International Journal of Intercultural Relations 73 (2019) 59–73
Contents lists available at ScienceDirect
International Journal of Intercultural Relations journal homepage: www.elsevier.com/locate/ijintrel
“They are all alike”: When negative minority outgroups are generalized onto superordinate inclusive outgroups
T
Flavia Albarelloa, , Francesco Foronib, Miles Hewstonec, Monica Rubinia ⁎
a b c
Alma Mater Studiorum-University of Bologna, Italy Australian Catholic University, Australia University of Newcastle, Australia
ARTICLE INFO
ABSTRACT
Keywords: Outgroup-to-outgroup generalization Intergroup threat Negative minority outgroups Outgroup projection
Two studies investigated outgroup-to-outgroup generalization, addressing whether members of negatively perceived minority outgroups are perceived as prototypical of larger partially-inclusive outgroups and whether this tendency is enhanced under intergroup threat. Both experimental studies were conducted with Italian undergraduate participants. Experiment 1 (N = 186) tested whether Roma are generalized onto (i.e., perceived as prototypical of) Romanians to a higher extent under intergroup realistic and symbolic threat than under no-threat. It also explored the direction of the phenomenon and its underlying mechanism. Experiment 2 (N = 90) provided additional evidence on the effect of threat on such generalization phenomenon considering a different pair of groups (Islamic terrorists, Arabs) and employing a less blatant measure of generalization. Overall, these studies provided consistent evidence of outgroup-to-outgroup generalization, clarifying the conditions of its occurrence. These results contribute to the understanding of the phenomenon of outgroup-to-outgroup generalization and of its impact on outgroup perception and prejudice.
Introduction “They are all alike” (Brown, 2011, p. 49) is the way outgroup members are often perceived and judged. This common sense inference derives from the outgroup homogeneity effect (Park & Rothbart, 1982; Simon, 1992) and forms the basis of the generalization tendency on which stereotypes are rooted (Judd & Park, 1988). Rothbart, Fulero, Jensen, Howard, and Birrell (1978) showed that people generalize individual negative exemplars to the whole group by providing experimental evidence that those members who were associated with extreme negative behaviours were disproportionally overestimated in the group impression. This evidence reveals that generalization processes on which stereotypes rely can be particularly enhanced when a social group is composed of extremely negative exemplars. Would a similar effect be found across group boundaries resulting in outgroup-to-outgroup generalization? A striking real-life example of this phenomenon concerns the tendency to consider Muslims, a huge human group, as if they were all Islamic terrorists. This tendency has been particularly pronounced after the 11th September terrorists attacks on the Twin Towers, New York, and the 2015 attack at the Charlie Hebdo magazine in Paris. The same generalization tendency appeared more recently after the terrorist attack at a Strasburg Christmas market by a French man who happened to have parents from Morocco. These events highlight that people, when threatened, tend to generalize negative minority outgroup categories onto another superordinate (partially) inclusive
⁎
Corresponding author at: Department of Psychology, Alma Mater Studiorum–University of Bologna, Viale Berti Pichat 5, 40126, Bologna, Italy. E-mail address:
[email protected] (F. Albarello).
https://doi.org/10.1016/j.ijintrel.2019.08.003 Received 26 June 2018; Received in revised form 22 July 2019; Accepted 25 August 2019 Available online 05 September 2019 0147-1767/ © 2019 Elsevier Ltd. All rights reserved.
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
outgroup category, thus leading to social exclusion of all people encompassed in the latter category. In this vein, Albarello and Rubini (2011) showed that people generalize negative stereotypes of Roma (Claps & Vitale, 2011; Kende, Hadarics, & Láštiková, 2017; Orosz et al., 2018; Villano, Fontanella, Fontanella, & Di Donato, 2017) to the larger (partially) inclusive group of Romanians. Moreover, Albarello, Foroni, Hewstone, and Rubini (2017) highlighted that intergroup threat (Stephan & Stephan, 2000) increased the generalization of Roma exemplars onto Romanians, that is, under threat participants estimated that a larger proportion of Roma exemplars were contained in a set of pictures of Romanians in a perception task. Such initial evidence suggests that when there is a minority-majority relationship (Albarello, Prati, et al., 2019; Brandstätter et al., 1991) between groups and the minority group is perceived more negatively, intergroup threat may lead individuals to generalize the negative minority onto the less negative superinclusive outgroup, as if they were all alike. The present contribution aimed at going deeper into this phenomenon by addressing the conditions of its appearance. First, relying on the evidence gathered by Rothbart et al. (1978) at the intragroup level and by Albarello and Rubini (2011; see also Albarello et al., 2017) at the intergroup level, it is likely that when an outgroup is evaluated in a very negative fashion and it can be included in a larger inclusive outgroup its negative characteristics are generalized onto the larger outgroups. Moreover, this phenomenon can be enhanced under intergroup threat (Albarello et al., 2017). In addition, at least three issues remain still unexplored. First, it is important to test whether the outgroup-to-outgroup generalization occurs from the minority negative outgroup to the superinclusive less negative outgroup and not vice versa. Secondly, the mediational mechanism of the phenomenon has still to be explored. Third it is also important to address the extent to which the phenomenon is extended to different groups that maintain the aforementioned necessary features. To address these issues, Study 1 focused on the negatively valued outgroup of Roma and the largerinclusive, less negatively valued outgroup of Romanians. Besides addressing the sine qua non conditions for outgroup-to-outgroup generalization to occur, Study 1 tested whether intergroup threat enhances the outgroup-to-outgroup generalization in terms of perceived prototypicality of exemplars of one group for the other (see Mummendey & Wenzel, 1999) and addressed the mediation of fear as stemming from intergroup threat and repulsion emotions elicited by groups on outgroup-to-outgroup generalization. Study 2 was run to provide convergent evidence for the outgroup-to-outgroup generalization using two different groups (i.e., Islamic terrorists, Arabs). Moreover an implicit linguistic measure of generalization was adopted to provide further support to the outgroup-tooutgroup generalization with a less controllable measure. Generalization processes Rothbart et al. (1978) showed that group members who were associated with extreme (deviant) negative behavior(s) were more available in memory and disproportionately represented in group impressions. Similar evidence was found by Haslam, Oakes, Turner, and McGarty’s (1995) work on outgroup prototypicality in which they showed that the perceived representativeness of extreme and moderate members of the same outgroup varied depending on the salience of intergroup division: that is, the extremist outgroup member was seen as more prototypical of the outgroup than the moderate one when intergroup division was made salient. Such evidence provides a basis to contend that generalization processes go from the more negative subcategory/minority to the less negative partially-inclusive/majority category. This contribution also suggests that one way to investigate how exemplars of one outgroup can be generalized onto another outgroup is through exemplar prototypicality. Mummendey and Wenzel (1999), in their work on ingroup projection, considered the prototypicality of ingroupers and outgroupers over the superordinate common category. Drawing from this, and going a step further by taking a group-to-group level of analysis, Albarello and Rubini (2011) showed that people generalize the content of stereotypes of Roma (as being characterized by dishonesty, lack of concern for their children and of refinement) to the larger partially-inclusive outgroup of Romanians. Albarello et al. (2017) showed that intergroup realistic threat (i.e., threats to the ingroup’s existence, economic