When Attribution of Consistency Depends on Group Value: Social Valorization of Preference for Consistency in Equivalent and Asymmetric Intergroup Relations

The social valorization of Preference For Consistency (PFC) was examined via a minimal group paradigm activating either equivalent or asymmetric intergroup relations. After an aesthetic judgment task and assignment to one of two groups, participants (N = 298) completed the PFC scale according to three instructions: on their own behalf, as an ingroup member, or as an outgroup member (order counter-bal-anced). Results showed that in equivalent intergroup relations, participants attributed greater PFC to the ingroup and to the self than to the outgroup. In asymmetric intergroup relations, participants attributed greater PFC to the high-status group and to the self than to the low-status group, and this was inde-pendent of the participants’ group membership. We discuss the contribution the minimal group paradigm can make to revealing the social valorization of psychological constructs and the mechanisms underlying the social value accorded to PFC.


Introduction
The desire to be consistent in one's choices, attitudes and decision-making has fascinated social psychologists since the 1950s . The tendency for consistency was first considered as a need inherent to human cognitive functioning (Festinger, 1957;Heider, 1946;Abelson & Rosenberg, 1958). This trend was supposed to control attitude organization and change (e.g., Mc Guire, 1960;Feldman, 1966), as well as interpersonal relationship management (e.g., Zajonc, 1968). However, several researchers, approaching the issue from two different but complementary perspectives, have called into question the supposed universality and origin of the need for consistency. One perspective regards consistency as a personality trait, an intrinsic and measurable Preference for Consistency (PFC) that varies among individuals (Cialdini, Trost, & Newsom, 1995;Sénémeaud, Mange, Fointiat, & Somat, 2014; see Guadagno & Cialdini, 2010 for a review). The second approach takes a more social perspective, suggesting that the need for consistency is largely determined by societal expectations; therefore, expressing a preference for consistency should be socially approved and valued (see Testé, Jouffre, & Somat, 2010, or Sénémeaud, Mange, Gouger, Testé, & Somat, 2011. Adopting this second perspective on consistency, the present study aims to examine the social valorization of preference for consistency.

The valorization of PFC and interpersonal relations
Researchers regularly postulate that the desire to be and to appear consistent is the product of specifically Western cultural values (Allen, 1968;Choi & Choi, 2002;Cialdini, 2009;Hoshino-Browne, 2012;Petrova, Cialdini, & Sills, 2007;Suh, 2002;Triandis, 1989Triandis, , 1994. In line with this idea, a few studies have examined the social value conferred on consistency and on people who display PFC. For example, Channouf and Mangard (1997;see Jouffre, 2007, andJouffre, Py, &Somat, 2001, for similar results with children) reported that individuals who were instructed to give "a good image of themselves" expressed greater PFC than individuals who were asked to present "a poor image of themselves". In other words, when a situation encourages individuals to highlight their social value, both adults and children choose to display a strong PFC. In addition, targets who express a strong PFC are consistently evaluated as more likeable and serious than targets who express weak PFC (Channouf & Mangard, 1997), and the likelihoods of being offered a job or of achieving professional (Louche, Pansu, & Papet, 2001) or academic (Louche, Hugues, & Sarrade, 2001) success are assessed as being higher for individuals who display a strong PFC than for people who display a weak PFC.
Nevertheless, studies that provide experimental evidence for the social valorization of PFC are rare and nearly all of them were conducted in the field of interpersonal relations. The objective of the present research was to confirm the social value of PFC by examining it through another field, that of intergroup relations, and basing our study on minimal groups of equal and unequal status.

