Abstract
Desecrating a national symbol is a powerful means of protest or of showing antipathy for a national group, but how do such actions impact ingroup favoritism? We investigated this issue via two field studies conducted prior to the France versus Ireland (Study 1, N = 72) and France versus Germany (Study 2, N = 165) matches at the Euro 2016 soccer tournament. We asked French participants to imagine the ingroup/competition outgroup flag being burnt by ingroup/competition outgroup perpetrators. Imagining the ingroup flag being burnt increased proingroup bias through increase in either ingroup favoritism (Study 1) or outgroup derogation (for all outgroups, including those unconnected with the threat; Study 2). Perpetrators’ group membership did not have the expected moderating effect. We discuss the implications of these results for social identity defense strategies and for the consequences of intragroup versus intergroup threats.
Keywords
The NFL has decided that it will not force players to stand for the playing of our National Anthem. Total disrespect for our great country!
In 2016 and 2017, many African American National Football League (NFL) players knelt during the national anthem in order to protest racial injustice in the United States, an act that rekindled debate about the desecration of national symbols, freedom of expression, and disloyalty to one’s country. Some Americans, including President Trump, reacted defensively to these protests, criticizing the players and affirming the country’s greatness. In France, whistling during the French national anthem before a soccer match between France and Algeria, in 2001, led the French government to introduce legislation protecting national symbols. These two examples suggest that desecration of a national symbol elicits protective attitudes among the ingroup. The present study examines for the first time the effect of desecrating a national symbol (i.e., a country’s flag) on proingroup bias, and whether this effect varies according to whether the perpetrators are ingroup or outgroup members.
National Symbols and Social Identity
According to social identity theory (Tajfel & Turner, 1986), categorization enables individuals to structure their environment and define their place within it by identifying with certain social groups. Comparing one’s own social group with salient outgroups allows individuals to acquire and maintain a distinct and meaningful social identity (Tajfel, 1974). In fact, ingroup distinctiveness often appears to be more important to an individual’s social identity than the positive nature of that identity (Mlicki & Ellemers, 1996). This need for distinctiveness is expressed via the boundedness of the ingroup, which results in a differentiation of the ingroup from outgroups that is expressed via proingroup bias, mostly in the form of more positive attitudes toward the ingroup (Tajfel & Turner, 1986). The need for distinctiveness can be partly fulfilled by national symbols, which strengthen both group cohesion and the ingroup’s distinctiveness (Callahan & Ledgerwood, 2016; Firth, 1973; Mach, 1993; Welch, 2000).
The role national symbols play in maintaining ingroup distinctiveness involves both intragroup and intergroup processes. For example, conscious or subliminal exposure to national symbols increases identification among national group members (Kemmelmeier & Winter, 2008), produces a shift toward conservative national ideologies (Carter, Ferguson, & Hassin, 2011a, 2011b), and reduces hostility toward ethnic or religious ingroup minorities (e.g., Arabs and Muslims; Butz, Plant, & Doerr, 2007). On the other hand, such exposure also increases hostility toward outgroups, such as increased prejudice against foreigners (Becker, Enders-Comberg, Wagner, Christ, & Butz, 2012; Butz, 2009) or greater accessibility of intergroup aggressive concepts (Ferguson, Carter, & Hassin, 2009). These findings suggest that national symbols are important for group life, as they strengthen ingroup distinctiveness by triggering intragroup strategies to increase ingroup cohesion and intergroup strategies to increase differentiation from outgroups.
Desecration of National Symbols and Identity Threat
Destroying or desecrating national symbols is considered immoral (Haidt, Koller, & Dias, 1993), even by children (Helwig & Prencipe, 1999). Because symbols are vectors of group distinctiveness, their desecration would be expected to trigger defensive responses from the group whose symbol is attacked. Wohl, Branscombe, and Reysen (2010) found evidence supporting this premise in one of a series of studies examining the threat the destruction of symbols poses to group distinctiveness. In this study (Study 1 in their paper), imagining their university’s sports team’s mascot was about to be banned resulted in greater collective angst among American university students. Wohl and colleagues accounted for these findings by suggesting that this type of threat tends to prompt intragroup defense strategies to strengthen the ingroup (e.g., donating to charities that serve the ingroup, teaching children about the ingroup’s culture), but they did not examine whether reactions vary according to the source of the threat or whether they include intergroup defensive strategies. This is the issue we address in the present research.
