You are either for us or against us: When are ambivalent in-group members sanctioned?

Abstract

The subjective group dynamics model describes conditions for sanctioning deviant in-group members. A description of a new or established group member's (the target) attitude toward “Obamacare” was provided to 136 Republican participants. The target expressed a group-normative, ambivalent, or deviant opinion. Participants indicated the self-relevance of Obamacare, and evaluated the target. Neither target status nor attitude affected evaluations for non-vested participants; however, highly vested participants evaluated new ambivalent targets more favorably than established ambivalent targets (p < .05): derogation or sanctioning of ambivalent and deviant targets, that is, was moderated by evaluators’ vested interest and longevity of the target's group membership.

Keywords:

• deviance
• ambivalence
• subjective norms
• vested interest
• in-group leniency

In the absence of objective means for reality testing, social consensus may be used to resolve attitudinal ambiguity (Festinger, 1954; Suls & Wheeler, 2012). Consistent with this position, Hogg (2007, 2012) found that people often look to subjective social group norms to resolve uncertainty about the self-concept. Information that renders norms unclear leaves uncertain individuals without social means to validate their attitudes or actions. Often, in-group members who deviate from subjective norms, thereby threatening their clarity or integrity, are derogated or ostracized (Marques, Abrams, & Serôdio, 2001; Marques & Páez, 1994).This black sheep effect (Marques, Yzerbyt, & Leyens, 1988) is central to the subjective group dynamics model (SGD), which describes reactions to in-group deviance on a group-defining trait. Because traits that define our social groups often are difficult to validate objectively, any model like the SGD must account for group members who are “on the fence” about identity-relevant attitudes. The current study was designed to test the conditions under which in-group members who were ambivalent about an important feature of group orthodoxy would be treated as black sheep and sanctioned, and to expand the SGD by considering the role that vested interest (VI) plays in the evaluation of attitudinally diverse in-group members.

Uncertainty regarding the self-concept is aversive (Lopes, 1987), and so individuals seek to reduce it. Identifying with distinctive and clearly defined social groups is an effective way to do so (Hogg, 2000). When people categorize themselves into social groups, they compare their own attitudes and behaviors against those embodied by the in-group prototype (the hypothetical ideal in-group member); thus, rendering the prototype unclear undermines its subjective validity (Hogg, Sherman, Dierselhuis, Maitner, & Moffitt, 2007). According to Pinto, Marques, Levine, and Abrams (2010), the black sheep of social groups are ostracized because in-group members who deviate from the norm destabilize the in-group definition, rendering in-group membership unreliable as a means of reducing self-uncertainty.

Subjective group dynamics

A foundational tenet of SGD is that the subjective validity of in-group norms is maintained through intra-group differentiation (distinguishing normative in-group members from deviant ones) and inter-group differentiation (distinguishing in-group from out-group members). The desire to validate subjective norms and uphold intra-group similarity and inter-group differences is believed to motivate sanctioning or derogation of deviant in-group members (Marques, Abrams, Paez, & Martinez-Taboada, 1998; Marques et al., 1988). Corollaries of this assumption are that in-group members should be scrutinized more closely and judged more harshly than out-group members for straying from established norms. Out-group members are expected to disagree with in-group norms, and their thoughts or actions are therefore “less crucial for the intragroup validation of social identity” (Pinto et al., 2010, p. 107). In-group members, however, are crucial to the validation of social identity, and for this reason they are subject to the black sheep effect, whereby they are judged more harshly than out-group members who hold the same deviant attitude or belief.

Whereas the black sheep effect refers to the relative evaluation of in-group versus out-group members who both hold an in-group deviant position, SGD more generally also posits that in-group deviants may be differentially evaluated relative to one another. Consequently, recent research into SGD considers factors that moderate the derogation of in-group deviants, such as membership status of the deviant. For example, Pinto et al. (2010) found participants upgraded new in-group deviant members relative to deviant “full” or established members, affirming that “people who have achieved full member status … who betray the group's standards … undermine the subjective validity of other in-group members’ social identity [and] should be more strongly derogated than … less central members” (p. 107). Their study suggests that time spent in the group might be an especially important cue to those evaluating ambivalent in-group members.

