Gabriel Mugny and his colleagues have intensively investigated factors moderating the influence of egalitarian and discriminatory norms on intergroup attitudes and behavior (e.g., Pérez & Mugny, 1993). Research on social influence showed that conformity to social norms depends on the extent to which norms are focal for the individual (see Cialdini, Kallgren & Reno, 1991). However, one of the key tenets of Mugny’s research on social influence is that, even when norms are salient, individuals do not blindly comply to group norms. This research initially analyzed conformity to ingroup norms as a function of the individuals’ perceived discrepancy between their own position and the norm (Muñoz-Rojas, Falomir-Pichastor, Invernizzi Gamba & Leuenberger, 2000; Sanchez-Mazas, Mugny & Jovanovic, 1996; Pérez, Sanchez-Mazas & Mugny, 1993). Moreover, this research has also investigated conformity to ingroup norms as a function of individuals’ perception of the norm as legitimate or not (i.e., as adapted or not to individual- or group-related values and motives). Indeed, social norms are perceived as legitimate to the extent they are congruent with individuals’ values (Zelditch, 2001), and that individuals can show non-conformity and even counter-conform when the norm violates important personal values (see Hornsey, Majkut, Terry & McKimmie, 2003).

The main goal of the present research is twofold. First, we aim at providing consistent evidence in support of Mugny and colleagues’ findings regarding conformity to ingroup norms when motivation to conform to group norms and motivation to defend ingroup identity are in conflict. Secondly, we also aim at increasing our understanding of these conformity processes by examining the relationship between ingroup identification and two indicators of intergroup differentiation—stereotyping and prejudice. To do so, we investigated whether ingroup identification predicted intergroup differentiation as a function of group norms (egalitarian versus discriminatory) and perceived threat to ingroup distinctiveness (low versus high).

Group identification and conflicting motives

Past research shows that both defensive reactions to ingroup threat and conformity to ingroup norms are intrinsically related to group dynamics (Tajfel & Turner, 1986), and that they are positively related to ingroup identification, which is defined as the strength of an individual’s ties with their ingroup identity (Ellemers, Spears & Doosje, 2002). Indeed, high group identification increases both intergroup differentiation (e.g., Branscombe, Schmitt & Harvey, 1999; Branscombe & Wann, 1994; Ellemers, et al., 2002; Esses et al., 1998; Levine & Campbell, 1972; Tajfel & Turner, 1986; Voci, 2006; see Brown, 2000, for a discussion) and conformity to ingroup norms (e.g., Christensen, Rothgerber, Wood & Matz, 2004; Jetten et al., 2002; Tajfel & Turner, 1986), even more so when ingroup threat is high. However, it is worth noting that these two processes may also be in conflict; for instance, when the group norm motivates group members to reduce intergroup differentiation (e.g., egalitarian ingroup norms) but perceived ingroup threat motivates group members to increase intergroup differentiation. Therefore, how do group members, and high identifiers in particular, negotiate the conflict between these two opposing motives?

Falomir-Pichastor, Mugny, and Gabarrot proposed the notion of loyalty conflict to account for this conflicting state (Falomir-Pichastor, Gabarrot & Mugny, 2009a, 2009b; see also Falomir-Pichastor & Mugny, 2013), and suggested that high identifiers solve this conflict by opposing the egalitarian norm (i.e., they show non-conformity or even counter-conformity) paradoxically as a way to show their loyalty to the group (see also Jetten & Hornsey, 2013; Packer, 2008). In a series of experimental studies. They examined the influence of egalitarian and discriminatory norms on discrimination against immigrants as a function of perceived threat (Falomir-Pichastor, Muñoz-Rojas, Invernizzi & Mugny, 2004; Falomir-Pichastor, et al., 2009a, 2009b; Falomir-Pichastor, Chatard, Selimbegovic, Konan & Mugny, 2013; Gabarrot, Falomir-Pichastor & Mugny, 2009; for a review, see Falomir-Pichastor & Mugny, 2013). For instance, egalitarian norms reduced discrimination against immigrants when immigrants were not perceived as threatening, but not when they were perceived as threatening ingroup prerogatives (i.e., when immigrants were described as increasing Swiss unemployment; Falomir-Pichastor et al., 2004). Further research showed that this selective conformity also appeared when the outgroup threatened ingroup identity (Falomir-Pichastor et al 2009b; Gabarrot et al., 2009) and when ingroup identification is high (Falomir-Pichastor et al., 2009a; Gabarrot et al., 2009). Finally, these findings also showed high identifiers’ loyalty towards their group in threatening contexts: while they actively opposed (rather than merely disregarded) ingroup egalitarian norm, high identifiers compensated by concurrently showing greater attachment to their group (Falomir-Pichastor et al., 2009a).