On the one hand, social identity theory suggests that ingroup defensive strategies in response to the desecration of a national symbol should occur whatever the source of the desecration (ingroup or outgroup). First, defensive strategies may appear when the threat comes from an outgroup member. Indeed, according to social identity theory, a threat to ingroup distinctiveness originating from an outgroup leads to enhancement of positive attitudes toward the ingroup, thereby increasing ingroup favoritism (e.g., Branscombe, Ellemers, Spears, & Doosje, 1999).
Second, when the source is endogenous, any threat to a national symbol constitutes a counternormative behavior or a criticism of the ingroup. However, two possible outcomes are usually found among other ingroup members as a function of the intergroup context and whether the criticism is viewed as constructive or not. Compared with an outgroup deviant, an ingroup deviant may be judged either less harshly, if the criticism is seen as constructive for the ingroup (following the intergroup sensitivity effect; Hornsey & Imani, 2004), or more harshly, if the deviance threatens the group’s image (as postulated by the black sheep effect theory; Marques, Yzerbyt, & Leyens, 1988). However, when criticism occurs in a conflictual intergroup context, the intergroup sensitivity effect disappears due to the need for the ingroup to be united (Ariyanto, Hornsey, & Gallois, 2010). Thus, desecrating a national symbol in a salient competitive intergroup context (in which symbols are needed to enhance group cohesion; von Scheve, Beyer, Ismer, Kozłowska, & Morawetz, 2014) is unlikely to be considered constructive criticism. Rather, it is likely to be viewed as counternormative criticism and, in line with the black sheep effect theory, ingroup members might chastise the deviant and take steps to strengthen the group’s cohesion (e.g., Hutchison, Abrams, & de Moura, 2013; Travaglino, Abrams, Randsley de Moura, Marques, & Pinto, 2014).
On the other hand, social identity theory also suggests that proingroup bias is manifested by both more positive attitudes toward the ingroup and more negative attitudes toward outgroups (e.g., increased hostility toward outgroups). Hostility toward outgroups emerges when the ingroup/outgroup distinction is threatened, especially when the threat comes from the outgroup, although it can also arise following an endogenous threat (e.g., a threat due to heterogeneity in the ingroup; Falomir-Pichastor & Frederic, 2013). In the case of an exogenous threat, hostility may be expressed in different ways, including increased derogation of the outgroup (e.g., Scheepers, Spears, Doosje, & Manstead, 2006a, 2006b), increased prejudice (e.g., Branscombe et al., 1999; Stephan & Stephan, 2000), and greater discrimination (e.g., Falomir-Pichastor, Gabarrot, & Mugny, 2009; Scheepers, Spears, Doosje, & Manstead, 2003).
To the best of our knowledge, no studies have directly examined the consequences of desecrating a national symbol for attitudes toward outgroups. According to social identity theory, threats to a national symbol in a competitive intergroup context should lead ingroup members to show greater hostility toward the threatening outgroup. However, ingroup members may also show increased hostility toward outgroups with which the ingroup is not directly confronted, as ingroup members may respond to a threat to their social identity by attempting to increase its distinctiveness from all relevant outgroups, which may not be solely those responsible for the threat. Such increased hostility toward a threatening outgroup combined with a generalization of this hostility to other outgroups has been reported by a number of studies (e.g., Brambilla & Butz, 2013; Branscombe & Wann, 1994). For example, Branscombe and Wann (1994) asked participants to watch a boxing match that was won by either an American boxer (nonthreatening condition) or a Russian boxer (threatening condition). The threatening condition resulted in an increase in negative attitudes not only toward Russians, but also toward other national outgroups who were “threat-irrelevant.” Hence, a threat to an ingroup symbol can increase group members’ negative attitudes toward outgroups, whether or not these outgroups are responsible for the threat.
Taken together, these studies suggest that desecrating an ingroup’s national symbol may lead to a generalized increase in ingroup bias toward outgroups, whether or not they are seen as threatening. Moreover, destruction of the ingroup symbol by an outgroup perpetrator may increase ingroup bias not only in terms of more positive attitudes toward the ingroup, but also in terms of more negative attitudes toward the perpetrating (threatening) outgroup.
Overview and Hypotheses
In the present research we examined the consequences of destroying a national symbol for proingroup bias as a function of the group membership of the perpetrators of the destructive act. We also investigated whether this effect leads to more positive attitudes toward the ingroup, or more negative attitudes toward a salient competitive outgroup and/or other outgroups not involved in the competition or the threat. We carried out two field studies with real groups of supporters (in fan zones) who had come to watch a soccer match between France and the Republic of Ireland (Study 1) or between France and Germany (Study 2). Testing our hypotheses in the context of a sports competition enabled us to make use of intergroup rivalries and their underlying social identities (Cheng et al., 2011; Johnson et al., 2006; Scheepers et al., 2003, 2006a; Torelli, Chiu, Tam, Au, & Keh, 2011). Moreover, this collective event and intergroup competition context allowed us to study a situation of high commitment to the ingroup (Castano, Yzerbyt, & Bourguignon, 2003; von Scheve et al., 2014).