Research on the social identity theory of leadership shows that like leaders, central group members are perceived as more prototypically in-group and more likely to uphold subjective norms (Hogg, van Knippenberg, & Rast, 2012). Furthermore, in some cases leaders are allowed more freedom to enact non-normative behavior. Abrams, Randsley de Moura, Marques, and Hutchison (2008) found that participants who read descriptions of a non-normative in-group member granted hypothetical future leaders (British politicians) greater license to innovate normative practice (i.e., change immigration laws in the UK) than deviant non-leader members or established leaders; deviance was tolerated only in future in-group leaders or soon-to-be prototypical in-group members, and not in either peripheral or fully socialized prototypical in-group members. In relation to SGD, new in-group members are easily described as soon-to-be prototypical, and therefore should be given innovation credit relative to established, already-prototypical members (Pinto et al., 2010). Whether defined as time spent in the group or prototypicality, membership status may determine whether a deviant (or innovator) is deemed a threat and sanctioned, or a novice whose misplaced opinions are understood if not accepted.

Moderators of SGD

Moderators of the processes described by SGD either augment the self-relevance of the subjective norm or the self-uncertainty regarding the norm. For instance, social identification has to do with the extent to which one's self-concept is tied to a particular group membership. When identification is high, there is a strong link between in-group definition and self-definition (Tropp & Wright, 2001); therefore, anything that undermines in-group definition should be threatening to the self, and research has shown that identification indeed intensifies derogation of deviant in-group members (Bègue, 2001) and reactions to threats to in-group distinctiveness or status (Alvaro & Crano, 1997; Spears, Doosje, & Ellemers, 1997). Another useful moderator, a focus of the present investigation, is vested interest (VI), which refers to the extent to which an attitude object is perceived as important and as having high “hedonic relevance” (Sivacek & Crano, 1982, p. 110) for the person espousing the corresponding attitude (Crano, 1995). As such, VI should accentuate reactions to in-group members who threaten the validity of a normative attitude.

Vested interest has been identified as an important moderator of attitude-behavior consistency (Johnson, Siegel, & Crano, 2014; Sivacek & Crano, 1982; Thornton & Tizard, 2010). While VI is not considered a predictor of attitudes in and of itself, highly vested attitudes have been found to be more strongly related to attitude-consistent actions than non-vested attitudes. For example, across a host of social policy issues, Lehman and Crano (2002) found that individuals for whom a legislative policy had important personal consequenceswere more likely to act on their attitudes than those for whom the issue was not vested. Further, research by Johnson et al. (2014) indicated that highly vested (vs. non-vested) people were more likely to take action on issues that directly affected their loved ones. Extending these findings, sanctioning or derogation of other in-group members by normative in-group members is expected when the others espouse counter-normative positions. VI is expected to bolster the deviance-derogation relationship, with the more highly vested sanctioning deviants to a greater degree than non-vested individuals.

Because of the implications of VI for the evaluation of attitudinally non-normative in-group members, it is investigated in the current study as an independent variable. Attitude certainty is another aspect of the attitude that is relevant to the sanctioning of attitudinally deviant in-group members. Certainty is a metacognitive attribute of attitudes, a component of attitude strength, and refers to the felt correctness of one's attitude. Just as self-uncertainty motivates stricter adherence to prototypical practices (Hogg, 2007), both attitudinal uncertainty and VI should increase the need for a clear conception of the group's position, and consequently may lead to harsher sanctioning of in-group members who muddy this conception. Identification and attitude certainty, both theoretically important moderators of the perception of in-group deviants, were assessed in the present research.

Method

Overview

Most operational definitions of attitudinal ambivalence refer to an individual simultaneously holding positive and negative evaluations of an attitude object (Conner, Povey, Sparks, James, & Shepherd, 2003, p. 706). This definition guided the design of the current study (see also Hohman, Crano, Siegel, & Alvaro, 2014). The study was designed to expand SGD by illuminating the circumstances under which targets with ambivalent attitudes were perceived as deviants or were evaluated as similar to normative group members. Based on the earlier research, we predict that new and established opinion-deviant in-group members will be evaluated less favorably than normative new and full members. Further, we hypothesize that new in-group members who hold an ambivalent stance on a key plank of the in-group's belief system will be evaluated more positively than established members who express a similarly ambivalent attitude on the group-relevant issue. The former prediction follows from research (e.g., Pinto et al., 2010) suggesting that in-group deviants, on the whole, are disliked and sanctioned relative to normative in-group members. The latter prediction rests on the idea that attitudinally ambivalent new in-group members, because they have not been fully socialized to group norms but are at least somewhat normative in their attitudes, will be upgraded relative to established members, just as deviant, future leaders were upgraded relative to deviant, incumbent leaders (Abrams et al., 2008).