In the present research, we aim to increase our understanding in how high identifiers deal with loyalty conflict by investigating conformity as a function of ingroup distinctiveness threat and the (descriptive versus evaluative) nature of the differentiation process.

Distinctiveness threat and descriptive versus evaluative differentiation

Social identity theory (Tajfel & Turner, 1986) contends that group members, and, in particular, high identifiers, are motivated to maintain a positive and distinctive social identity. Intergroup distinctiveness is defined as the perceived difference between the ingroup and the outgroup on such a dimension (Jetten, Spears & Postmes, 2004). However, the effect of perceived ingroup distinctiveness (or perceived similarity) between the ingroup and a relevant outgroup on intergroup relations is complex. Past research show that high distinctiveness (or low intergroup similarity) is associated with intergroup differentiation (i.e., stereotyping, prejudice and discrimination), while low distinctiveness (or high intergroup similarity) is expected to either increase or decrease differentiation (see Jetten et al., 2004; see also Costa-Lopes, Vala & Judd, 2012, for a recent review). As such, intergroup differentiation may be a reflection of intergroup distinctiveness, or a reaction against the threat associated to a lack of distinctiveness (Jetten & Spears, 2003; Jetten et al., 2004). Indeed, the perception of high intergroup similarity makes it harder to achieve a positively distinct social identity (see Branscombe, Ellemers, Spears & Doosje, 1999), and group members would be motivated to restore ingroup distinctiveness by strengthening intergroup differentiation (e.g., Diehl, 1988; Moghaddam & Stringer, 1988; Roccas & Schwartz, 1993; Spears, Doosje & Ellemers, 1997).

Whereas research has often investigated intergroup similarity in terms of shared attitudes, status, values or characteristics, similarity can also be made salient by social policies or norms, contributing in turn to the de-emphasizing of intergroup differences in a specific social context (Falomir-Pichastor & Hegarty, 2014; Falomir-Pichastor, Mugny & Berent, 2017; Gabarrot et al., 2009). For instance, Gabarrot et al. (2009) examined the influence of egalitarian and discriminatory ingroup norms on prejudice and discrimination (e.g., ingroup bias) as a function of distinctiveness threat (i.e., perceived intergroup similarity; Studies 1 and 2) and ingroup identification (Study 2). Ingroup identification increased prejudice and ingroup bias both when perceived similarity was low and the group norm was discriminatory, and when perceived similarity was high and group norm was egalitarian. This last finding is consistent with Falomir-Pichastor et al.’s (2009a) loyalty conflict, and suggests that group identification increases both motivation to group conformity and to restore group distinctiveness, but that group members resolve this conflict showing counter-normative behaviour. However, this last study does not provide evidence in support of the aforementioned compensatory mechanism.