and political power, or physical or material well-being; Stephan & Stephan, 2000) and symbolic threat (i.e., threats related to the outgroup’s different morals, values, beliefs and standards; Stephan & Stephan, 2000) led participants to overestimate the number of Roma exemplars out of Romanians in experimentally presented samples. These findings by Albarello and colleagues underline that intergroup threat enhances people’s tendency to assimilate less negative outgroupers as if they were all belonging to the more negative minority outgroup. Very likely, the appearance of outgroup-to-outgroup generalization is due to the enhanced salience of intergroup distinctions (cf. Haslam et al., 1995) under intergroup threat, which provides the basis for restricting intergroup boundaries (Pereira, Vala, & Costa‐Lopes, 2010) and leads to exclusion of ambiguous targets (e.g., categorization of mixed-race targets as outgroupers under intergroup threat; Ho, Sidanius, Cuddy, & Banaji, 2013). In this vein, threat might enhance the accessibility of the available negative Roma-Romanian prototype–to which people are exposed by media (see Albarello & Rubini, 2011)–and may lead perceivers to make false positive errors (i.e., adopting a belief that is not true) rather than false negative errors (i.e., failing to adopt a theory that is true; Haselton & Nettle, 2006), very likely for the role of individuals’ defensive motivation to avoid the risks of not recognizing more negative outgroup members. In this vein, this phenomenon is assumed as being limited to negatively perceived outgroups or their negative characteristics. This is in line with previous evidence by Albarello and Rubini (2011) showing that positive traits of Roma (e.g., being musical) were not generalized onto Romanians. Intergroup threat and emotions It has consistently been shown that intergroup threat is an antecedent of outgroup prejudice (for a review, see Rios, Sosa, & Osborn, 2019). A recent trend in intergroup research is to consider the role of emotions as processes through which intergroup threat exacerbates intergroup discrimination (Chang, Krosch, & Cikara, 2016). In this context, two lines of research are relevant. 60
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
Firstly, research has underlined the connection between threat and fear responses, as an aversive emotional state stemming directly from threat (Mobbs et al., 2009; Öhman, 2000). In this vein, threat might elicit fear in perceivers and this perception might, in turn, explain an enhanced tendency to generalize negatively valued minority outgroups onto a less negatively valued outgroup in order to avoid the risks of not recognizing people from the negative minority outgroup. This can be conceived as a defensive strategy that individuals follow under threat. In this vein, it is plausible to propose that fear triggered by intergroup threat mediates the effects of threat on outgroup-to-outgroup generalization. Secondly, we know that emotional appraisals of groups are linked to intergroup cognition and behaviour (Mackie, Devos, & Smith, 2000) and that specific emotions elicited by groups (e.g., admiration, contempt) mediate the effect of predictors of prejudice (groups’ perceived competence or warmth/negative interdependence) on behavioural tendencies towards groups (Cuddy, Fiske, & Glick, 2007). In this vein, Albarello and Rubini (2011) showed that both Roma and Romanians are victims of contemptuous prejudice (Fiske, Cuddy, Glick, & Xu, 2002)—a peculiar pattern of prejudice affecting low status groups. Repulsion emotions such as disgust and contempt are indeed very important in order to tackle the motivational roots of outgroup-to-outgroup generalization. Disgust is a basic emotion characterized by revulsion and withdrawal (Rozin, Haidt, & McCauley, 2000) and also a predictor of negative intergroup attitudes. It is associated with avoiding negative/offensive stimuli (Rozin et al., 2000). Negative emotions such as disgust also instigate defense mechanisms to protect the ingroup in intergroup domains (Hodson & Costello, 2007). In this vein, under intergroup threat, repulsion emotions can work as a further mediator of threat over outgroup-to-outgroup generalization. Negatively valued outgroups Anti-Roma hostility is very high in Europe (European Commission, 2014) and Roma are one of the most discriminated outgroups in all European Countries (FRA, 2018). In Italy,1 for example, many cases of discrimination and aggression toward Roma have been reported (e.g., FRA, 2008) and rejection of Roma is a current political issue on which extreme positions are taken, such that Italian vice-Prime Minister, Matteo Salvini, recently expressed his willingness to compose a list of Roma people and expel them (Povoledo & Pianigiani, 2018). Not only Roma, but also the national group of Romanians (who represent the most numerous foreign population in Italy; ISTAT, 2018) is targeted with heinous prejudice both in Italy (see also Moscatelli, Menegatti, Pratto, & Rubini, 2019) and in Europe (e.g., Ion, 2011). Interestingly, Woodcock (2007) highlighted that a substantial proportion of Romanians attribute the discrimination they suffer to the association of their national group with the Roma minority. This may be due to a confounding between the two groups, given a relative similarity in the category labels defining these two groups (i.e., Roma-Romanians).2 This led the Romanian government to fund a campaign stressing all the positive characteristics of Romanians living in Italy (Nadotti, 2008). Prejudice against Muslims and people from Arab countries is very widespread in Western countries too (FRA, 2017; Ogan, Willnat, Pennington, & Bashir, 2014) and such outgroups are also very negatively perceived (Bruneau, Kteily, & Laustsen, 2018). Importantly, perceived Arab support for terrorist attacks has been shown to predict prejudice against immigrants, in general, and Muslims (Doosje, Zimmermann, Küpper, Zick, & Meertens, 2009). Generalization of terrorists onto such groups also appears in the Western popular culture (e.g., in movies; Malik, 2009) and in public opinion. Overall, this evidence forms the rationale for considering the phenomenon of outgroup-to-outgroup generalization with specific reference to these negatively perceived groups. The present contribution In the light of the literature summarised above, Study 1 examined whether evaluating Roma more negatively than Romanians is a precondition for the occurrence of outgroup-to-outgroup generalization and whether the minority group has to be perceived as a subgroup of the partially-inclusive majority outgroup. Then, Study 1 addressed whether the outgroup-to-outgroup generalization goes from the negatively valued minority group of Roma to the majority group of Romanians, but not in the reverse direction. Moreover, it considered the mediating role of emotional consequences of threat (Chang et al., 2016) in terms of both fear elicited by threat and of repulsion emotions elicited by the groups (Hodson & Costello, 2007; Rozin et al., 2000) on the effects of threat on outgroup-to-outgroup generalization. Study 2 sought to provide further evidence for the enhancement of outgroup-to-outgroup generalization tendency under intergroup threat by considering a second set of outgroups, namely, Islamic terrorists and Arabs. Study 2 employed both an explicit and an implicit linguistic measure of generalization among outgroups based on linguistic abstraction (Semin & Fiedler, 1988; see also Maass, Castelli, & Arcuri, 2000). Finally, it assessed the mediating role of emotions on the outgroup-to-outgroup generalization. Study 1 Study 1 analysed outgroup-to-outgroup generalization processes considering Roma and Romanians. In contrast to Albarello et al.’s (2017) work, this study assessed generalization in terms of perceived prototypicality of the members of one outgroup onto another 1 In Italy, the numerical relation is of 1.6 Roma individuals to 100 Romanians and only a minority of Roma have Romanians origins (i.e., 8.32%; Council of Europe, 2018). It is reported that 6.1% of Italian Roma population live in Emilia-Romagna. This population is settled in 80% of EmiliaRomagna Municipalities (ISTAT, 2018). Romanians represent 17% of the foreign population in Emilia-Romagna and 23% of the foreign population in Italy (ISTAT, 2018); 7.65% of Romanians in Italy live in Emilia-Romagna. 2 In Italian, “Rom” is the noun for the ethnic group, while “Rumeni” refers to the people from Romania.