Design and participants

This study employed a 2 (target member status: new, established) × 3 (target attitude: deviant, ambivalent, normative) × 2 (vested interest: high, low) between-subjects factorial design. Self-designated Republican participants (N = 136) read a description of a target and evaluated him. The target's Republican Party membership status and attitude toward the Obama administration's Affordable Care Act, popularly called Obamacare, were manipulated independently. The dependent measure was participants’ evaluation of the target. All recruitment and participation occurred online: The sample of adults (61 female, 75 male), registered or identifying as Republicans or Tea-party members, were compensated ¢45 for their participation. The study was advertised through Mechanical Turk (a crowdsourcing website hosted by Amazon.com^®). It was described as a study of how “we understand and evaluate other people's politics.” Approximately 85% (n = 118) of the participants were Caucasian/White; 18 participants belonged to another ethnic category. Age and education level both appeared to be normally distributed (skewness = 0.20 and − 0.38, respectively), with the majority of participants being between the ages of 25 and 54 years, and having either some undergraduate experience, a Bachelor's, or an Associate's degree.

Procedure and measures

Mechanical Turk (MTurk) members (“workers”) can search through a database of surveys or other tasks, called HITs, and complete them for monetary compensation. Empirical evidence supports the use of MTurk in psychological research (e.g., Crump, McDonnell, & Gureckis, 2013). Members of MTurk who searched for HITs by key terms Republican, politics, health care reform, or Obamacare were directed to the description of this study and its inclusion criteria. Those who judged themselves eligible to participate followed a hypertext link to the study itself, and were asked to indicate their feelings toward health care reform and four other legislative issues: legalizing gay marriage, increased access for women to reproductive/family planning services, stricter gun control legislation, and providing education for non-citizen immigrants (1 Very negative, 9 Very positive). This operation was not used to assess participants’ attitudes, but rather to familiarize them with the attitude measure.

Participants read about the political background and attitudes of a 30-year-old Republican man named Jack S., the target of their evaluations. Depending on experimental condition, the target was described either as having been a Republican “since he was first old enough to vote” (12 years prior to the present study) or having been an Independent until registering as Republican for the 2008 presidential election; this constituted the manipulation of the target's membership status (new/established). As for his attitude toward health care reform, Jack was described as being either normative for a Republican (i.e., as opposed to Obamacare), as deviant for a Republican (as supporting Obamacare), or as ambivalent (“just as much for reforming the health care system as he is against it”).

These vignettes also described the arguments Jack used to support his deviant, ambivalent, or normative position. Only arguments rated seven or higher by all 24 Republicans sampled in a pilot study on both of two 11-point rating scales (relevant and convincing) were used in the main study (in which the pilot study respondents did not participate). The preliminary study ensured that the target's position would not be perceived to be so weak as to be regarded as irrelevant to in-group norms. Strong arguments for and against health care reform were both presented when the target was described as ambivalent. The order of presentation was counterbalanced across participants.

After the vignettes, participants viewed a carefully constructed screen-shot of the response scale the target had used to indicate his political attitudes. In the deviant target attitude condition, the photo indicated that Jack had marked the eighth of nine response options (the ninth option was labeled “Very positive”). In the normative attitude condition, the photo showed the second response option marked (the first option was labeled “Very negative”). In the ambivalent target attitude condition, both the second and eighth response options were checked. After reading the vignette and considering Jack's scaled attitude responses, participants were asked to evaluate him by indicating their “overall impression of Jack” (1 Very negative, 7 Very positive) and rating him on nine trait dimensions using 7-point semantic differential items (bad/good; foolish/intelligent; shy/sociable; lazy/ambitious; narrow-minded/open-minded; weak/strong; treacherous/trustworthy; mean/kind; immoral/moral) (α = 0.88). These bipolar adjectives were adapted from Asch and Zukier (1984, p. 1231). The nine semantic differential items and the single global evaluative item were combined into a single scale (α = .89). Participants then completed measures of in-group identification, attitude certainty, and VI.