Intergroup biases as indicators of the social valorization of judgments
The intergroup context has been used to show the social valorization of different judgments, such as intrinsic motivation (Adam & Louche, 2009) or internality (Pansu, Tarquinio, & Gilibert, 2005). In this context, measures of social valorization relied on ingroup bias (Tajfel, 1978(Tajfel, , 1981, which is defined as a "tendency to favor the ingroup over the outgroup, in behaviors, attitudes, preferences or perception" (Turner, Brown, & Tajfel, 1979, p. 187). Ingroup bias has been highlighted by a variety of measures, such as rewards (Jetten, Spears, & Manstead, 1996;Turner et al., 1979), performance evaluations (Sherif, 1967) or trait rating (Doise & Sinclair, 1973). This process interacts with group status (see Bettencourt, Door, Charlton, & Hume, 2001 for review). Namely, high-status group members almost systematically favor their own group and sometimes derogate the outgroup, whereas low-status group members often exhibit outgroup favoritism, i.e., favoring the high-status group (Brown, 1995, for review;Guimond, Dif, & Aupy, 2002). On this basis, authors (Dubois & Beauvois, 1996;Dubois, Beauvois, Gilibert, & Zentner, 2000;Pansu et al., 2005) have considered that, as people tend to attribute more positive judgments to their ingroup in equivalent intergroup relations, then a judgment systematically attributed to an ingroup should be considered socially valued. Similarly, as people tend to attribute more positive judgments to a high-status group in asymmetric intergroup relations, then a judgment systematically attributed to the high-status group should be considered socially valued. For example, as it is socially valued, internality is systematically attributed to the ingroup in studies activating equivalent intergroup relations (Dubois et al., 2000) and to the high statusgroup in studies based on asymmetric intergroup relations (Dubois & Beauvois, 1996;Beauvois, Gilibert, Pansu, & Abdelaoui, 1998;Pansu et al., 2005).

Intergroup biases as indicators of the social valorization of PFC
Only one study examined the social valorization of PFC in an intergroup context (Sénémeaud et al., 2011). The results showed that people tend to allocate greater PFC to members of the most socially valued group (high-status group). However, one major concern with this study, implemented in a natural setting with real-life groups, is that the results do not allow us to determine whether PFC has been allocated to groups according to their social status or according to their intrinsic characteristics or their social history. Therefore, we decided to use a minimal group paradigm (Tajfel, Billig, Bundy, & Flament, 1971) with equivalent and asymmetric group status to clearly determine the social valorization of PFC in the field of intergroup relations.
First, the minimal group paradigm, which consists in creating two artificial groups in a purely experimental context, neutralizes the intrinsic characteristics and/or social history of the groups (Dubois, 2009). Second and above all, the experimental manipulations of equivalent and asymmetric intergroup relations in this paradigm (e.g., Lorenzi-Cioldi, F. (2008) Sachdev & Bourhis, 1987, 1991 allows to draw conclusions about the value of psychological constructs with more certainty. If a construct is socially valorized, then intergroup attributions of this construct should be a function of the social value of the group mentioned. Consequently, all modifications in the social value attributed to the group should lead to similar modifications in the construct attribution to groups. In our specific case, if expressing PFC is considered socially approved and valued, PFC will then be allocated to minimal groups based on the value conferred to these groups by the participants. Hence, in equivalent intergroup relations, as people tend to attribute more value to the ingroup, we predict higher PFC will be attributed to the ingroup compared to the outgroup. In asymmetric intergroup relations, as people tend to attribute more value to a high-status group, higher PFC attributions to the high-status group compared to the low-status group should be observed, whatever the participants' initial group membership (low-vs. high-status group).

The present research
To test our predictions, a study was conducted in a minimal group paradigm (Tajfel et al., 1971) and was divided into two sessions. During the first session, the participants carried out an aesthetic judgment task. In the second session, the participants were randomly assigned to one of two groups. These minimal groups were created by means of bogus feedback concerning the participants' aesthetic taste evaluated during the aesthetic task. In the equivalent intergroup relation condition, no other information on groups activated was delivered. In the asymmetric intergroup relation condition, the feedback included first a description of the groups' cognitive ability in aesthetic judgments (strong cognitive ability for high-status group vs. poor cognitive ability for low-status group, cf. Sachdev, & Bourhis, 1987, 1991Scheepers, Branscombe, Spears, & Doosje, 2002). Afterwards, the participants filled out the PFC scale (Cialdini et al., 1995) according to three instructions (identification paradigm; see for review Gilibert & Cambon, 2003): on their own behalf vs. as an ingroup member vs. as an outgroup member (order counter-balanced for the two last instructions).
First, we expected to observe higher PFC attributions to the ingroup than to the outgroup in equivalent intergroup relations (hypothesis 1a) and higher PFC attributions to the high status group than to the low status group in asymmetric intergroup relations (hypothesis 2a). Second, as research in the field of interpersonal relations has demonstrated that PFC is used by people to value themselves (Channouf & Mangard, 1997, study 2;, then the score of PFC conferred to the self should always be similar to the one conferred to the most valued group. In other words, self PFC scores should be identical to ingroup PFC scores in equivalent intergroup relations (hypothesis 1b) and self PFC scores should be identical to high-status group PFC scores in asymmetric intergroup relations (hypothesis 2b).