In both studies we manipulated the threat to national identity posed by the burning of either the ingroup’s or the outgroup’s national flag. The competition outgroup was the national group (Irish or German) whose team was involved in the soccer match against France. We manipulated the group to which the perpetrators belonged by presenting them as either ingroup members or competition outgroup members. Following these inductions, we measured attitudes toward the French ingroup, the competition outgroup (outgroup involved in the match), and noncompetition outgroups (national and transnational groups not involved in the competition).
Our review of the literature led us to formulate the following hypotheses:
H1: Destruction of the ingroup’s (vs. outgroup’s) national symbol will enhance proingroup bias against both the competition outgroup and noncompetition outgroups.
H2: Proingroup bias against the competition outgroup will be even stronger when the ingroup symbol is destroyed by a member of the competition outgroup (rather than a member of the ingroup).
Moreover, in order to examine whether proingroup bias takes the form of more positive evaluations of the ingroup or more negative evaluations of the outgroups, we examined the evaluations of each group (ingroup, competition outgroup, and noncompetition outgroups) separately.
Study 1
Study 1 was conducted during the Euro 2016 soccer tournament, prior to the match between France and the Republic of Ireland. We created three experimental conditions on the basis of the flag that was burnt and the group responsible for burning it.
Method
Population
We recruited 77 participants in the fan zone on the Champ de Mars, in Paris, prior to the match between France and the Republic of Ireland. We eliminated four participants who were not born in France and one participant who did not complete all of the measures. The final sample comprised 72 French people (30 women, 42 men), aged between 18 and 56 years (M = 24.67, SD = 5.43). This sample allowed us to detect a low to moderate effect size (f = .18), with an alpha of 0.05%, in a mixed ANOVA design with a between-subjects variable (three groups) and a within-subjects variable (two measures), and with a power of .78 (G*Power; Faul, Erdfelder, Lang, & Buchner, 2007).
Procedure
Participants were interviewed a few hours before the match began. After obtaining the participants’ consent and collecting identification information, 1 we asked them to imagine seeing a national flag being burnt. We created three experimental conditions. Participants in the ingroup symbol–ingroup perpetrators condition read the following instruction: “Imagine the following scenario occurs this evening: During the match, a group of young French people deliberately burn a French flag in the middle of Paris.” In the other conditions, a French flag was burnt by Irish people (ingroup symbol–outgroup perpetrators) and an Irish flag was burnt by French people (outgroup symbol–ingroup perpetrators). Participants then answered questions on how they felt about the situation they had been asked to imagine, completed a feeling thermometer measure, and provided sociodemographic information. Finally, we debriefed them and thanked them for their time.
Measures
We used a feeling thermometer measure to determine indices for attitude toward the ingroup, attitude toward the competition outgroup, and attitude toward noncompetition outgroups. Participants had to rate on 11-point scales (1 = very unfavorable, 11 = very favorable) 10 social groups, including the ingroup, the competition outgroup, four noncompetition outgroups, and four distractor groups. 2 We analyzed their evaluations of the ingroup (the French group; M = 9.69, SD = 2.03), the competition outgroup (in this study, the Irish group; M = 9.99, SD = 1.73), and noncompetition outgroups (mean for the national and transnational outgroups not involved in the competition; α = .84, M = 9.21, SD = 1.92; e.g., Dambrun & Guimond, 2001). Means and standard deviations within conditions are reported in Table 1.
Observed means and standard deviations of evaluation of the ingroup, competition outgroup, and noncompetition outgroups for each experimental condition (Study 1; N = 72).
Results
Preliminary analyses
Preliminary analyses showed that participants’ age, F(2, 71) = 0.25, p = .781, η2p = .01; and sex, χ2(2) = 1.00, p = .607, did not vary as a function of experimental conditions.
Proingroup bias against the competition outgroup
We ran a 3 (condition: ingroup symbol–ingroup perpetrators vs. ingroup symbol–outgroup perpetrators vs. outgroup symbol–ingroup perpetrators) x 2 (group evaluated: ingroup vs. competition outgroup) ANOVA with condition as a between-subjects factor and group evaluated as a within-subjects factor. The main effects of group evaluated, F(1, 69) = 1.19, p = .279, η2p = .02, and condition, F(2, 69) = 1.88, p = .161, η2p = .05, were not significant, but the group evaluated by condition interaction effect was significant, F(2, 69) = 8.91, p < .001, η2p = .21. Decomposing this interaction showed that the interaction effect was due to a pro-outgroup bias when the outgroup flag was burnt by an ingroup member, F(1, 69) = 17.98, p < .001, η2p = .21. We did not find any significant bias in any of the ingroup symbol conditions, all Fs < 1.11, all ps > .296, all η2p s < .02 (see Figure 1).