The present research intends to show the existence of such a compensatory mechanism in loyalty conflict situations by distinguishing among different dimensions of intergroup differentiation. In their meta-analysis, Jetten et al. (2004) distinguished between judgmental and behavioral measures of differentiation. Typically, judgmental measures are trait attributions, whereas, within the social identity theory framework, behavioral measures correspond to points or reward allocations. These authors proposed that intergroup distinctiveness would be more easily reflected on judgmental measures, whereas intergroup similarity would enhance differentiation on behavioral measures. However, previous research showed very similar patterns of results on both behavioral and judgmental measures (respectively ingroup bias and modern prejudice scale; see Gabarrot et al., 2009). Therefore, further research is needed for a better understanding of the different ways in which distinctiveness may relate to intergroup differentiation.

To address this issue, in the present research we suggest a distinction between judgmental measures related to ratings on stereotypical traits (e.g., competent, warm) and judgmental measures involving evaluative traits (e.g., nice, aggressive). More specifically, we propose that the cognitive dimension of intergroup differentiation would be more effectively captured by ratings on stereotypical traits—i.e., more stereotypical than counter-stereotypical trait attributions for both groups—whereas the motivational dimension would be more promptly captured by differential attribution of positive (vs. negative) traits to the ingroup and to the outgroup. When intergroup similarity is low and the group norm is discriminatory, we expect group identification to increase both stereotype and prejudice (i.e., reflective distinctiveness; Gabarrot et al., 2009). More importantly, this distinction may be crucial in understanding the way high identifiers deal with loyalty conflict. Indeed, when conformity and ingroup defense motives are in opposition (i.e., when ingroup distinctiveness is threatened and the group norm supports intergroup equality), we postulate that group identification should increase prejudice allowing the required defense of the group against distinctiveness threat (i.e., reactive distinctiveness; Gabarrot et al., 2009). However, and according to the hypothesized compensatory mechanism (Falomir-Pichastor et al., 2009a), group identification should concurrently increase conformity to the egalitarian norm through decreased stereotyping.

The present research

The present research aimed at extending our understanding of conformity processes in threatening intergroup contexts by examining the effect of distinctiveness threat on descriptive versus evaluative differentiation measures, and by focusing specifically on the role of ingroup identification. To do this, we assessed participants’ ingroup identification, manipulated both intergroup similarity (low vs. high) and ingroup norms (discriminatory vs. anti-discrimination), and finally measured descriptive (stereotype) versus evaluative (prejudice) forms of intergroup differentiation. We predicted an identification × similarity × norm interaction effect on both prejudice and stereotyping. We expect ingroup identification to be positively related with prejudice when both intergroup similarity is low and group norm is discriminatory—which would correspond to the reflective distinctiveness process (H1a)—and when intergroup similarity is high and group norm is egalitarian (H1b)—which would correspond to the reactive distinctiveness process and the loyalty conflict it elicits. More importantly, we expect group identification to be negatively related with stereotyping, when intergroup similarity is high and the norm is egalitarian, due to the hypothesized compensatory mechanism (H2).

Method

Participants and procedure

Eighty-two students were recruited on the campus of a French University (67 women, mean age = 22.18, SD = 7.49). All of them were French native speakers, born in France, with French nationality. A male experimenter invited participants to complete a questionnaire concerning the image of different social groups living in France. The questionnaire was presented as assessing the social images of French nationals and North-African immigrants living in France. It was specified that the term ‘North-African immigrants’ referred to people from Algeria, Tunisia, or Morocco who live in France without having French nationality. Participants were randomly assigned to one out of four conditions in a 2 (Intergroup similarity: High vs. Low) × 2 (Ingroup norm: Egalitarian vs. Discriminatory) design. Each questionnaire contained a measure of ingroup identification, the manipulation of ingroup norm and intergroup similarity, and a representativeness estimate task (see below; see also Wolsko, Park, Judd & Wittenbrink, 2000). The questionnaire also contained manipulation-checking questions. The main dependent variables, computed on the basis of participants’ responses on the representativeness estimate task, were prejudice and stereotyping.