Method Participants and Design
The participants were 298 French undergraduate students (108 males, 190 females; Mean age = 19.49, SD age = 1.51). The design was mixed with type of intergroup relation (equivalent vs. asymmetric) and participant's group membership (the Margueray group -people having supposedly preferred a painting by Margueray vs. the Rimbert group -people having supposedly preferred a painting by Rimbert) as between-subjects factors and instructions (self-description vs. ingroup identification vs. outgroup identification) as a repeated measure.

Procedure and Measures
The study was introduced as a two-part survey on "aesthetic judgments in an academic setting" in classes of 19 to 33 students.
In the first phase, participants performed the aesthetic judgment task (cf. Dubois et al., 2000). One of the two experimenters explained to students that they would see a set of paintings by contemporary artists presented in pairs. The participants were asked to choose the painting they preferred from each pair and to indicate their choices on an answer sheet. The experimenter collected the answer sheets and then left the classroom to analyze the participants' answers and determine their profiles. Students were randomly categorized into one of two groups, the Margueray group or the Rimbert group. The experimenter came back at the end of the class session to give the bogus feedback on the students' aesthetic profiles. In the equivalent intergroup relation condition, he told the students "The pictures you evaluated were by two painters, in chronological order, Margueray and Rimbert. Interestingly, some people appear to have systematically preferred Margueray's paintings, while others systematically chose Rimbert's paintings. In other words, there were two groups in the classroom: the Margueray group and the Rimbert group".
In the asymmetric intergroup relation condition, he added: "Margueray is well-known for his artistic qualities and pictorial technique, whereas Rimbert is known for his lack of aesthetic qualities and poor pictorial tech-nique… Some people systematically preferred Margueray's paintings, thereby indicating that they have very good cognitive abilities when it comes to making aesthetic judgments. Other people systematically chose Rimbert's paintings, thereby showing they have poor cognitive abilities with respect to making aesthetic judgments." Then the experimenter told the students in both conditions that they would be confidentially notified of their group membership on an individual sheet of paper. Hence, in the "asymmetry" condition, participants who were supposed to have preferred Margueray were considered the highstatus group, whereas participants who were supposed to have preferred Rimbert were considered the low-status group. A pilot study was conducted to check the efficiency of the manipulation in terms of both group identification and status perception (equivalent vs. asymmetric; see Appendix for details).
The second phase was presented as being designed to "examine the link between aesthetic preferences of people and their general attitudes and behaviors." The first page of a booklet distributed to each of the participants described the participant's bogus membership as follows: "Result analysis demonstrated that some people systematically preferred Margueray's paintings, whereas others systematically chose Rimbert's paintings. Analysis of your choices showed that you belong to the group that preferred Margueray (vs. Rimbert)." Participants were then required to write the name of their group on the following page, in order to prime their membership (Doise & Sinclair, 1973). The next three pages contained a PFC scale with specific instructions for completing it: according to one's own opinion vs. how a member of one's group would complete it vs. how a member of the other group would complete it (order was counter-balanced for the last two instructions). Finally, the participants were thanked and debriefed.
Dependent measure. The dependent measure was the level of PFC measured using Channouf and Mangard's (1997) French version of the PFC scale (Cialdini et al., 1995). The French PFC scale consists of fifteen items that investigate the desire to be consistent, to appear consistent to others and for others to be consistent, as well as the consistency between two attitudes or two behaviors, or between attitude and behavior. All items were assessed on a 9-point scale (from 1 = strongly disagree to 9 = strongly agree). The responses were averaged to give a single PFC value (alpha = .81). The higher the score on the scale, the stronger the participant's PFC.