Evaluation of the ingroup and competition outgroup for each experimental condition (Study 1).
Proingroup bias against noncompetition outgroups
We conducted a 3 (condition: ingroup symbol–ingroup perpetrators vs. ingroup symbol–outgroup perpetrators vs. outgroup symbol–ingroup perpetrators) x 2 (group evaluated: ingroup vs. noncompetition outgroups) ANOVA with evaluations of the ingroup and noncompetition outgroups as within-subjects factors and condition as a between-subjects factor. The main effect of condition was not significant, F(2, 69) = 1.95, p = .150, η2p = .05. However, the analysis showed a significant main effect of group evaluated, F(1, 69) = 8.86, p = .004, η2p = .11, with the ingroup (M = 9.78, SD = 0.23) being evaluated more positively than the noncompetition outgroups (M = 9.18, SD = 0.23). The group evaluated by condition interaction effect was also significant, F(2, 69) = 8.97, p < .001, η2p = .21. Decomposing this interaction showed a proingroup bias when the ingroup flag was burnt by outgroup members, F(1, 69) = 17.84, p < .001, η2p = .21. We did not observe any bias when the ingroup flag was burnt by ingroup members, F(1, 69) = 3.81, p = .055, η2p = .05, or when the outgroup flag was burnt, F(1, 69) = 2.21, p = .142, η2p = .03 (see Figure 2).
Evaluation of the ingroup and noncompetition outgroups for each experimental condition (Study 1).
Additional analyses of the separate evaluations of groups
We conducted additional decomposition analyses of the group evaluated by condition interaction for separate evaluations of the different groups. The French group was evaluated more positively when the French (vs. Irish) flag was burnt, regardless of the perpetrators’ group membership (Ingroup symbol–ingroup perpetrators versus outgroup symbol–ingroup perpetrators: F(1, 69) = 7.78, p = .007, η2p = .10; ingroup symbol–outgroup perpetrators versus outgroup symbol–ingroup perpetrators: F(1, 69) = 9.73, p = .003, η2p = .12; ingroup symbol–ingroup perpetrators versus ingroup symbol–outgroup perpetrators: F(1, 69) = 0.31, p = .578, η2p = .01). However, condition had no significant effect on evaluations of the Irish group, all Fs < 0.32, all ps > .574, all η2p s < .01, or on evaluations of the noncompetition outgroups, all Fs < 1.31, all ps > .255, all η2p s < .02.
Discussion
The results of Study 1 partially support Hypothesis 1. As expected, we found more positive evaluation of the ingroup when the ingroup symbol was burnt than when the outgroup symbol was burnt. No effect of the flag being burnt was observed on evaluations of the outgroups. Analyses on biases revealed a proingroup bias against noncompetition outgroups when the French (vs. Irish) flag was burnt. However, this effect appeared only when the ingroup flag was burnt by outgroup members, but not by ingroup members, nor when the Irish flag was burnt. Proingroup bias against the Irish group was not significant when the ingroup flag was burnt (although it was higher than in the Irish flag condition). Our results do not fully support Hypothesis 2. We observed a significant effect on proingroup bias when ingroup flag was burnt by outgroup but not by ingroup members. Moreover, we did not obtain any significant effects as a function of perpetrators’ group membership on ingroup evaluation.
This first study had at least two limitations. First, a football match between France and the Republic of Ireland does not seem to have been a very threatening intergroup context for participants, which may have led to a lower level of perceived competition. In fact, participants’ overall evaluations of the Irish group were positive, so we did not obtain the expected proingroup bias score, and we actually obtained a pro-outgroup bias when ingroup perpetrators burnt the Irish flag. French people’s mainly positive views of Irish people (the competition outgroup) may, at least partly, explain why we failed to find consistent effects on bias scores, namely as a function of the perpetrators. Second, the design we used included only three experimental conditions. The failure to include an outgroup flag–outgroup perpetrators situation may have created an imbalance in conditions that impacted the outgroup evaluation, thereby limiting our appraisal of the processes investigated. In order to address these limitations, we conducted a second study in a more competitive and threatening context, and in which we included this fourth, entirely outgroup condition (outgroup flag–outgroup perpetrators).