61
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
one. Before conducting the study, the assumed necessary conditions for outgroup-to-outgroup generalization were considered in two preliminary studies. Given the higher perceived negativity of Roma compared to Romanians and given that intergroup threat enhances the tendency to restrict intergroup boundaries (Ho et al., 2013; Pereira et al., 2010), we expected that under threat Roma would be generalized onto Romanians to a higher extent than in the no-threat condition (Hypothesis 1). Consistent with evidence that negative exemplars shape the overall group impression (Rothbart et al., 1978), we expected that only Roma exemplars would be considered as prototypes of Romanians and not vice versa (Hypothesis 2). Finally, we examined emotional experiences underpinning this phenomenon by taking into account two possible mediators that might both impact the effect of threat. Building on evidence on fear as stemming from threat (Öhman, 2000), we expected that perceived fear would mediate the effect of threat in enhancing the perception of Roma as prototypes of Romanians (Hypothesis 3a). Moreover, drawing from evidence on the link between emotional appraisals of groups and related cognition (Cuddy et al., 2007), and considering that both Roma and Romanians elicit contemptuous/repulsion emotions (Albarello & Rubini, 2011)), we expected that repulsion emotions (Rozin et al., 2000) elicited by groups would mediate the effects of threat on perceived prototypicality of groups (Hypothesis 3b). Method Participants and design One hundred and eighty-six Italian undergraduate students (Mage = 18.88, SDage = 6.15; 70.4% female) participated in the study on a voluntary basis. All participants came from the same north-central region of Italy (Emilia-Romagna). In this area the presence of both Roma and Romanians is quite similar to the national data (cf. Footnote 1). The study used a 3 (threat: no-threat, realistic-threat, symbolic-threat) × 2 (target group: Roma, Romanians) between-participants experimental design.3 Assessing the preconditions of outgroup-to-outgroup generalization Two preliminary studies tackled the preconditions of outgroup-to-outgroup generalization. The first preliminary study considered whether the minority outgroup was evaluated more negatively than the partially-inclusive one. The second preliminary study analysed whether intergroup threat enhanced outgroup-to-outgroup generalization considering Roma and another low status outgroup (i.e., Turks, who are perceived as similar to Roma in terms of physical appearance), that it, is less linked to Roma by a sub/ superinclusion relation, given that very few Roma exemplars have Turkish origins (i.e., 3.78%; Council of Europe, 2018). In the first preliminary study, Italian undergraduate students (N = 40; Mage = 23.20, SDage = 3.36; 57.5% female) evaluated separately Roma and Romanians on a 7-point scale (1 = very negative; 7 = very positive). A paired samples t-test showed that Roma were rated more negatively (M = 1.63, SD = 0.87) than Romanians (M = 2.13, SD = 0.88), t(39)= -4.42, p = .001, η2 = .320. The evaluations of both Roma and Romanians were lower than the mid-point of the scale, respectively t(39)= -17.31, p < .001, d = 2.74; t(39) = -13.44, p < .001, d = 2.13. This evidence suggests that generalization from Roma to Romanians is in line with the assumption that the more negative outgroup is generalized onto less negatively perceived outgroups. The second preliminary study (N = 82 Italian undergraduate, Mage = 20.14, SD = 3.91; 69.5% female) considered whether the negative minority outgroup of Roma could be generalized onto Turks. It also considered whether threat enhanced generalization of Roma onto Turks. After being exposed to the threat manipulations (for details, see Method of Study 1), participants rated the extent to which a “typical Roma is similar to a typical Turk” (1 = not at all; 7= very much). The analysis of variance (ANOVA) (threat: nothreat, realistic-threat, symbolic-threat) on prototypicality scores showed no significant difference (M no-threat = 2.12, SD = 1.93; M 2 realistic-threat = 2.11, SD = 1.71; M symbolic-threat = 2.04, SD = 1.32), F(2, 79) = 0.02, p = .983, η = .000. All pair-wise comparisons (Bonferroni-corrected) were non-significant (ps = 1.00). Overall, such evidence suggests that generalization tendencies might not apply to all outgroups, unless they are perceived as linked to each other by a sub-superinclusion relation (Albarello & Rubini, 2011). Procedure and materials Participants were tested in small groups of 5–7 in a laboratory and informed that the study focused on impression formation about members of national groups and of their subgroups. In order to assure participants’ knowledge of the groups, participants were reminded that Roma are an ethnic group and Romanians are a national group. Participants were exposed to a slide which, in the Romanians condition, reported that the session concerned the group of Romanians (described as people from Romania), some of whom could also be Roma; in the Roma condition, the slide reported that the session concerned the group of Roma (described as a minority ethnic group some of whom could also be Romanians). Realistic and symbolic threat (Stephan & Stephan, 2000) were manipulated by projecting a slide with one of two previously pretested (Albarello & Rubini, 2018) scenarios which referred to: (a) the threat posed by immigrants in terms of unemployment and costs of health and social welfare (realistic threat), or (b) research confirming cultural differences between Italians and immigrants in habits, traditions, ideologies and moral values (symbolic threat). In the no-threat conditions, only the slide describing the groups was presented. 3 Data were collected in two phases. First, all the Roma conditions and the no-threat Romanians conditions were collected. Participants were randomly assigned to one of these four experimental conditions. The remaining two conditions were collected later and participants were randomly assigned to one of these two experimental conditions. The sample collected in phase one (N = 131, Mage = 18.29, SD = 7.10, 64.9% females) was similar to the one collected at phase two (N = 55, Mage = 20.29, SD = 2.25; 83.6% females).