Materials

Ingroup identification. Ingroup identification was assessed using a five-item scale (see Doosje, Spears & Ellemers, 2002). Participants were asked to indicate on a seven-point scale (from 1 = ‘Not at all’ to 7 = ‘Absolutely’) the extent to which they agreed with five statements (‘At this moment, I identify with my group, the French, ‘At this moment, I see myself as belonging to my group, the French’, ‘At this moment, I am happy to belong to my group, the French’, ‘At this moment, I feel committed to my group, the French’, and ‘At this moment, I feel solidarity with my group, the French’). An ingroup identification score was computed by averaging participants’ responses to each item (M = 4.57, SD = 1.54, Cronbach’s α = 0.94).

Ingroup norm. Manipulation of ingroup norm was identical to Gabarrot et al. (2009). Participants were informed about the results of an alleged previous study—similar to the one they were participating in—carried out with a representative sample of French nationals. Results were displayed in graphical form, using percentages of responses (i.e., ‘Yes’, ‘No’, and ‘I don’t know’) to four questions. In the egalitarian norm condition, participants were informed that most of the French nationals allegedly polled in this study did not consider favoring French people against North Africans in terms of social welfare (unemployment benefits, disability or health insurance), nor in terms of housing or education benefits, to be legitimate (respectively, 82.25 and 79.21% of ‘No’). Furthermore, participants were informed that this sample did not actually favor their ingroup relatively to the outgroup when asked to allocate these resources (respectively, 89.26, and 87.33% of the participants opted for an egalitarian distribution of resources). In the discriminatory norm condition, high percentages were associated with discriminatory responses (i.e., 82.25 and 79.21% of participants of the alleged study considered ingroup favoritism on social resources to be legitimate, and 89.26 and 87.33% of these participants allegedly opted for an ingroup favoring distribution of resources).

Intergroup similarity. Induction of similarity was identical to the one used in Gabarrot et al. (2009, Study 2). In the high similarity condition, both groups were allegedly similar on four traits (sociable, organized, cooperative, and self-reliant): two of the traits were highly present in both the French and the North-African samples, whereas the two remaining traits were barely present in the two samples. For instance, a majority of French nationals were allegedly reported to have described themselves as sociable (84.2%) and organized (75.3%), while only a minority described themselves as cooperative (15.1%) or self-reliant (13.1%). Similarly, a majority of North-African people were allegedly reported to have described themselves as sociable (85.6%) and organized (73.6%), while only a minority described themselves as cooperative (14.3%) and self-reliant (16.3%). In the low similarity condition, both groups were presented as being similar on two traits and different on the other two traits. For instance, a majority of French nationals was allegedly reported to have described themselves as sociable (84.2%) and organized (75.3%), while only a minority described themselves as cooperative (15.1%) or self-reliant (13.1%). In contrast, a majority of North-African people were allegedly reported to have described themselves as sociable (85.6%) and cooperative (73.6%), while a minority described themselves as organized (14.3%) and self-reliant (16.3%). All of these traits were pretested to have a positive valence and were counterbalanced so that differences between the two experimental conditions would not be attributable to the description of the groups.

Dependent measures. Participants were presented with a list of 16 traits—all different from those used in the similarity induction—and were asked to indicate whether they thought these traits were representative of French nationals in general, and of North African immigrants in general, on a 7-point scale (from 1 = ‘not representative’ to 7 = ‘very representative’). The traits used in this experiment were borrowed from Dambrun and Guimond (2004) and were selected based on pilot studies not presented here (see Gabarrot et al., 2016). Half the traits were stereotypic of North Africans (and counter-stereotypic of the French; positive traits: warm, straightforward, original, stick together; negative traits: aggressive, insolent, threatening, violent) and half were stereotypic of the French (and counter-stereotypic of North Africans; positive traits: ambitious, cultured, polite, hard-working; negative traits: egoistic, grumpy, pessimistic, withdrawn). We computed two indexes corresponding to the main dependent variables: prejudice and stereotyping (see Table 1).