Preliminary analysis
We began by performing ANOVAs to test the possible effects of "instruction order" and sex. We did not find a significant effect on PFC scores of either "instruction order" or sex, F (1, 290) < 1; ns and F (1, 290) = 1.24, p = .27, respectively. Neither did we find any effect of age 1 . Consequently, we excluded these three variables from subsequent analyses and they are not discussed further.

Main analysis
A 2 (intergroup relation type: equivalent vs. asymmetric) x 2 (group membership: Margueray vs. Rimbert) x 3 (instruction: self-description vs. ingroup identification vs. outgroup identification) ANOVA, with repeated measures on the third factor, did not reveal a main effect of either intergroup relation or of participant's group membership, F(1, 294) < 1; ns. However, it did reveal a significant main effect of instruction, F(2, 588) = 35.71, p < .001, h 2 = 0.11, as the PFC score obtained in the ingroup identification condition (M = 6.22, SD = 1.2) was significantly higher than the score obtained in the outgroup identification condition (M = 5.38, SD = 1.51; LSD test, p < .001) but not different from the score obtained in the self-description condition (M = 6.21, SD = 1.01, p = .93). The self-description condition also significantly differed from the outgroup identification condition (p < .001). This instruction main effect was qualified by a participant group membership x instruction interaction, F(2, 588) = 15.41, p < .001, h 2 = 0.05, itself moderated by the predicted threeway interaction of intergroup relation, participant group membership and instruction, F(2, 588) = 16.35, p < .001, h 2 = 0.06. The mean results for each experimental condition are presented in Table 1.
We tested our specific hypotheses by conducting two planned orthogonal contrasts for each "intergroup relation x participant group membership" condition. We began by testing the same two contrasts for the Margueray group and the Rimbert group in the equivalent intergroup relation condition. First, we compared the ingroup and self PFC scores with the outgroup PFC scores (ingroup = 1, self = 1, outgroup = -2). Second, we compared the ingroup PFC scores with the self PFC scores (ingroup = 1, self = -1, outgroup = 0). The first contrast was significant for both groups (Margueray: t = 4.91, p < .001; Rimbert: t = 5.03, p < .001), indicating that stronger PFC was attributed to the ingroup (M Margueray = 6.19, M Rimbert = 6.3) and to the self (M Margeray = 6.2, M Rimbert = 6.23) than to the outgroup (M Margueray = 5.27, M Rimbert = 5.34). The second contrast was not significant: there was no difference between the PFC scores attributed to the self and to the ingroup (t = -0.67, p = .49 for Margueray and t = 0.1, p = .91 for Rimbert). Hence, in accordance with hypotheses 1a and 1b, PFC was used to express value, whether for the self or the ingroup.
Second, we conducted the same two orthogonal contrasts in the asymmetric intergroup relation condition and for the Margueray high-status group, (1, 1, -2, and 1, -1, 0, respectively, for the ingroup, self, and outgroup conditions). The first contrast was significant (t = 6.09, p < .001), demonstrating that greater PFC was attributed to the ingroup (M = 6.61) and to the self (M = 6.3) than to the outgroup (M = 4.89). Interestingly, the second planned comparison was also significant (t = -2.11, p < .04): ingroup PFC scores were higher than self PFC scores, suggesting that PFC score increase when the value of the group is strengthened.
Third, we calculated two new contrasts in the asymmetric intergroup relation condition and for the Rimbert low-status group. The first contrast compared the ingroup PFC scores with the outgroup and self PFC scores; the second contrast compared the outgroup PFC scores with the self PFC scores (-2, 1, 1, and 0, -1, 1, respectively, for the ingroup, self, and outgroup conditions). The first contrast was significant (t = 2.55, p < .02), revealing that, as expected, the PFC scores for the high-status outgroup and for the self (M = 6.08) were significantly higher than the PFC scores for the low-status ingroup (M = 6.23), and that there was no significant difference between the outgroup PFC scores and the self PFC scores (t = 0.55, p = .58). Hence, as postulated in hypotheses 2a and 2b, PFC was used to express the value of the high-status group, whatever the participants' initial group membership (low-vs. high-status group).