Study 2
Study 2 included two main changes compared with Study 1. First, it was conducted prior to the match between France and Germany, an outgroup that is generally perceived in France as quite threatening due to the history of conflict between the two countries (e.g., Germond, 2014), and because of their long-standing soccer rivalry (Museux, 2016). Accordingly, a pilot study (N = 53) confirmed significant differences in evaluations of Irish and German people in general, with German people being perceived as less warm, more competent, and more threatening. 3 Second, in this study we manipulated the symbol and the perpetrators as orthogonal factors in a 2 x 2 experimental design.
Method
Population
We recruited 177 participants in the fan zone on the Champ de Mars, in Paris, during the Euro 2016 soccer tournament. Eliminating participants who were not born in France (n = 12) gave us a final sample of 165 people (58 women, 107 men), aged between 18 and 49 years (M = 23.21, SD = 5.46). This sample allowed us to detect a low to moderate effect size (f = .18), with an alpha of 0.05%, in a mixed ANOVA design with a between-subjects variable (four groups) and a within-subjects variable (two measures), and with a power of .98 (G*Power; Faul et al., 2007).
Procedure
We interviewed participants a few hours before the France versus Germany match began. The procedure and measures were identical to Study 1, the only change being the content of the induction, with a 2 (perpetrators: French ingroup vs. German outgroup) x 2 (symbol: French ingroup vs. German outgroup) experimental design.
Measures
We followed the same procedure we used in Study 1, which gave us an ingroup score (French people; M = 9.79, SD = 1.76), competition outgroup score (German people; M = 8.55, SD = 2.93), and noncompetition outgroups score 4 (α = .84, M = 8.86, SD = 2.19). Means and standard deviations within experimental conditions are reported in Table 2.
Observed means and standard deviations of evaluation of the ingroup, competition outgroup, and noncompetition outgroups for each experimental condition (Study 2; N = 165).
Results
Preliminary analyses
There were no differences between the experimental conditions in terms of participant sex, χ2(3) = 1.86, p = .602, but the conditions did vary in terms of age, F(3, 161) = 4.43, p = .005, η2p = .08. A Tukey post hoc test revealed that the participants in the ingroup symbol–ingroup perpetrators condition were younger than the participants in the outgroup symbol–ingroup perpetrators, b = −3.82, se(b) = 1.14, p = .006, and outgroup symbol–outgroup perpetrators, b = −3.12, se(b) = 1.16, p = .040, conditions. As age is significantly correlated to evaluations of the ingroup, r = −.18, p = .024, we introduced age as a control variable in the following analyses. 5
Proingroup bias against the competition outgroup
We ran a 2 (symbol: ingroup vs. outgroup) x 2 (perpetrators: ingroup vs. outgroup) x 2 (group evaluated: ingroup vs. competition outgroup) ANCOVA, with symbol and perpetrators as independent variables, group evaluated as a within-subjects factor, and age (standardized score) as a control variable. This analysis showed a significant main effect of group evaluated, F(1, 160) = 36.26, p < .001, η2p = .19, indicating a proingroup bias due to the ingroup (M = 9.80, SD = 0.14) being evaluated more positively than the competition outgroup (M = 8.60, SD = 0.22). The main effects of symbol, F(1, 160) = 5.20, p = .024, η2p = .03, and of perpetrators, F(1, 160) = 4.12, p = .044, η2p = .03, were also significant. Evaluations of the groups were more positive when the German (vs. French) flag was burnt (ingroup symbol: M = 8.85, SD = 0.21; outgroup symbol: M = 9.55, SD = 0.22) and when the perpetrators were French (vs. German; ingroup perpetrators: M = 9.53, SD = 0.23; outgroup perpetrators: M = 8.88, SD = 0.22). The symbol by perpetrators interaction effect was not significant, F(1, 160) = 0.78, p = .379, η2p = .01. Neither the group evaluated by perpetrators interaction, F(1, 160) = 2.79, p = .097, η2p = .02, nor the group evaluated by symbol by perpetrators interaction, F(1, 160) = 1.25, p = .265, η2p = .01, had a significant effect.
More importantly, the interaction between group evaluated and symbol was significant, F(1, 160) = 9.80, p = .002, η2p = .06. Although proingroup biases were significant in both conditions, the difference between evaluations of the French group and evaluations of the German group was larger when the ingroup flag was burnt, F(1, 160) = 43.75, p < .001, η2p = .22, than when the outgroup flag was burnt, F(1, 160) = 4.00, p = .047, η2p = .02 (see Figure 3).
Evaluation of the ingroup and competition outgroup when the ingroup or outgroup flag was burnt (Study 2).