Four items (adapted from Hogg et al., 2007) assessed each participant's strength of identification with the Republican Party. Participants rated on 9-point scales the importance of their party membership, their level of identification with the party, the extent to which they felt pride in being a party member and the extent of felt similarity to other members (α = .92). Four items adapted from Petrocelli, Tormala, and Rucker (2007) measured attitude certainty, with participants rating the extent to which they perceived their own position on Obamacare to be rational, stable, clear, and correct (e.g., 1 Not clear at all, 7 Perfectly clear) (α = .84). A targeted measure of VI used in earlier research by Lehman and Crano (2002) employed an item that tapped the extent to which participants felt they would be affected in their access to health-care services by the Affordable Care Act's passage (1 Not at all, 7 Extremely). In addition to assessing social identification, attitude certainty, and VI, six items taken from Knowles, Lowery, and Schaumberg (2010) assessed participants’ attitudes toward health care reform legislation. This scale measured the extent to which six possible outcomes of health care reform (taxes being raised, a move from capitalism to socialism, increased cases of euthanasia, benefits to undeserving populations, health care rationing, and long delays in receiving medical treatment) were worrisome or concerning (1 Not concerned at all, 5 Extremely concerned) (α = .85). Higher values on these items correspond to more negative evaluations of Obamacare. Finally, participants indicated their age, gender, educational level, and ethnicity, and were compensated for participation.

Results

Identifying covariates

A 2 (target membership status: new, established) × 3 (target attitude: deviant, ambivalent, normative) between-subjects multivariate analysis of variance on the three measured moderator variables (identification, participant attitude, and attitude certainty) revealed no significant effects. However, participants’ attitudes were significantly and negatively correlated (r = − .22, p = .01) with target evaluation (see Table 1) and therefore were included as a covariate in further analyses.

Table 1 Inter-correlations among dependent variable, demographics, and measured covariates.

Measure	α		1	2	3	4	5	6	7
1. Evaluation of target (10 items)	.89	4.86	–
2. Age		–	–	.11	–
3. Education		–	–	− .05	.13	–
4. Vested interest (1 item)	–	4.25	− .02	− .01	.004	–
5. Participant attitude (6 items)	.85	3.82	− .22^*	.14	.04	.23^**	–
6. Identification (4 items)	.92	5.67	− .08	.20^*	.07	.27^**	.56^**	–
7. Attitude certainty (4 items)	.84	5.43	.08	.08	.06	.32^**	.30^**	.48^**	–

Note. ^*p < .05, ^**p < .01.

Also noteworthy was the positive relation between social identification and the VI measure (r = .23, p = .007) suggesting that Obamacare was an identity-relevant attitude object, as the extent to which participants identified with the Republican Party was linked with the perceived self-relevance of the issue. The strongest correlation involved the link between participants’ attitudes toward Obamacare and their level of social identification with the Republican Party (r = − .56, p < .001), further supporting the proposition that this attitude object was important to participants’ perceptions of themselves as members of the Republican Party. Breaking down the correlations by experimental condition, of note are the statistically significant inverse associations between: participants’ attitudes toward the Affordable Care Act and evaluation of deviant, new (r = − .66, p < .001), deviant, full (r = − .55, p = .01), and ambivalent, full (r = − .57, p = .01) in-group targets. This suggests that attitudes toward “Obamacare” only come to bear on the evaluation of deviant targets that blatantly contradict in-group norms, and of ambivalent established targets, who are expected no longer to be “on the fence” in their evaluation of the Affordable Care Act. Party identification was inversely correlated with evaluation of deviant, new member targets (r = − .37, p = .04), and marginally correlated with evaluation of deviant, full member targets (r = − .37, p = .10). This pattern corroborates previous findings of high in-group identifiers more harshly derogating in-group members (Bègue, 2001).

Central analyses

The median response to the item assessing VI was 5; a median split on this variable was used to create high and low VI conditions for the following analyses. A 2 (target membership status: new, established) × 3 (target attitude: deviant, ambivalent, normative) × 2 (vested interest: low, high) between-subjects Analysis of Covariance (ANCOVA), with participants’ attitude toward Obamacare as a covariate, and evaluation of the target as the dependent measure, revealed a significant main effect of target attitude, F (2, 117) = 6.82, p = .002, and a marginally significant interaction of target attitude with target status, F (2, 117) = 2.69, p = .07 (Figure 1). Analysis of the main effect of target attitude revealed that deviant targets (M = 4.55) were evaluated less favorably than either ambivalent (M = 4.98), F (1, 117) = 4.97, p = .03, or normative targets (M = 5.18), F (1, 117) = 13.06, p < .001. Ratings of normative and ambivalent targets were not significantly different.

Figure 1 Evaluation of in-group target predicted by target attitude and target membership status in high vested participants, controlling for participant attitude.

