientiousness was not moderated by condition (b = .08, t =.20, p >.80). Follow-up analyses showed that Conscientiousness was positively associated with dissent among individuals high in Openness in both the collective and individual harm conditions (ps <.02); further, Conscientiousness and dissent were unrelated among individuals low in Openness in both conditions (ps >.70). Thus, whereas the positive effect of collective identification on dissent was contingent on a group norm being regarded as collectively harmful, the effect of these personality variables was not. Individuals high in both Openness and Conscientiousness were, in other words, responsive to the possibility of both collective and individualistic forms of harm. The interactive effect of Conscientiousness and Openness was not qualified by collective identification (b = −.08, t = −.45, p >.60) and no other interaction terms were significant, indicating once again that the effect of these personality variables was not contingent on being either strongly or weakly identified with the group.

General Discussion

Across three datasets, Conscientiousness and Openness to Experience exerted an interactive effect on willingness to influence one's group, in particular via expression of dissenting opinions. In Study 1, Conscientiousness predicted feelings of responsibility to influence the group only among those individuals who reported being high in Openness; that is, among individuals likely to encounter situations in which their opinions deviate from those of the group and in which exertion of influence is necessary. Studies 2 and 3 extended this finding to show the same interactive pattern on reported willingness to dissent, as well as on actual expressions of disagreement regarding group norms. This recurrent pattern is consistent with the prediction that two factors are necessary for expression of dissent in groups: a capacity for individuals to form and maintain alternate perspectives, and a motivation to try to shift group opinion in the direction of these perspectives once they have been formed. Employing a social identity approach, past research has shown that identification with a group provides one source of motivation for dissent—specifically among members who perceive a group norm to be harmful to the collective (Packer & Chasteen, 2010)—and has further shown that identification may make group members vigilant to possible collective harms (Packer, 2009). The current research demonstrates that personality variables can also influence these two factors and thereby affect non-conformist opinion expression within groups.

These data further indicate that, at least with respect to Openness and Conscientiousness in these particular groups, social identity and personality variables can operate independently. The absence of significant interactions between identification and the personality variables in any of these studies suggests that neither the functional antagonism nor the functional agonism hypothesis adequately captures the relationship between collective and personal aspects of identity in this context. Identifying with a group did not suppress or enhance the influence of these personality traits; as such, there was room for idiosyncratic personality variables to exert systematic influence on willingness to express dissenting opinions among all group members, whether they were strongly or weakly identified.

With respect to formulating a broad characterization of the relationship between social identity and personality variables, I suspect that it is unlikely be a question of which of the three hypotheses (i.e., functional antagonism, independence, functional agonism) are correct, but rather a question of when each of them apply. It is likely that all three patterns apply in different contexts, and an important direction for future research will be to specify the conditions under which each hypothesis operates. The functionally antagonistic pattern may be most likely when a certain character trait is central to a group prototype. For instance, in a group in which the prototypic member is expected to behave in an agreeable fashion, Agreeableness at the individual level is only likely to predict behavior among weakly identified members; strong identifiers should behave agreeably as a function of their level of identification, regardless of the extent to which they hold this trait as individuals. As such, the more domains of thought, feeling, and action for which a group has specified norms, the greater the shift in characteristics strongly identified members are likely to display when they transition from a personal to a collective level of identity (and the less predictive individuals’ personality profiles are likely to be having made such a shift). In contrast, the independence pattern is more likely to occur for traits about which a group has not specified a clear norm, or for traits on which it is normative to vary (see Hornsey & Jetten, 2005; Jetten, Postmes, & McAuliffe, 2002). Finally, the functionally agonistic pattern may be most likely in situations when members are motivated to make an individual impression on their group. This may be particularly likely, for instance, during group formation and/or when hierarchies are being negotiated (e.g., Postmes et al., 2005; Reicher & Hopkins, 2001). In this situation, those members for whom the group is especially important may be the most likely to display unique characteristics that they believe will contribute to the benefit of the collective and thereby garner them membership-related rewards (e.g., status, leadership positions; Blau, 1964).