Proingroup bias against noncompetition outgroups
We ran a 2 (symbol: ingroup vs. outgroup) x 2 (perpetrators: ingroup vs. outgroup) x 2 (group evaluated: ingroup vs. outgroup) ANCOVA with symbol and perpetrators as between-subjects factors, group evaluated as a within-subjects factor, and age as a covariate. The analysis showed a significant main effect of group evaluated, F(1, 160) = 40.22, p < .001, η2p = .20, with the ingroup being evaluated more positively (M = 9.80, SD = 0.14) than the noncompetition outgroups (M = 8.90, SD = 0.17). The effects of perpetrators, F(1, 160) = 3.74, p = .055, η2p = .02; symbol, F(1, 160) = 2.31, p = .131, η2p = .01; the symbol by perpetrators interaction, F(1, 160) = 0.69, p = .409, η2p < .01; the group evaluated by perpetrators interaction, F(1, 160) = 2.54, p = .113, η2p = .02; and the group evaluated by perpetrators by symbol interaction, F(1, 160) = 1.48, p = .226, η2p = .01, were all not significant.
Finally, the effect of the group evaluated by symbol interaction was significant, F(1, 160) = 5.37, p = .022, η2p = .03. In both conditions, participants displayed a significant proingroup bias but this bias was greater when the French flag was burnt, F(1, 160) = 39.15, p < .001, η2p = .20, than when the German flag was burnt, F(1, 160) = 7.76, p = .006, η2p = .05 (see Figure 4).
Evaluation of the ingroup and noncompetition outgroups when the ingroup or outgroup flag was burnt (Study 2).
Additional analyses of the separate evaluations of groups
We decomposed the group evaluated by symbol interaction effects with respect to the evaluations of each group. The flag being burnt had no effect on evaluations of the ingroup, F(1, 160) = 0.08, p = .777, η2p < .01, but denigration of the German group was greater when the ingroup (vs. outgroup) flag was burnt, F(1, 160) = 9.02, p = .003, η2p = .05 (ingroup symbol: M = 7.94, SD = 0.31; outgroup symbol: M = 9.27, SD = 0.32). We found a similar effect for noncompetition outgroups, which were evaluated less favora-bly when the ingroup (vs. outgroup) flag was burnt, F(1, 160) = 4.92, p = .028, η2p = .03 (ingroup symbol: M = 8.53, SD = 0.23; outgroup symbol: M = 9.27, SD = 0.24).
Discussion
The results of Study 2, which was conducted in a more competitive and threatening intergroup context than Study 1, again provide consistent evidence in support of Hypothesis 1. Imagining the French (vs. German) flag being burnt increased ingroup favoritism against both the competition outgroup and noncompetition outgroups. Furthermore, when the ingroup flag was burnt, this effect was due to more negative evaluations of both the competition outgroup and noncompetition outgroups. Finally, as in Study 1, and despite a full 2 x 2 experimental plan, the results of Study 2 did not support Hypothesis 2.
General Discussion
The present research examined whether destroying a national symbol in a context of intergroup competition has an impact on ingroup favoritism. We expected an increase in ingroup bias against both the competition and noncompetition outgroups when the ingroup (rather than the outgroup) symbol was destroyed (H1). We also expected ingroup bias against a specific outgroup to increase when the symbol was destroyed by members of this outgroup (H2).
Both studies provided consistent evidence supporting H1, according to which proingroup bias increases when the ingroup (vs. outgroup) symbol is destroyed. More specifically, proingroup bias against noncompetition outgroups increased in both studies, and against the competition outgroup in Study 2. Moreover, complementary analyses showed that this increased bias was due to more positive attitudes toward the ingroup in Study 1, but the result of less positive evaluations of the outgroups in Study 2. However, the results of neither study supported H2, as knowing the ingroup versus outgroup identity of the perpetrators and that the symbol was destroyed by a member of the competition outgroup did not affect participants’ evaluations. We discuss these issues in the following sections.
Ingroup Favoritism Versus Outgroup Derogation
Our two studies show that ingroup members react defensively by increasing proingroup bias when an ingroup symbol is desecrated. However, attitudes toward groups differed between the studies. Although in Study 1 this bias took the form of more positive attitudes toward the ingroup, in Study 2 it was the result of less positive attitudes toward both the competition and noncompetition outgroups. There are a number of possible explanations for this finding.