62
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
After the manipulation, participants completed a paper-and-pencil questionnaire. On a 7-point scale (1 = not at all; 7 = very much) participants reported—as a manipulation check—the extent to which they felt threatened by the information provided at the beginning of the research session (i.e., description of the groups and information on immigrants). Using the same 7-point scale they then rated the perceived prototypicality of groups. In the Roma conditions participants rated “the extent to which a typical Roma is similar to a typical Romanian” and “the extent to which a typical Roma is a representative exemplar of Romanians”. In the Romanians condition they evaluated “the extent to which a typical Romanian is similar to a typical Roma” and “the extent to which a typical Romanian is a representative exemplar of Roma”. The scores of these pairs of items were then averaged to yield a more reliable composite score (respectively, r = .888, p < .001, and, r = .758, p < .001). Finally, participants rated on the same 7-point scale the extent to which they experienced “fear” and “concern” after having read the initial information slide(s) and the extent to which Roma (Romanians) elicit “disgust” and “contempt”. The scores of these items were averaged in two mean scores expressing, respectively, perceived fear (r = .883, p < .001) and repulsion emotions elicited by groups (r = .740, p < .001). At the end of the experimental session participants were thanked and debriefed. Results and discussion Unless otherwise indicated, we computed 3 (threat: no-threat, realistic-threat, symbolic-threat) × 2 (target group: Roma, Romanians) ANOVAs; see Table 1 for means and standard deviations. The ANOVA on the manipulation check highlighted a significant effect of threat, F(2, 180) = 74.33, p < .001, η2 = .452. Bonferroni post-hoc comparisons revealed that in the no-threat conditions participants felt threatened to a lesser extent than in the realistic- and symbolic-threat conditions (ps < .001). The realistic- and symbolic-threat conditions did not differ from each another (p = .636). The ANOVA on prototypicality of groups showed a significant effect of threat, F(2, 180) = 4.85, p = .009, η2 = .051 (Mno-threat = 3.67, SD = 1.63; Mrealistic-threat = 4.50, SD = 1.49; Msymbolic-threat = 4.22, SD = 1.70). The ANOVA also showed that this effect was not qualified by the two-way interaction of threat and target group, F (2, 180) = 1.54, p = .462, η2 = .009. However, since we hypothesised that threat would enhance Roma-to-Romanians generalization only (Hypothesis 1), we conducted simple main effects analysis in the Roma condition. Bonferroni post-hoc comparisons showed that in the realistic- and symbolic-threat conditions Roma were rated as prototypical of Romanians to a higher extent than in the no-threat condition (ps ≤ .052). The realistic- and symbolicthreat conditions did not differ (p = 1.00). Importantly, in the Romanians condition, Bonferroni post-hoc comparisons highlighted no significant differences between all threat conditions (ps ≥ .476). Regarding Hypothesis 2, the ANOVA also revealed a significant effect of the target group, F(2, 180) = 52.30, p = < .001, η2 = .225, showing that prototypicality scores were lower in the Romanians than Roma conditions (MRoma = 4.81, SD = 1.53; MRomanians = 3.30, SD = 1.32). As shown by one-sample t-tests, prototypicality of Roma over Romanians was higher than the mid-point of the scale, t(99) = 5.26, p < .001, d = 0.35, whereas that of Romanians for Roma was significantly lower, t(86) = -4.09, p < .001, d = 0.53. Together, these findings suggested that perceived prototypicality of one group over the other went from the negatively valued minority outgroup to the less negative partially-inclusive one. The mean score of perceived fear varied significantly as a function of threat, F(2, 185) = 13.66, p < .001, η2 = .132. Bonferroni post-hoc comparisons revealed that perceived fear was lower in the no-threat than in the realistic- and symbolic-threat conditions (Mno-threat = 3.85, SD = 1.27; Mrealistic-threat = 5.04, SD = 1.39; Msymbolic-threat = 4.76, SD = 1.22), ps < .001. The realistic and symbolic threat conditions did not differ from each another (p = .714). Repulsion emotions elicited by outgroups varied as a function of threat (Mno-threat = 3.35, SD = 1.21; Mrealistic-threat = 5.56, SD = 1.62; Msymbolic-threat = 4.85, SD = 1.36), F(2, 185) = 21.75, p < .001, η2 = .195. Bonferroni post-hoc comparisons revealed that repulsion emotions elicited by groups were higher in the realistic- and symbolic-threat conditions (ps < .001), whereas they did Table 1 Perceived Threat, Prototypicality of Groups, Fear, and Repulsion Emotions as a Function of Threat and Target Group (Study 1). Threat
Target group
No-threat
Roma Romanians
Realistic-threat
Roma Romanians
Symbolic-threat
Roma Romanians
Perceived threat
Prototypicality of groups
Fear
Repulsion emotions
2.49 (1.40) 1.96 (0.80)
4.21 (1.63) 3.06 (1.25)
3.63 (1.21) 4.10 (1.31)
3.67 (1.34) 2.98 (0.94)
4.93 (1.36) 4.76 (1.51
5.22 (1.15) 3.60 (1.36)
5.26 (1.55) 4.76 (1.13)
4.79 (1.19) 4.28 (2.02)
4.58 (1.53) 4.67 (1.30)
5.03 (1.57) 3.30 (1.32)
4.91 (1.48) 4.58 (0.82)
5.15 (1.32) 4.52 (1.35)
Note: Standard deviations are presented in parentheses. 63
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
Table 2 Total, Direct, Indirect, and Omnibus Effects of Predictors of Prototypicality of Roma over Romanians (N = 101) (Study 1). Total effect
Direct effect
Indirect effect 95% CI through fear
Predictor
B
SE
p
B
SE
p
D1 D2 Omnibus effect
1.00 0.82
0.36 0.36
.006 .024 .014
0.36 0.11
0.38 0.38
.343 .776 .601
B
SE 0.29 0.23 0.03
0.20 0.17 0.03
Indirect effect 95% CI through repulsion emotions
LL
UL
B
−0.05 −0.04 −0.01
0.74 0.63 0.10
SE
0.35 0.47 0.06
0.20 0.24 0.04
LL 0.05 0.08 0.01
LL 0.85 1.05 0.16
Note: D1 = no-threat (0), realistic-threat (1), symbolic threat (0); D2 no-threat (0), realistic-threat (0), symbolic-threat (1); B = unstandardized beta weights; SE = standard error; CI = confidence interval; LL = lower limit; UL = upper limit. Table 3 Total, Direct, Indirect, and Omnibus Effects of Predictors of Repulsion Emotions Elicited by Roma (N = 101) (Study 1). Total effect
Direct effect
Indirect effect 95% CI through fear
Predictor
B
SE
p
B
SE
p
D1 D2 Omnibus effect
1.09 1.48
0.32 0.31
.001 .000 .000
0.61 1.08
0.33 0.31
.068 .001 .004
B
SE 0.24 0.19 0.03
0.17 0.14 0.02
LL −0.03 −0.02 −0.00
UL 0.65 0.53 0.08
Indirect effect 95% CI through prototypicality of Roma over Romanians B 0.25 0.20 0.02
SE 0.15 0.15 0.02
LL 0.04 0.01 −0.00
LL 0.65 0.63 0.07
Note: D1 = no-threat (0), realistic-threat (1), symbolic threat (0); D2 no-threat (0), realistic-threat (0), symbolic-threat (1); B = unstandardized beta weights; SE = standard error; CI = confidence interval; LL = lower limit; UL = upper limit.