Table 1

Contrast weights used for the computation of stereotyping and prejudice indexes.

Target group French North Africans

Trait stereotypicality French North Africans French North Africans

Contrast 1 – Prejudice
Trait valence: Positive +1 +1 –1 –1
Trait valence: Negative –1 –1 +1 +1
Contrast 2 – Stereotyping
Trait valence: Positive +1 –1 –1 +1
Trait valence: Negative +1 –1 –1 +1

The first index (‘prejudice’) represents differentiation on the evaluative dimension regardless of trait stereotypicality, and thus corresponds to a target group × trait valence interaction. A positive score on this contrast indicates that positive traits are more readily associated with the French than negative traits and that negative traits are more readily associated with the North-Africans than positive traits. The second index (‘stereotyping’) represents intergroup differentiation regardless of trait valence, and thus corresponds to a target group × trait stereotypicality interaction. A positive score on this contrast indicates that North Africans-stereotypic traits are more readily associated with North Africans than French-stereotypic traits, and conversely that French-stereotypic traits are more readily associated with the French than North Africans-stereotypic traits.1

Manipulation checks. One item assessed participants’ perceived intergroup similarity both just after the similarity manipulation and at the end of the questionnaire (‘Personally, you think that the French and North Africans are:’ 1 = ‘Absolutely different’ and 7 = ‘Absolutely similar’). Responses to these two items were correlated, r = .48, p < 0.001, and the two scores were combined to obtain a single score of perceived similarity. At the end of the study, two items assessed the perceived ingroup norm: ‘To what extent do people allocate more social resources to French people than to North African people?’ (1 = ‘As much to the French as to North Africans’ to 7 = ‘Much more to the French than to North Africans’) and ‘To what extent do people think it is legitimate to favor French people over North African people in terms of social resources?’ (1 = ‘Not legitimate at all’ to 7 = ‘Very legitimate’). The score between the two measures were averaged (r = .68, p < 0.001).

Results

Manipulation checks

A 2 (intergroup similarity: High vs. Low) × 2 (ingroup norm: Egalitarian vs. Discriminatory) ANOVA was performed on manipulation check scores. Regarding perceived similarity, the main effect of the similarity induction was significant, F(1, 77) = 43.86, p < 0.001, ηp2 = 0.363. French nationals and North Africans were perceived as more similar in the high similarity condition (M = 5.34, SD = 1.13) than in the low similarity condition (M = 3.93, SD = 0.74). Neither the main effect of ingroup norm, F(1, 77) = 0.353, p = 0.554, ηp2 < 0.01, nor the similarity by norm interaction effect, F(1, 77) = 0.129, p = 0.72, ηp2 < 0.01, reached statistical significance.

Regarding the perceived ingroup norm, the main effect of the norm manipulation was significant, F(1, 77) = 73.10, p < 0.001, ηp2 = 0.49. Ingroup norm was perceived as more discriminatory in the discriminatory norm condition (M = 5.70, SD = 0.96) than in the egalitarian norm condition (M = 3.51, SD = 1.35). The results also revealed a significant main effect of the similarity manipulation, F(1, 77) = 6.16, p = 0.015, ηp2 = 0.07. Discrimination was perceived as less normative in the high similarity condition (M = 4.16, SD = 1.72) than in the low similarity condition (M = 4.85, SD = 1.46). The similarity by norm interaction effect, F(1, 77) = 0.778, p = 0.38, ηp2 = .01 did not reach statistical significance.

Dependent variables

Prejudice and stereotyping were regressed on intergroup similarity induction (coded: –0.5 = low distinctiveness; +0.5 = high distinctiveness), ingroup norm induction (coded: –0.5 = egalitarian; +0.5 = discriminatory), identification (centered), and all possible interactions between these three variables.