First, our results suggest that desecration of a symbol does not lead group members to blindly defend their group; instead, they adopt a strategic identity defense, which may vary as a function of the context and the competing outgroup. Thus, participants evaluated the ingroup more positively when the competing outgroup was Irish people (seen as a warm and low-competence outgroup), but derogated all outgroups when the competing outgroup was German people (seen as a cold and high-competence outgroup). Although our focus on natural groups prevented us testing which feature of the salient outgroup determines the defense strategy adopted, our results are in line with research into the consequences of the perceptions of outgroups. On the one hand, perceiving outgroups as having low warmth is related to higher perceived competition between groups (Fiske, Cuddy, Glick, & Xu, 2002) and to higher perceived symbolic threat (Kervyn, Fiske, & Yzerbyt, 2015). This higher perception of threat may lead to increased derogation of outgroups (e.g., Stephan, Ybarra, & Rios, 2016). On the other hand, higher perceived competence, which is linked to high perceived status (Fiske et al., 2002), may also lead to greater outgroup derogation, especially in competitive contexts (Lehr, Ferreira, & Banaji, 2017; Scheepers et al., 2006a). Thus, perceptions of Germans as having a combination of low warmth and high competence may have led to an increase in outgroup derogation. Our findings are also in accordance with Scheepers et al.’s (2006a) theoretical proposal that ingroup bias may accomplish different functions depending on the intergroup context. For example, ingroup bias may have an instrumental function when the ingroup is comparing itself with a high-status outgroup, but an identity function when the ingroup is comparing itself with a low-status outgroup. This pattern can be seen in our results. When the French ingroup faced a high-status outgroup (Germans), it showed an instrumental bias expressed via outgroup derogation (e.g., Lehr et al., 2017). Conversely, when the French ingroup faced a lower status outgroup (Irish people), the increased bias was expressed via ingroup favoritism, which may have been motivated by the desire to show ingroup superiority, that is, by an identity confirmation function (Scheepers et al., 2003). The desecration of symbols elicits defense strategies which increase distinction and evaluative biases between the ingroup and the outgroup, but the strategies adopted differ according to the intergroup context.
Second, the increased ingroup bias triggered by the desecration of our participants’ group symbol affected noncompetition and irrelevant outgroups, as well as the competition outgroup. Although we did not find evidence of outgroup derogation in Study 1, both the competition and noncompetition outgroups were the targets of derogation in Study 2. These results suggest that ingroup members respond to threats to their social identity by derogating outgroups, even those that are not the source of the threat. Some studies have observed such an extension of defense strategies to outgroups that were not responsible for a threat, most notably via the scapegoat phenomenon (deprecation of a low-status outgroup when an identity threat emanates from another, higher status outgroup; Cadinu & Reggiori, 2002; Dollard, Miller, Doob, Mowrer, & Sears, 1939; Marcus-Newhall, Pedersen, Carlson, & Miller, 2000; Zawadzki, 1948), and in the case of macro-level threats (threats that cannot be attributed to a specific outgroup, such as climate change or the loss of cultural values, which lead to deprecation of outgroups stereotypically related to the macro-level threat; Becker, Wagner, & Christ, 2011; Brambilla & Butz, 2013; Butz & Yogeeswaran, 2011). Both of these phenomena (scapegoat and macro-level threat) involve deprecating outgroups that are not responsible for a threat, either because it is difficult to derogate the threatening outgroup (scapegoat phenomenon) or because there is no designated threatening outgroup (macro-level threat). Our studies contribute to this research by showing that deprecation may be generalized even when there is a salient and “disparagable” threatening outgroup. Some threats, such as attacking a group’s distinctiveness by desecrating its symbol, may elicit generalized defense strategies, even toward outgroups not involved in the threatening context. Interestingly, participants in Study 1 did not devaluate any outgroup, suggesting that the positive perception of the Irish group acted as a buffer against derogation of outgroups in response to a social identity threat. Further studies are needed to ascertain which dimension (warmth or competence) of outgroups determines whether ingroup defense strategies involve ingroup valorization or outgroup derogation, and to examine which outgroups (source of threat or not) are derogated.