not differ in the latter two conditions (p = .752). The analysis also highlighted a significant effect of the target group, showing that Roma elicited repulsion emotions to a higher extent than Romanians (MRoma = 4.52, SD = 1.43; MRomanians = 3.90, SD = 1.60), F(1, 185) = 9.09, p = .003, η2 = .048. Mediational analyses To test the mediational hypotheses (Hypotheses 3a and 3b), we focused on the Roma conditions (N = 101) only. Bootstrapping multiple mediational analysis (5000 re-samples) was conducted using the method described by Hayes and Preacher (2014) for mediational models employing multicategorical independent variable. The independent variable was recoded as two dummy variables: D1 contrasted the no-threat (coded 0), realistic-threat (coded 1) and symbolic threat (coded 0) conditions; D2 compared the nothreat (coded 0), realistic threat (coded 0), symbolic-threat (coded 1) condition. D1 and D2 were entered simultaneously as predictors in the regression model. The PROCESS 2.15 macro for SPSS (model 4) was used, since it produces omnibus tests of total, direct, and indirect effects indicating whether there is an overall effect of the two dummy variables on the outcome variable besides their specific effects. This analysis (for detailed results, see Table 2) showed that D1 and D2 significantly predicted fear (see also Fig. 1). The omnibus test of total effects of D1 and D2 was significant, F(2, 98) = 4.47, p = .014, R2 = 0.08. The relative total effects of D1 and D2 were significant. When the mediators were included in the model, the omnibus test of direct effects of D1 and D2 was not significant, F(2, 96) = 0.51, p = .601, R2 = 0.01, and the relative direct effects of D1 and D2 on the dependent variable became non-significant. Contrary to expectations (Hypothesis 3a), the omnibus test of indirect effects through fear was not significant, with the 95% Confidence Interval (CI) including zero. The specific relative indirect effects of D1 and D2 through fear were non-significant. Instead, as expected (Hypothesis 3b), the omnibus indirect effect through the emotions elicited by Roma was significant. The specific relative indirect effects of D1 and D2 through emotions were significant, thus showing that repulsion emotions elicited by Roma mediated the effects of threat.4 This study investigated outgroup-to-outgroup generalization considering Roma and Romanians. Overall, preliminary studies showed that Roma were rated as a more negative outgroup than Romanians, and that Roma were not generalized onto Turks. Then, Study 1 showed that threat, overall, enhanced the outgroup-to-outgroup generalization effect. Importantly, simple main effects analysis conducted to test Hypothesis 1 revealed that under threat, perceived prototypicality of highly negative (out)group minority members is extended to members of the partially-inclusive superordinate group. Adding to the evidence by Albarello et al. (2017), the study also investigated the direction of the phenomenon and showed that Roma (minority) were viewed as prototypical of Romanians 4 An alternative mediational model was tested entering prototypicality of Roma over Romanians and fear as the mediators of threat’s effect on emotions elicited by Roma (Table 3). The omnibus test of direct effects of D1 and D2 was still significant after including mediators in the model, F(2, 96) = 5.95, p = .004, R2 = 0.09, whereas the omnibus indirect effect of D1 and D2 through the mediators was not significant (with 95% CIs including zero). This mediational model was less convincing, as revealed by larger CIs for the predicted model and by the higher effect size of the direct effect of the independent variables in the alternative model.
64
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
Fig. 1. Multiple mediation model of the effect of threat (D1 and D2) on protopypicality of Roma over Romanians via fear elicited by threat and repulsion emotions elicited by Roma (Study1). Note: *p < .05. ** p < .01. *** p < .001. D1 = no-threat (0), realistic-threat (1), symbolic threat (0); D2 no-threat (0), realistic-threat (0), symbolicthreat (1); Coefficients are unstandardized regression weights.
(partially-inclusive superordinate group) to a greater extent than Romanians were of Roma, suggesting that outgroup-to-outgroup generalization goes from the more negative minority onto the less negative partially-inclusive superordinate outgroup and thus it does not simply depend on threat, but it is related to the groups’ characteristics (i.e., higher negativity of the minority group from which judgments are projected). Importantly, perceived prototypicality of Romanians over Roma was not affected by threat. Additionally, this study increased our understanding of social psychological mechanisms underlying outgroup-to-outgroup generalization. Although, unexpectedly, fear due to threat did not mediate the effect of threat on outgroup-to-outgroup generalization, repulsion emotions elicited by Roma acted as the mediator on perceived prototypicality of Roma over Romanians. Since the literature suggests that such emotions are related to avoidance and defensive motives (Rozin et al., 2000), this evidence improves our understanding of the effects of threat in restricting intergroup boundaries (Ho et al., 2013) and rejecting potential negative outgroupers. Study 2 Study 2 aimed to test whether the outgroup-to-outgroup generalization effect is a general phenomenon, and not restricted to the case of Roma and Romanians. In order to maintain a partial sub/superinclusion relation as a necessary condition for the outgroup-tooutgroup generalization tendency to occur (as shown in Study 1), Arabs (described as the people from the 22 member states of the Arab League5) and Islamic terrorists were used as groups. Real life events show that only a small minority (subgroup) of Arabs are terrorists, and not all Islamic terrorists are from member states of the Arab League (e.g., Islamic terrorists from Chechnya). These two groups were chosen not only because they are linked through the sub/superinclusion relation criterion which—as shown in Study 1—is a necessary condition for the occurrence of the outgroup-to-outgroup generalization, but also because they are loaded with a very negative evaluation. As shown in Study 1, negative evaluation is another precondition of outgroup-to-outgroup generalization; and this negative evaluation is even more pronounced for Islamic terrorists than Roma. The more negative evaluation of Islamic terrorists than of Arabs was tested in a pretest (see Method section below). Due to absence of differences between the threat conditions in Study 1, in this study we focused on no-threat versus symbolic-threat, in view of the emphasis given by the current Italian government on the different values endorsed by immigrants (Squires, 2018). With the aim of testing whether outgroup-to-outgroup generalization occurs even at a less blatant level, we used a linguistic measure relying on variations in linguistic abstraction, given that Maass et al. (2000) contended that linguistic devices form implicit means of stereotypes transmission and persistence. We used verbs and adjective following the Linguistic Category Model (LCM; Semin & Fiedler, 1988, 1991) which distinguishes among linguistic terms used in interpersonal and intergroup domains (Maass, Salvi, Arcuri, & Semin, 1989; Maass, 1999; see also Albarello & Rubini, 2015; Moscatelli, Albarello, Prati, & Rubini, 2014; Moscatelli, Albarello, & Rubini, 2008; Rubini, Menegatti, & Moscatelli, 2014; Rubini, Moscatelli, Albarello, & Palmonari, 2007) relying on the fact that abstract terms such as adjectives (e.g., Mark is aggressive) imply higher generalizability across different situations and time of the characteristics associated with the described person, whereas the use of verbs (e.g., descriptive action verbs, such as Mark hits a person) implies that the described characteristics are less generalizable and more context-dependent (Semin & Fiedler, 1988). Besides replicating evidence of Study 1 on the group prototypicality measure, given that adjectives imply higher generalizability 5