Discussion
The aim of the present research was to further examine the social valorization of preference for consistency by using a minimal group paradigm with groups of equivalent and differential status. In equivalent intergroup relations, results showed that greater PFC was attributed to the ingroup and to the self compared with the outgroup, with no difference between the PFC attributed to the ingroup and to the self. In asymmetric intergroup relations, participants attributed greater PFC to the high-status group and to the self than to the low-status group, and this was the case for participants in both the high-status and the low-status groups. Moreover, low-status group members attributed equivalent levels of PFC to the self and to high-status group members, whereas high-status group members attributed greater PFC to the ingroup than to the self. The following discussion is based on two points: the interest of using the minimal group paradigm and manipulating group status to explore the social valorization of psychological constructs, and the contribution of our results to furthering understanding of the mechanisms underlying the social valorization of PFC.
Our study showed that the minimal group paradigm can be used with groups of equal and unequal status to explore the social value of psychological constructs (see also Adam & Louche, 2009, or Pansu et al., 2005. As noted in the introduction to this paper, manipulating group status within the minimal group paradigm can generate accurate predictions against which the social valorization of constructs can be tested. On the one hand, demonstrating that modifications in the intergroup relation, and consequently in the social value attributed to the groups,  leads to similar modifications in the construct attribution between groups, is a way of determining the social valorization of a construct. In this regard, it is precisely because the attribution of PFC varies according to the type of intergroup relation and to the value people wish to confer to the target group that we can conclude that PFC is socially approved and valorized. On the other hand, the manipulation of group status in this paradigm allows us to preclude alternative explanations of intergroup bias that are not based on the social value of PFC. More precisely, in the equivalent intergroup relation condition, the PFC scores attributed to the self and to the ingroup were similar. We believe this result is indicative of a self-presentation strategy adopted by the participants in order to obtain social approval in the experimental situation (e.g., Dubois, 1991;Dubois et al. 2000) because PFC provides a way of expressing value, whether for the self or for the ingroup. However, an alternative interpretation to the valorization of PFC is that participants' responses to the instructions were intended to accentuate similarity between themselves and their (bogus) ingroup. If this were the case, participants would respond to the ingroup identification instructions by basing their attributions on their own personality traits or their own needs (for consistency), without necessarily trying to value their ingroup.