Simple effects analysis of the marginal target attitude x target status interaction indicated that established targets’ attitudes toward Obamacare had no effect on the evaluations they received; ratings of established deviant members (M = 4.68), that is, were not different from those of established ambivalent (M = 4.69), or normative (M = 5.03) members. However, target attitude had a significant effect on participants’ evaluations of new target members, F (2, 117) = 9.24, p <  .001. New members holding deviant attitudes were derogated (M = 4.41) relative to ambivalent new members (M = 5.26), F (1, 117) = 9.72, p = .002, and new members holding normative attitudes (M = 5.03), F (1, 117) = 15.40, p < .001; evaluations of ambivalent and normative new members were not significantly different.

The hypothesized second-order interaction effect was not statistically significant, F (2, 117) = .12, ns; however, because we had theoretical reasons to predict that VI would moderate the reactions of in-group members’ attitudes toward the target, we assessed the simple effects of the interaction of target attitude and status on evaluation of the target within the high and the low VI groups (Rosnow, Rosenthal, & Rubin, 2000). Decomposition of this interaction is critically relevant to considerations of the black sheep effect. The analysis revealed that within the low VI condition, neither of the simple main effects (target status and target attitude), nor their simple interaction were statistically significant.

Within the high VI condition, both the simple main effects of target status, F (1, 117) = 5.08, p = .03, and target attitude, F (2, 117) = 6.21, p = .003, were statistically significant. Highly vested participants rated new members more positively (M = 5.20) than full members (M = 4.74), F (1, 117) = 5.07, p = .03. Deviant members (M = 4.50) were rated less favorably than either ambivalent (M = 5.14), F (1, 117) = 6.60, p = .01, or normative members (M = 5.28), F (1, 117) = 9.96, p = .001. Normative and ambivalent members were rated similarly, F (1, 117) = .33, ns.

Among the highly vested participants, simple effects analysis of the marginally significant target status x attitude interaction, F (2, 117) = 2.51, p = .09, revealed that deviant new (M = 4.44) and established members (M = 4.56) were evaluated similarly. Evaluations of new normative (M = 5.59) targets were slightly more positive than those of established normative members (M = 4.97), F (1, 117) = 3.24, p = .07. As predicted, new ambivalent members (M = 5.57) were evaluated (by highly vested participants) significantly more positively than established ambivalent members (M = 4.70), F (1, 117) = 5.12, p = .03.

Discussion

This study assessed the conditions under which an in-group target expressing ambivalence on a group-relevant position would be sanctioned by members of the group. Prior research on the black sheep effect has not considered reactions to in-group members expressing ambivalent positions on issues of group relevance, nor have the moderating effects of VI in the topic at issue been studied. Participants of high or low VI on the issue of Obamacare read a vignette describing a new or established member of the Republican Party who was either in favor of, opposed to, or ambivalent about some aspects of health care reform legislation, and evaluated this individual. It was predicted that deviant targets would be derogated relative to targets expressing a normative position regardless of their membership status. This result was obtained when assessed across VI and membership status conditions, thus confirming the most basic tenet of SGD. Adding VI and membership status (new vs. established) to the predictive mix refined the model. The analysis revealed that target attitude (normative, ambivalent, or deviant) did not affect participants’ evaluations when the issue was viewed as not highly self-relevant. This expectation has neither been tested nor confirmed before this study.

Under conditions of high VI, a series of planned comparisons increased the theoretical reach of the SDG model. The decomposition of the second-order interaction revealed that (1) deviants were derogated irrespective of their status (new vs. established) in the in-group, (2) participants’ evaluations of new and established in-group members holding normative group-relevant attitudes differed only marginally, and (3) when participants were highly vested in the issue, their evaluation of an established group member who expressed an ambivalent attitude was significantly less positive than their evaluation of a new group member who expressed the identical ambivalent belief. Neither the marginally significant difference in evaluations of new vs. established targets holding normative in-group attitudes nor the non-significant difference in evaluation of deviant new and deviant established members were predicted. The former trend toward a significant difference in evaluation of new and established normative in-group members (by highly vested Republicans), may be attributable to participants’ beliefs that an incumbent in-group member who believes as the rest of the group does may be easily written off as a sheep complying with group norms.

Though the latter finding of a nonsignificant difference in evaluation of new in-group deviants and established in-group deviants runs contrary to Pinto et al.'s (2010) results, this apparent inconsistency can be explained by a difference in the studies’ operationalization of membership status. In the current experiment, new members were defined as having been registered as Republicans for up to 4 years (i.e., one election cycle), whereas established members had been party members for more than a decade, and had voted in three elections. Pinto and colleagues’ new members had been in-group for only 6 months. This places both our new and established members in their full member category, and may explain the discrepancy between our results and theirs. Our participants viewing deviant targets might have viewed them all as established members who should have known better than to express non-normative attitudes. Why then were ambivalent targets differentiated on the basis of membership status? Perhaps because ambivalence, by definition, makes in-group members difficult to categorize as normative or deviant, necessitating the use of cues such as length of time in-group to enable such categorization.