The 22 member states of the Arab League include Morocco, Jordan, Algeria, Syria, Oman, Tunisia, United Arab Emirates, Qatar, Egypt, etc. 65
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
than verbs (Semin & Fiedler, 1988), we expected that threat would lead participants to agree more strongly with sentences employing adjectives (e.g., “group X and group Y are identical”) to describe the relation between the two groups than with sentences employing verbs (e.g., “group X and group Y overlap one another” (Hypothesis 1). In line with evidence of Study 1, we also expected that repulsion emotions elicited by Islamic terrorists would mediate the effects of threat on the measures of outgroup-to-outgroup generalization (Hypothesis 2). Method Pretesting evaluation of target outgroups Perceived negativity of Islamic terrorists and of Arabs was pretested (N = 45 Italian undergraduate; Mage = 20.67; SDage = 4.20; 68.1% female). Participants were informed that the research focused on impression formation about members of social groups and that the current section regarded the group of Arabs (defined as people from member states of the Arab League) or of Islamic terrorists (defined as members of political groups that make use of terrorist methods in order to promote their political/religious ideas; adapted from Treccani, 2017). Perception of Arabs and Islamic terrorists was rated on a single item from 1 (very negative) to 6 (very positive). An independent samples t-test showed that Islamic terrorist were rated more negatively than Arabs (MIslamic terror2 ists = 1.33, SD = 0.48, MArabs = 3.80, SD = 0.83), t(45) = 12.80, p < .001, η = .785. Participants and design Ninety undergraduate Italian university students (Mage = 19.64, SDage = 3.13; 63.3% females) participated in the study on a voluntary basis. The study employed a 2 (threat: no-threat, symbolic threat) × 2 (target group: Islamic terrorists, Arabs) betweenparticipants experimental design. Participants were randomly assigned to one of the four experimental conditions. Procedure and materials The procedure was the same as for the previous studies. Participants were told that the research focused on impression formation concerning members of social groups and that the current section regarded mainly the group of Arabs (in the condition wherein Arabs were the main target group to consider) or of Islamic terrorists (in the condition in which Islamic terrorists were the main target group to consider). The no-threat and symbolic-threat conditions were manipulated as in Study 1. Manipulation check and dependent variables In order to test the effectiveness of the threat manipulation, as in Study 1, participants rated the extent to which they felt threatened by the information provided at the beginning of the research session (i.e., description of the groups and information on immigrants). Group prototypicality and repulsion emotions were measured as in Study 1. In addition, we developed a measure of linguistic generalization of Islamic Terrorists onto Arabs based on the LCM (Semin & Fiedler, 1988). Participants rated the extent to which they agreed (1 = completely disagree; 7 = completely agree) with four sentences describing the relation between the two groups in which the semantic meaning was systematically maintained but abstractness of linguistic terms varied. Two of these sentences were formulated with verbs (e.g., “Arabs and Islamic terrorists overlap one another”; “Islamic terrorists and Arabs match one another”), the other two were worded with adjectives (e.g., “Islamic terrorists and Arabs are identical”; “Arabs and Islamic terrorists are indistinguishable”).6 Since people are usually unaware of the concreteness/abstractedness of the terms they use to describe others and of the cognitive implications of these terms with regard to generalization of the described features (Semin & Fiedler, 1988; cf. Fino, Menegatti, Avenanti, & Rubini, 2016), this provides a less blatant measure of generalization among groups. Two mean scores were computed averaging the items containing verbs (r = .537, p < .001) and those containing adjectives (r = .836, p < .001). These two scores were significantly positively correlated (r = .529, p < .001). Results and discussion Means and standard deviations for all the study variables are reported in Table 4. As for perceived threat, a 2 (threat: no-threat, symbolic-threat) × 2 (target group: Islamic terrorists, Arabs) ANOVA revealed only a significant main effect of threat showing that participants reported higher levels of perceived threat in the symbolic- than in the no-threat condition (Mno-threat = 2.05, SD = 1.08; Msymbolic-threat = 4.10, SD = 1.99), F(1, 86) = 25.97, p < .001, η2 = .232. A 2 (threat) × 2 (target group) ANOVA on perceived group prototypicality highlighted significant effects of threat (Mno2 threat = 2.15, SD = 0.85; Msymbolic-threat = 3.29, SD = 2.04), F(1, 86) = 11.88, p = .001, η = .121, revealing higher perceived prototypicality under threat than no-threat, and of target group (MIslamic terrorists = 3.39, SD = 2.05; MArabs = 2.18, SD = 0.88), F(1, 6 The two verbs (i.e., “overlap”; “match”) and the two adjectives (i.e., “identical”; “undistinguishable”) employed in the four sentences comprised in the ad hoc created measure of linguistic generalization among different outgroupers were pretested. A sample of undergraduate Italians (N = 25, Mage = 20.48, SDage = 2.55; 79.2% female) rated on a series of semantic differentials the extent to which each term described the overlap between two fictional groups (i.e., group X, group Y) in concrete (1) versus abstract (7) terms. Means of ratings referring to the two verbs (r = .644, p = .001) and to the two adjectives (r = .858, p < .001) were averaged and then compared through a paired-sample t-test. The sentences containing verbs were rated as describing the relations between the two groups in more concrete terms than sentences using adjectives (Mverbs = 1.16, SD = 0.43; Madjectives = 5.78, SD = 1.51), t(24) = -4.62, p < .001, η2 = .471.