Equivalent intergroup relation Asymmetric intergroup relation
Results obtained in asymmetric intergroup relations rule out this alternative because they show that PFC selfattribution scores were equivalent to those of the most socially valued group and not systematically to those of the ingroup. This was the case for low-status group members, who attributed the same PFC scores to themselves and to the outgroup. These findings contradict a simple similarity accentuation principle but concord with attributing PFC as a way of according value.
Finally, this line of reasoning may apply to many explanations of minimal ingroup bias (see Spears & Otten, 2012 for a review), including the classic need to maintain or achieve a positive social identity (Tajfel & Turner, 1986), explanations based on expectations of reciprocity within the ingroup (e.g., Gaertner & Insko, 2000), and explanations that stress the role of the self in PMG effects (e.g. Cadinu & Rothbart, 1996). In this final case, the minimal ingroup bias could arise from the projection of self-positive aspects onto the ingroup (Clement & Krueger, 2002;Otten, & Wentura, 2001) or onto the most valued group (Otten, 2005). The fact that our results show that PFC is projected onto the ingroup and the high-status group indicates that PFC corresponds to a positive and valued attribute. In short, whatever the mechanism driving the minimal ingroup bias and its moderation by group status, the mere presence of this phenomenon through the attribution of PFC is enough to confirm the social value of PFC. Hence, although it has been under used in social valorization research, the minimal groups paradigm with manipulation of group status is a valuable tool for exploring or confirming the social value of constructs.
A second important aspect of our findings is that they allow us to further examine possible reasons why endorsement of a strong PFC should be valued in Western society. The most common explanation refers to cross-cultural point of view and is based on variations in self-system between Western and East Asian cultures (e.g., Hoshino-Browne, 2012;Markus & Kitayama, 1991, 2010Markus, Mullally, & Kitayama, 1997) 2 . Another complementary explanation is based on the assumption that high PFC is socially valued because it is useful for the social system (Beauvois, 1991).
More precisely, the prevailing model in the literature on social judgment suggests the existence of two fundamental dimensions underlying individual and group judgments. These "Big Two" (Abele & Wojciszke, 2013 dimensions are referred to under a variety of names, including warmth and competence (Fiske, Cuddy, Glick, & Xu, 2002), communality and agency (Abele & Wojciszke, 2013), and social desirability and social utility (Beauvois & Dubois, 2001). The first dimension refers to a person's social and moral qualities, capacity to be appreciated by others and ability to function in social relations. The second dimension relies on a person's ability to achieve objectives, succeed in a social system and obtain the most valued positions (Abele & Wojciszke, 2007;Darnon, et al. 2009;Dubois & Beauvois, 2001, 2005Testé & Perrin, 2013). Research suggests that this second dimension is associated with high-status groups (Fiske et al., 2002;Fragale, Rosen, Xu, & Merideth, 2009;Oldmeadow & Fiske, 2010), and related to an individual's status (Carrier, Louvet, Chauvin, & Rohmer, 2014;Dubois, 2010).
The link between high-PFC and high-status revealed by the present study gives food for thought about the potential social utility of PFC. Indeed, high-PFC people, who should be considered predictable and stable in their way of thinking and behaving (Channouf & Mangard, 1997), should also be perceived as more able to succeed and bring progress to the social system than moderate-or low-PFC people. However, although some studies have reported that high-PFC targets were evaluated as more socially useful than low-PFC targets (Channouf & Mangard, 1997;Sénémeaud et al., 2011), other studies have not found this difference (Beauvois, 2003;Testé et al., 2010). Even though the primary aim of the present research was not to settle this issue, our study revealed new data favoring the social utility, rather than the social desirability, of PFC. Viewed from this perspective, the social function of consistency may be comparable to that of other constructs that are socially valued in individualistic societies and anchored in social utility dimension, such as an internal locus of control (Dubois, 2003;Jellison & Green, 1981;Perrin & Testé, 2010), self-sufficiency (Beauvois & Dubois, 2005), and belief in a just world (Alves & Correia, 2008, Testé & Perrin, 2013. Future studies are needed to clarify the social value associated with PFC.

Conclusion
The present study confirms the social valorization of PFC by demonstrating that people use PFC to display value to groups in conformity to the social status of the group. This result questions psychological and social consequences, especially for members of low-status groups. More precisely, studies in the field of interpersonal relations examining the social valorization of PFC have shown that, in addition to being less positively valued, low-PFC people might be less likely to achieve academic and professional success than high-PFC people (Channouf & Mangard, 1997;. In our study, the low-status group was systematically attributed less PFC, even by low-status group members, suggesting that members of low-status groups contribute to perceptions of their reduced success and weaker opportunities for social progress by allocating lower PFC to their own group. By doing so, they may help legitimize and maintain existing social inequalities, consistent with the idea shared by some contemporary theories in the intergroup field (e.g., Eagly, 1987;Glick & Fiske, 2001;Jost & Banaji, 1994;Oldmeadow & Fiske, 2010). From this last perspective, PFC attribution would not be an innocuous social "tool" as it participates to legitimize status systems.

Notes
1 Regression analysis indicated no significant effects of age on PFC scores, whether in the self-description instruction, ß = .10, t (296) = 1.7, p = .09, the ingroup instruction, ß = .03, t (296) = 1, ns, or the outgroup instruction: ß = .03, t (296) = 1, ns. 2 More precisely, ideal self-concepts in Western societies, which include independence, unicity, rationality, and stability, could be the source of the social approbation conferred to consistency by individuals. In contrast, PFC should be valued less highly among East Asian cultures, which favor an interdependent, rather than independent, self-system. Hence, they attach greater importance to maintaining harmonious interpersonal relationships with significant others and to basing decisions on other people, rather than on their own past choices, behavior or beliefs.