The current research was limited by its reliance on datasets collected primarily for other purposes and the fact that personality measures were always collected at the end. Given the collective contexts in which these scales were administered, it is possible that when participants were asked to report their personality traits, they answered questions not in terms of who they were as individuals, but rather in terms of who they were as group members. However, to the extent that this should be more likely for strong identifiers, the independence observed between identification and personality traits indicates that it was probably not a significant issue. Nevertheless, future research would do well to pre-measure personality variables prior to putting participants into a collective context. In addition, while the current research employed measures of actual opinion expression in Study 2, the relatively anonymous chat-room environment may have facilitated expressions of dissent by reducing the social costs anticipated to result from non-conformity (Spears, Lea, Corneliussen, Postmes, & Ter Haar, 2002). Although participants in Study 3 reported being willing to dissent in more public contexts (e.g., by giving a speech), direct observation of dissenting behavior in riskier situations is an important next step.

Finally, I have suggested that Conscientiousness serves a motivational function when it comes to dissent. Importantly, however, as with all the Big Five factors, this personality trait is composed of distinct sub-components (e.g., Barrick & Mount, 1991; Stewart, 1999). An achievement sub-component corresponds directly with the volitional aspect of Conscientiousness that I have described. However, Conscientiousness also contains a sub-component related to order and organization, which corresponds less to motivation and more to abilities relevant to enacting goals. As such, the observed relationships between Conscientiousness and expressions of dissenting opinions among highly Open individuals may be due to its volitional characteristics and/or its ability-related characteristics (e.g., the ability to organize one's thoughts into a coherent expression, develop a plan regarding how to persuade others, etc.). It will be important in future research to tease apart the relative contributions of these sub-components.

Conclusion

Although lack of dissent remains a critical problem in collective decision making, this research suggests that there is reason for optimism. Individuals who are inclined to generate alternate perspectives may, if they also have the requisite motivation, express non-conformist opinions regarding group behavior. Adding to past research showing that collective identification plays a role in motivating dissent, the current data illustrate that individuals high in Openness to Experience and Conscientiousness are also more likely than others to dissent. Importantly, as demonstrated by Study 3, the dissenting viewpoints articulated by individuals with these personality traits are likely to be broader in scope than those motivated by high levels of identification. Whereas dissent by strongly identified group members is triggered specifically by perceptions that norms are collectively harmful (Packer & Chasteen, 2010), highly Open and Conscientious individuals are willing to express dissent in response to both collective and individual forms of harm. In either case, however, alternate perspectives are given voice, and the challenge for groups is to capitalize on their articulation in order to improve collective decision-making.

Notes

¹Nor was the interactive effect moderated by participants’ attitudes toward gay marriage (p >.60), the focus of the study in which these measures were administered.

²Reported analyses excluded two multivariate outliers with Cook's Ds greater than 4/n (i.e., 4/33 =.12). Including these participants affected significance levels but not the pattern of results.

³It appeared that there could be a trend for an interaction between identification and Conscientiousness (b = −.72, t = −1.61, p =.12). Exploratory investigation revealed a weak pattern consistent with the functional antagonism hypothesis, such that the effect of Conscientiousness on dissent may have been somewhat stronger among weakly identified (b = 1.02, p =.03) than strongly identified (b =.84, p =.10) members.

⁴Reported analyses excluded one outlier on identification and one on Openness (<3 SD from these means). However, including these participants did not change the results.

⁵To avoid accidental inclusion of overlapping data in future meta-analyses, regression coefficients for this analysis are not included in text. For interested parties, however, these relationships in the current dataset were as follows: condition by identification interaction, b = .46, t = 3.12, p <.01; effect of identification on dissent in collective harm condition, b = .31, t = 3.76, p <.001; effect of identification on dissent in individualistic harm condition, b = −.15, t = −1.29, p >.20.