Perpetrators and Symbol Desecration
Participants’ evaluations of groups did not vary as a function of the perpetrators’ group membership in either of our studies, but this does not mean that the processes underlying the ingroup’s defense strategy were the same in each scenario. Although we did not examine how participants interpreted the flag-burning, participants’ understanding of the threat may have varied according to the identity of the perpetrators. Research in sociology and law shows that flag-burning has become a way of criticizing and protesting actions and events within a society (Goode & Ben-Yehuda, 2009; Welch, 2000). However, people interpret criticism differently according to the critic’s group membership, with ingroup critics being perceived more positively than outgroup critics, whose criticism is perceived as hostile (e.g., Hornsey & Imani, 2004). We addressed this issue with respect to our studies by conducting a post hoc study in which we asked 176 students of French nationality to imagine a flag-burning during a sports competition. Overall, they perceived an intragroup flag-burning (i.e., ingroup flag burnt by ingroup members) more as criticism than aggression, and an intergroup flag-burning (ingroup flag burnt by outgroup members) more as aggression than criticism. Thus, in our two studies, the intergroup flag-burning and the intragroup flag-burning are likely to have been interpreted differently, even though they led to similar defensive responses. This may be because destroying a symbol constitutes a distinctiveness threat per se, no matter who the perpetrators are (Firth, 1973; Welch, 2000), and ingroup bias is one of the most prevalent defensive reactions to threats to ingroup distinctiveness (e.g., Jetten, Spears, & Postmes, 2004). The ways in which ingroup versus outgroup threats to the ingroup impact intergroup relations and processes have rarely been examined (Greenaway & Cruwys, 2018). Therefore, how and when flag-burning perceived as criticism and flag-burning perceived as aggression, or situations involving ingroup versus outgroup flag burners, lead to similar or different intergroup effects would be an interesting subject for future studies.
Limitations and Future Perspectives
Some characteristics of our research should be highlighted and deserve further attention. First, we investigated the consequences of threats to ingroup symbols for ingroup bias through an imagined scenario, instead of presenting this threat as real. Our methodology relied on mental imagery, which has been shown to elicit similar cognitive and motivational responses to real situations (Dadds, Bovbjerg, Redd, & Cutmore, 1997; Garcia, Weaver, Moskowitz, & Darley, 2002). In fact, simulated intergroup contact has been widely shown to lead to attitudinal and even behavioral intergroup responses to a variety of outgroup targets (e.g., Birtel & Crisp, 2012; Brambilla, Ravenna, & Hewstone, 2012; Ioannou, Hewstone, & Al Ramiah, 2017; Kuchenbrandt, Eyssel, & Seidel, 2013; Vezzali et al., 2015). Hence, inducing a social identity threat should trigger similar processes and responses as a function of whether the threat is induced via mental imagery or presented as real. Nevertheless, further research is needed to determine whether or not this actually holds true in the case of social identity threats.
Second, using real groups (rather than a student sample, for example) and a real context constitutes one of the major advantages of our study because it increases the external validity of the results (Brewer & Crano, 2014). Our field studies confirm the importance of research into how group members react to threats to ingroup symbols and, more broadly, into symbolic behaviors, which are still understudied in social psychology (Butz, 2009; Hobson, Schroeder, Risen, Xygalatas, & Inzlicht, 2017). Because the context in which we conducted our studies enhanced people’s need to express ingroup identities and their tendency to deprecate outgroups (Lehr et al., 2017; Scheepers et al., 2003; Weisel & Böhm, 2015), future research should include laboratory studies so it can more easily control internal validity and determine whether the same processes function in other contexts.
Our studies focused on the destruction of the French flag. However, as noted in the introduction, other symbols, such as a national anthem, can also be desecrated, without being destroyed. The significance of a symbol (e.g., one’s national flag) can differ between countries (e.g., Becker et al., 2017), just as the significance of different national symbols can differ within a country (e.g., Finell, Olakivi, Liebkind, & Lipsanen, 2013). Gilboa and Bodner (2009) showed that a national flag and a national anthem activate similar national concepts. However, in the context of a sports competition, anthems may serve a more functional goal, becoming a song sung to support one’s team (Scheepers et al., 2003; Slater, Haslam, & Steffens, 2018), whereas rituals and collective events may serve similar functions to flags (increase affiliation and loyalty; Hobson et al., 2017). Further studies are needed to examine similarities and differences in the consequences of attacks on national symbols according to a symbol’s function and the context in which it is attacked.
Finally, the impact desecrating a group’s symbols has on identity and intergroup relations is an important issue. Our study shed light on a nation’s defense reactions, which involve increased bias against outgroups whether or not they are directly linked to the threat. Our studies also showed that the effect of such a threat in a context of intergroup competition is similar whether the threat comes from an outgroup or the ingroup. Studying symbols and their desecration is complex and opens new opportunities for future research. Our results now need to be combined with further research in order to understand how and why hostility toward different types of outgroups emerges when the ingroup’s symbols are threatened. Such research into intergroup attitudes could provide interesting insights into the resurgence of populism and nationalism around the world, and of hostility toward ethnic and religious minorities (e.g., Human Rights Watch, 2017). Understanding these mechanisms will allow for the development of strategies to mitigate the defensive attitudes individuals adopt and thereby minimize the harmful intergroup consequences of identity threats.