66
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
Table 4 Perceived Threat, Prototypicality of Groups, Linguistic Generalization Verbs and Adjectives) and Repulsion Emotions as a Function of Threat and Target Group (Study 2). Threat
Target group
No-threat
Islamic terrorists
Perceived threat
Prototypicality of groups
Linguistic generalization (verbs)
Linguistic generalization (adjectives)
Repulsion emotions
1.94 (1.00) 2.14 (1.67)
2.11 (0.83) 2.18 (0.88)
1.67 (0.87) 1.43 (0.60)
1.28 (0.55) 1.55 (0.51)
3.94 (1.92) 1.50 (0.65)
4.22 (1.97) 3.96 (2.05)
4.24 (2.19) 2.17 (1.10)
2.26 (1.25) 1.48 (0.68)
2.74 (2.33) 1.46 (0.60)
5.13 (2.00) 1.76 (1.73)
Arabs Symbolic-threat
Arabs Islamic terrorists
86) = 10.51, p = .002, η2 = .109, showing that Islamic terrorists were perceived as prototypical of Arabs to a higher extent than Arabs were perceived as prototypical of Islamic terrorists. The two-way interaction was significant, F(1, 86) = 12.06, p = .001, η2 = .123, and post-hoc comparisons (Bonferroni) highlighted that threat affected the extent to which Islamic terrorists were perceived as prototypical of Arabs (p < .001), but did not affect the extent to which Arabs were perceived as prototypical of Islamic terrorists (p = .985). In the symbolic threat-condition, Islamic terrorists were perceived as prototypical of Arabs to a higher extent than Arabs were perceived as prototypical of Islamic terrorists (p < .001). No difference emerged in the no-threat condition (p = .878). A 2 (threat) × 2 (target group) × 2 (type of terms: verbs, adjectives) mixed-model ANOVA with repeated measures on the last factor was conducted on the measures of linguistic generalization as the most parsimonious analytical approach. As for withinparticipant effects, the Threat × Target group × Type of terms interaction approached conventional levels of statistical significance, F(1, 86) = 3.65, p = .059, η2 = .041. In line with evidence of Study 1, pairwise comparisons (Bonferroni) revealed a significant effect of threat in the Islamic terrorists condition both on verbs and on adjectives, respectively, p = .036, and p = .001. That is, under symbolic-threat participants showed stronger agreement with sentences stating that Islamic terrorists and Arabs overlap one another than in the no-threat condition. Even though means of agreement were below the mid-point of the scale, threat led participants to perceive higher overlap between the members of the two groups. Most importantly, pairwise comparisons revealed that in the symbolic-threat condition, when Islamic terrorists were the target group, participants rated adjectives as more suitable than verbs to describe the relation between Islamic terrorists and Arabs (p = .046) (Hypothesis 1). No significant difference was found in either the no-threat Islamic terrorists condition (p = .185), the no-threat Arabs condition (p = .667), or the symbolic-threat Arabs condition (p = .933). In the symbolic threat-conditions, when the research was presented as focused mainly on Islamic terrorists, participants agreed to a higher extent with sentences describing the relation between Islamic terrorists and Arabs both with verbs (p = .003) and with adjectives (p = .001) than when the research was presented as regarding Arabs. In other words, participants who were told that the research concerned Islamic terrorists tended to perceive an overlap between Islamic terrorists and Arabs to a higher extent than participants who were told that they would focus on Arabs during the research session. Overall, these data suggested that under threat conditions people were more prone to generalize members of the negatively perceived outgroup of Islamic terrorists onto the superordinate partially-inclusive outgroup of Arabs. Finally, the aforementioned mixed-model ANOVA highlighted significant between participants effects of threat (Mno-threat = 1.48, SE = 0.16; Msymbolic-threat = 1.98, SE = 0.14), F(1, 86) = 5.76, p = .019, η2 = .063, and of target group (MIslamic terrorists = 1.99, SE = 0.15; MArabs = 1.48, SE = 0.15), F(1, 86) = 5.87, p = .017, η2 = .064, and a significant two-way interaction between Threat × Target group, F(1, 86) = 6.26, p = .014, η2 = .068. These effects are not discussed here since they are of less theoretical relevance as they do not allow to distinguish between verbs and adjectives.7 As in Study 1, repulsion emotions significantly varied as a function of threat, F(1, 86) = 4.09, p = .046, η2 = .045. That is, participants expressed higher repulsion emotions to the outgroups in the symbolic- than in the no-threat condition (Mno-threat = 2.06, SD = 1.83; Msymbolic-threat = 3.58, SD = 2.52), and to Islamic terrorists than to Arabs (MIslamic terrorists = 5.13, SD = 2.01; MArabs = 1.62, SD = 1.31), F(1,86) = 66.11, p < .001, η2 = .435. As in Study 1, the two-way interaction was not significant, F(1, 86) = 1.67, p = .200, η2 = .019. Mediational analyses To test the mediational hypothesis that repulsion emotions elicited by Islamic terrorists mediate the effects of threat on generalization among Islamic terrorists and Arabs (Hypotheses 2) bootstrapping multiple mediational analyses (5000 re-samples) were conducted in the Islamic terrorists conditions (N = 45) using the method described by Preacher and Hayes (2004) for simple mediational models. The PROCESS 2.15 macro for SPSS (model 4) was used. Mediational analysis on group prototypicality (for detailed results, see Fig. 2) showed that the effect of threat on the mediator 7
Data available from first author. 67
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
Fig. 2. Mediational model (N = 45) of the effect of threat on prototypicality of Islamic terrorists over Arabs via repulsion emotions elicited by Islamic terrorists (Study 2). Note: *p < .05. **p < .01. #p = .055. Coefficients are standardized regression weights.