Prejudice. The analysis revealed a significant main effect of ingroup identification, B = 0.60, F(1, 70) = 8.42, p = 0.005, ηp2 = 0.11. The more participants were identified with their ingroup, the higher their prejudice score. Furthermore, this effect was qualified by an intergroup similarity × ingroup norm × identification interaction, B = 2.62, F(1, 70) = 9.85, p = 0.002, ηp2 = 0.12. Identification was positively related with prejudice in the low similarity/discriminatory norm condition, B = 1.30, F(1, 70) = 7.70, p = 0.007, ηp2 = 0.10 (H1a), and in the high similarity/egalitarian norm condition, B = 1.22, F(1, 70) = 11.66, p = 0.001, ηp2 = 0.14 (H1b). In both conditions, the more participants were identified with their group, the more prejudiced they were. Identification was not related to prejudice either in the low similarity/egalitarian condition, B = 0.086, F(1, 70) = 0.05, p = 0.83, ηp2 < 0.01, or in the high similarity/discriminatory condition, B = –0.01, F(1, 70) = 0.001, p = 0.98, ηp2 < 0.01. Results are depicted in Figure 1.

Figure 1 

Prejudice score as a function of intergroup similarity (high vs. low), ingroup norm (egalitarian vs. discriminatory) and identification.

Stereotyping. The analysis only revealed a significant interaction between the three factors, B = –1.58, F(1, 70) = 4.42, p = 0.04, ηp2 = 0.061. Ingroup identification was negatively related to stereotyping only in the high similarity/egalitarian norm condition, B = –0.98, F(1, 70) = 9.28, p = 0.003, ηp2 = 0.12 (H2). Congruent with the hypothesized compensatory mechanism, the more participants identified with their ingroup, the lower their stereotyping score in the loyalty conflict condition. Identification was not related to stereotyping in any other condition: in the similarity/discriminatory norm condition, B = –0.09, F(1, 70) = 0.04, p = 0.83, ηp2 < 0.01; in the low similarity/egalitarian norm condition, B = 0.47, F(1, 70) = 1.39, p = 0.17, ηp2 = 0.03; in the high similarity/discriminatory norm condition, B = 0.04, F(1, 70) = 0.09, p = 0.92, ηp2 < 0.01. Results are depicted in Figure 2.

Figure 2 

Stereotyping score as a function of intergroup similarity (high vs. low), ingroup norm (egalitarian vs. discriminatory) and identification.

Discussion

The present research investigated how group members deal with loyalty conflict (i.e., their opposing motivation to conform to an egalitarian norm and defend their group from an intergroup threat) on different dimensions of intergroup differentiation. Accordingly, we examined the relationship between ingroup identification, ingroup norm and intergroup similarity on stereotyping and prejudice. First, our loyalty conflict hypothesis, and the compensatory mechanism it implies, suggest that high identifiers should seek to restore ingroup positive distinctiveness while simultaneously conforming to their ingroup norm. Accordingly, ingroup identification was positively related to prejudice, but negatively related to stereotyping, when the groups were presented as similar and the norm was egalitarian. On the one hand, being motivated to both defend their ingroup and to conform to its norm, high identifiers should experience a loyalty conflict. This conflict, we argue, should be resolved by defending the ingroup against this distinctiveness threat on one dimension of differentiation while conforming to the ingroup norm on the other.

While high identifiers are concerned with the fate of their ingroup and are likely to defend the ingroup against intergroup threats, low identifiers, on the other hand, are less invested in the group, and are more concerned with their personal identity. They are more likely to take a more opportunistic and individualistic stance towards group membership, and would only maintain a relationship with the group if it serves their self-interests. Threatened by a lack of distinctiveness between them and the outgroup at the individual level, even more so if the group norm is proscribing intergroup differentiation, low identifiers could be at risk of a categorization threat (Branscombe et al, 1999), which would result in their strategical distancing from the outgroup (Jones & Pittman, 1982), hence a heightened differentiation on the perceptual level (e.g., stereotyping).