These findings imply that newer group members are not expected to be fully committed or socialized to in-group norms, and that a target described as having recently moved into in-group status may be allowed to show some ambivalence vís-a-vís group norms and practices. New ambivalent in-group members are tolerated because the normative component of their ambivalent attitude suggests at least some degree of allegiance to in-group norms and beliefs. The work of Abrams et al. (2008) is relevant to the current findings. It was suggested that new in-group members were functionally equivalent to the future leaders evaluated in the research of Abrams et al. The current results require that this statement be tempered, as new and full member deviant targets were both derogated relative to new and full ambivalent targets and new and full normative targets. The current findings suggest that ambivalent new targets, relative to established ambivalent targets, were evaluated similarly to the future leaders in Abrams et al. Being ambivalent and therefore holding both normative and deviant evaluations of a group-relevant attitude object is akin to being a deviant future leader, where the target is assumed to have the in-group's best interest at heart, but is not a full member or leader yet, and so is allowed some leeway (but not full deviance) in his or her position on group-relevant attitudes.

These results have implications for the wielding of social influence, and suggest that new members, or group members who are afforded leniency for some other reason, may be tolerated and ultimately, persuasive. The implication of this possibility is that in being tolerated, new in-group members also may inject more balanced or flexible thinking into debates over in-group beliefs. Many models of attitude change and persuasion, from information processing approaches like the elaboration likelihood model (Petty & Cacioppo, 1996) to message learning theories (Hovland, Janis, & Kelley, 1953), hold that once a persuasive message is processed systematically and open-mindedly, it may produce change in targeted or related attitudes. Associated (or indirect) attitude change on attitudes related to the attitude targeted by a persuasive message is predicted by the Context/Comparison - Leniency Contract (CCLC) model (Alvaro & Crano, 1997; Crano & Hemovich, 2011), which describes conditions under which non-normative opinions will be heard and cognized. Focused primarily on minority influence, the theory describes the persuasive rules that must hold if minority group members are to prevail, and their influence is to produce lasting change (Crano & Seyranian, 2009). It holds that deviant or anti-normative positions are more likely to be influential when presented by in-group minority members, as the larger group extends leniency to such individuals.

In the present research, this extension of leniency was suggested by in-group members’ responses toward neophyte members who expressed an ambivalent attitudinal position on an important feature of group orthodoxy (rejection of Obamacare).This leniency was not extended to deviant members, no matter their group status. This result, too, is consistent with predictions of the CCLC model, which holds that minority positions that threaten the very existence of the group will be rejected, and the minority relegated to out-group status. As new members, a semi-aberrant, ambivalent attitude did not threaten the stability of the group. However, established members were not afforded such leniency. They were expected to uphold the majority's normative position, and as established group members, their equivocation posed a threat to the integrity of the group.

The findings suggest that as long as a persuasive message espousing an ambivalent attitude clearly entails a (partly) normative evaluation of the attitude object, and this ambivalence can be attributed to some temporary property of the message source, such as recency of membership status, the ambivalent position will be heard. Other types of in-group members that may be heard should they choose to express ambivalence include those who hold fast to the most central tenets of the in-group prototype, who also may be granted leeway in expressing ambivalence on less central issues. A Democrat who is pro-environment, pro-choice, and in support of social welfare programs might be allowed to be ambivalent on the issue of gun control, for example. Finally, some may be expected by virtue of their membership in other groups to show some ambivalence or uncertainty about a subjective norm. For instance, a prominent Republican administrator whose daughter has publicly come out as being gay might be hard pressed to oppose gay marriage, and the circumstances surrounding the administrator's reluctance may be understood, if not entirely by the majority of the in-group.

While the present study provides preliminary insight into the effect of a transient source characteristic, membership status, on evaluation of attitudinally ambivalent in-group members (and the moderating effect of VI on this relationship), it gives no direct evidence that attributing ambivalence to some transient source characteristic may lead people also to process this ambivalent message, thereby offering a pathway toward introducing more balanced arguments into political, inter-group discourse. This possibility bears scrutiny because research has shown that indirect and, subsequently, delayed focal change may be a consequence of processing such communications.

Disclosure statement

No potential conflict of interest was reported by the authors.