Second ingroup identification was also positively related to prejudice when the norm was discriminatory and intergroup similarity was low. We argue that it would be the case for three main reasons: (a) mere conformity to the ingroup norm, (b) mere distinctiveness, or (c) combination of both conformity to a discriminatory norm and intergroup distinctiveness. In this condition, one could argue that differentiation merely reflects conformity to the ingroup norm. As noted earlier, ingroup identification has been consistently shown to facilitate conformity to ingroup norms (e.g., Christensen, et al., 2004; Jetten, Postmes & McAuliffe, 2002; Tajfel & Turner, 1986). However, the mere conformity effect seems implausible regarding our results. Such an effect should have implied a 2-way ingroup norm by ingroup identification interaction. Identification should also be related to prejudice in the high similarity/discriminatory norm, even if to a lesser extent. Contrary to this prediction, though, our results failed to unveil an effect of identification in the latter condition.

Increased differentiation in the low similarity/discriminatory ingroup norm condition can also be attributed to the mere effect of distinctiveness. A salient intergroup distinctiveness may lead high identifiers to perceive both outgroup and ingroup as more homogeneous and distinct from one another than low identifiers (Doosje, Spears et al., 1999; Haslam, Oakes, Turner & McGarty, 1995, 1996). As high identifiers would hold more positive perceptions of their ingroup than low identifiers, it is plausible to expect differences between low and high identifiers with respect to subsequent levels of perceived group distinctiveness and intergroup differentiation. Like the mere conformity hypothesis, however, this mere distinctiveness hypothesis implies a 2-way intergroup distinctiveness by ingroup identification interaction. Again, our results failed to support such an effect.

Even though these findings provide consistent evidence in support of our main hypotheses, and, in particular, of how group members deal with loyalty conflict through compensatory mechanisms, several limitations need to be addressed. One first limitation may refer to reasons why participants show conformity to the egalitarian ingroup norm (i.e., compensation effect) through decreasing stereotyping instead of prejudice. Indeed, one may argue that in same conditions participants could have protected their group from distinctiveness threat by increasing stereotyping, whilst showing conformity to ingroup norms (i.e., compensation effect) through prejudice. Even if stereotyping can also constitute a relevant response to distinctiveness threat, the present findings suggest that prejudice fulfills ingroup defensive goals to a greater extent than stereotyping. This finding is consistent with both Jetten et al.’s (2004) reactive distinctiveness hypothesis and Gabarrot et al.’s (2009) finding that intergroup similarity induces more ingroup bias and modern prejudice. Further research is needed to investigate the conditions under which stereotyping versus prejudice may represent a more appropriate strategy to defend the ingroup from intergroup threats.

A second limitation regards the fact that group identification did not predict stereotyping in the low similarity and discriminatory norm condition. Since ingroup norm and intergroup distinctiveness are consistent, one should expect identification to be positively related to differentiation in this condition. An explanation would be that, as intergroup distinctiveness is high and additionally confirmed by the ingroup norm, neither high identifiers should feel the need to differentiate/stereotype the outgroup to reinforce it, nor would low identifiers be at risk to be confused with the outgroup. As such, this result is consistent with Mugny’s perspective on social influence which goes beyond mere compliance, but would rather result from the elaboration of conflicting (or not conflicting) situational, normative, and individual factors (e.g., Pérez & Mugny, 1993). Future research is welcome in order to clarify whether participants in this condition were satisfied with highlighted intergroup differences and did not experience any kind of conflict regarding discriminatory ingroup norms.

A third limitation may regard the fact that overall differentiation—in particular, prejudice—is relatively low. Indeed, high identifiers seem to show little prejudice against the similar or the different outgroup in the egalitarian norm condition. One first explanation could rely on the fact that this study compared conditions of high intergroup similarity (i.e., the two groups were similar on all the traits in the similarity manipulation) to relatively lower (but not absolutely low) similarity (i.e., the two groups were similar on half the traits, and different on the other half of the traits, used in the manipulation of similarity). Therefore, a more extreme manipulation of intergroup distinctiveness might have increased prejudice in these specific conditions. A second explanation would be that the study is carried out in a highly ideological and politicized context, and French nationals are not motivated to show high levels of prejudice in order to provide a positive self-image. Finally, a third explanation may relate to the sample of university students enrolled in psychology studies (i.e., a hierarchy-attenuating university major; Dambrun, Guimond & Duarte, 2002). Therefore, our sample does not constitute a representative sample of French population regarding national identification and prejudice level. In sum, further research is needed in order to investigate this issue.

Another limitation regards the specific mechanism activated by the similarity manipulation, which may not be clear enough. For instance, is there a confound between the similarity/distinctiveness manipulation and groups’ stereotypes, or even a threat to ingroup stereotype? Were participants influenced by intergroup similarity, or instead threatened by the apparent refutation of the stereotypes they possess on both their ingroup and the outgroup? How do participants typically rate the groups onto the traits that were manipulated? Although future research should clarify the specific mechanism behind the present induction, we would like to emphasize that the traits on which the groups were presented as being similar or different were counter-balanced, which reduces concerns regarding the possibility that some traits were a priori stereotypical of one group or the other. Furthermore, if the manipulation had threatened ingroup or outgroup stereotypes, participants should have tried to restore these stereotypes (and thus increase stereotyping), specifically in the high similarity condition, and even more so when they identify with their ingroup. The results regarding stereotyping seem to contradict this interpretation.

A final limitation may relate to the two measures used to assess differentiation. Intergroup differentiation was assessed through both the relative attribution of positive and negative traits to the ingroup and the outgroup (i.e., prejudice) and differences in stereotypical and counter-stereotypical attribution for each group (i.e., stereotyping). However, one could wonder whether the effects on the prejudice measure reflect an increase in the attribution of positive traits to the ingroup (ingroup favoritism) or an increase of the attribution of negative traits to the outgroup (outgroup derogation). Similarly, one could wonder whether the effects on the stereotyping measure reflect an increase in the attribution of stereotypical traits to the ingroup (ingroup homogeneity effect) or an increase of the attribution of stereotypical traits to the outgroup (outgroup homogeneity effect). We chose to compute only two measures of intergroup differentiation because the purpose of our paper was to examine the effect of intergroup similarity, ingroup norm and ingroup identification on differentiation and not the mechanisms underlying differentiation. Further research, however, would provide a more fine-grained account of how the loyalty conflict is elaborated using these more focused indicators.

To conclude, in Western societies, there seems to be a cultural norm proscribing prejudice and discrimination, promoting tolerance and social equality (Eurobarometer, 2015). Yet, despite the power usually attributed to norms to regulate human behaviour and intergroup relations (e.g., Crandall, Eshleman & O’Brien, 2002; Louis, Duck, Terry, Schuller & Lalonde, 2007; Jetten, Spears & Manstead, 1996), discrimination toward immigrants persists, and is even considered as the most widespread form of discrimination in the EU (Eurobarometer, 2015). Consequently, this issue leaves open the question of the limits of the influence of egalitarian norms on intergroup attitudes. The present research sheds light on this issue by providing evidence in support of the general hypothesis that individuals conform to egalitarian social norms only to the extent that these norms are congruent with alternative motivations based on individuals’ understanding of the situation (Pérez & Mugny, 1996; see also Hornsey, et al., 2003). Indeed, when immigrants are perceived as too similar to the national group (i.e., distinctiveness threat), national identification increases motivation both to conform to the egalitarian norm and to restore intergroup positive distinctiveness. In order to deal with this loyalty conflict, high identifiers defended their ingroup identity on an evaluative dimension (i.e., prejudice) while showing conformity to the egalitarian norm on a descriptive dimension (i.e., stereotype).