(i.e., repulsion emotions) approached the conventional level of statistical significance. The total effect of threat on the outcome variable was significant. When the mediator was included in the model, it significantly predicted prototypicality of Islamic terrorists over Arabs. The direct effect of threat was still significant. The indirect effect of threat through the mediator was significant, 95% CI [0.01, 0.47]. Importantly the effect size of the mediational model was significant, R2 = 0.09, 95% CI [0.01, 0.24].8 As for the linguistic measure of generalization based on verbs, the mediational analysis (SPSS macro, model 4) showed (see Fig. 3) that the total effect of threat on the outcome variable was not significant. The direct effect of threat was not significant when the mediator was included in the model. The indirect effect of threat via the mediator was also not significant [-0.00, 0.23].9 The mediational analysis on generalization via adjectives (see Fig. 4) showed that the total effect of threat on the outcome variable was significant. After inclusion in the model, the mediator predicted the outcome variable. The direct effect of threat on the outcome variable was not significant. Importantly, the indirect effect of threat through the mediator was significant, 95% CI [0.03, 0.45]. The effect size of the mediational model was significant, R2 = 0.06, 95% CI [0.01, 0.18].10 Overall, these results provided convergent evidence on the role of repulsion emotions in explaining the effect of threat on generalization of Islamic terrorists onto Arabs relying both on a blatant and less blatant measure of generalization. Study 2 aimed to provide a conceptual replication of the effects of intergroup threat in enhancing outgroup-to-outgroup generalization by considering another pair of outgroups and employing an explicit and a less controllable/blatant measure of generalization relying on linguistic abstraction. In line with evidence of Study 1, findings on the explicit generalization measure confirmed that threat enhanced generalization of Islamic terrorists onto Arabs, and that Arabs were not generalized onto Islamic terrorists. As expected, results also showed that people displayed higher linguistic generalization (Semin & Fiedler, 1988) among the two groups when subjected to intergroup threat than no-threat by considering sentences with adjectives rather than verbs as more suitable to describe the relation between Islamic terrorists and Arabs. Thus, this study provided converging evidence with Study 1, using also a less blatant measure of generalization. Importantly, the indexes of linguistic generalization through both verbs and adjectives were also positively correlated with prototypicality of groups, suggesting that such measures were both tapping generalization processes. This study also provided further evidence of the mediational role of repulsion emotions elicited by Islamic terrorists on the effects of threat on both the blatant measure of generalization and on the less blatant, linguistic measure of generalization (based on use of adjectives). In this latter case, even though the mediational evidence is not as strong as that of Study 1, the findings obtained are impressive given that implicit measures are rarely correlated with explicit measures (cf. Albarello, Crisp, & Rubini, 2018; Maass et al., 2000). Thus, this evidence underlines the strong power of adjectives in expressing generalization among groups. General discussion The present contribution aimed to deepen knowledge on outgroup-to-outgroup generalization tendencies by shedding light on the antecedents and processes through which individuals reach the conclusion that some outgroups “are all alike”. First of all, in a preliminary study, we showed that the outgroup that is generalized onto another group was evaluated more negatively than the other group. That is, both Roma (Study 1) and Islamic terrorists (Study 2) were rated more negatively than Romanians and Arabs, respectively. Thus, an enhanced negative value associated with the smaller group can be considered as an antecedent or necessary condition for outgroup-to-outgroup generalization to occur. The second necessary condition highlighted here is the sub/superinclusion relation between the two groups, such that the negatively-valued minority outgroup is generalized to the superordinate inclusive one. In this respect, a preliminary study revealed that 8 The alternative mediational model conducted entering prototypicality of Islamic terrorists over Arabs as the mediator of the indirect effects of threat on moral emotions through the mediator was significant, as shown by the 95% CI [0.07, 0.84]. However, the effect size of this mediational model was not significant, R2 = 0.07, 95% CI [-0.01, 0.25]. 9 The alternative mediational model considering linguistic generalization (verbs) as the mediator showed that the indirect effect of threat through the mediator was not significant, 95 % CI [-0.00, 0.18]. 10 An alternative mediational model was tested entering linguistic generalization through adjectives as the mediator of threat’s effect on moral emotions elicited by Islamic terrorists. The indirect effect of threat through the mediator was significant, 95% CI [0.07, 0.84]. However, the effect size of this mediational model was not significant, R2 = 0.06, 95% CI [-0.00, 0.17], thus showing that the alternative mediational model was less robust than the hypothesized one.
68
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
Fig. 3. Mediational model (N = 45) of the effect of threat on linguistic generalization (verbs) among Islamic terrorist and Arabs via repulsion emotions elicited by Islamic terrorists (Study 2). Note: *p < .05. #p = .055. §p = .088. Threat: No-threat = 0; Symbolic-threat = 1. Coefficients are unstandardized regression weights.
Fig. 4. Mediational model (N = 45) of the effect of threat on linguistic generalization (adjectives) among Islamic terrorist and Arabs via repulsion emotions elicited by Islamic terrorists (Study 2). Note: *p < .05. #p = .055. Threat: No-threat = 0; Symbolic-threat = 1. Coefficients are unstandardized regression weights.
Roma were not generalized onto Turks, who share a similar physical stereotype with Roma, but are less clearly recognizable as a superinclusive outgroup for Roma, than Romanians are (indeed, the percentage of Turkish Roma is lower than that of Romanian Roma, respectively 3.78% and 8.32%; Council of Europe, 2018). Sub-superinclusion as a necessary condition for outgroup-to-outgroup generalization was observed also in Study 1 with Roma and in Study 2 with Islamic terrorists generalized onto the larger, partially-overlapping superordinate outgroups, respectively of Romanians and Arabs. This was also further corroborated by evidence of Study 1 and Study 2 highlighting that generalization is not a bidirectional phenomenon between groups given that Romanians were not perceived as prototypes of Roma, and Arabs were not perceived as prototypes of Islamic terrorists. This result adds to previous evidence of individual-to-group generalization obtained by Rothbart et al. (1978), considering a group-to-group level and showing that direction of generalization is from the negatively-valued minority to the superordinate outgroup and not vice-versa. Moreover, besides its main effect observed in both studies, intergroup threat enhanced the outgroup-to-outgroup generalization effect only in the expected direction, that is, from the negative minorities (i.e., Roma and Islamic terrorists) to the less negative partiallyinclusive superordinate outgroups (i.e., Romanians and Arabs). Intergroup threat had a consistent significant effect, both on an explicit measure of prototypicality and on a less blatant linguistic measure of generalization. This finding is particularly noteworthy since, as argued by Maass et al. (2000), people are not supposed to be aware of the different implications in terms of generalizability of the characteristics attributed to targets by verbs or adjectives (which imply duration and invariance across time and contexts). Thus, the use of sentences formulated with terms varying in abstractness allowed us to detect the effect of intergroup threat in leading people to generalize a negatively-valued outgroup onto another less negative outgroup. In this respect, it is important to stress that the method we developed, being based on agreement with sentences formulated at different levels of abstraction (Semin & Fiedler, 1988; see also Maass et al., 1989, Study 2), differed from classical work on language abstraction based on participants’ free use of terms. In this latter case, scholars sometimes computed differential abstraction scores (positive minus negative) to summarise the overall level of discrimination expressed towards a target (e.g., Albarello & Rubini, 2018) after having provided weights to the different terms depending on their classification according to Semin and Fiedler’s model. We believe that a difference score would not suit the aim of our study, however, since we did not address positive versus negative statements, but were interested in understanding whether agreement with statements varied across experimental conditions, while keeping constant the meaning of the sentence (i.e., overlap of the two groups). In this vein, the repeated measures ANOVA performed on the linguistic measures seems a suitable analytical tool to compare agreement with verbs and adjective sentences across conditions. Overall, findings confirm that linguistic abstraction can be a very sensitive tool to assess not only biased evaluations of others (e.g., Albarello & Rubini, 2015; Rubini et al., 2014), but also rather unexplored phenomena such as outgroup-to-outgroup generalization processes and the role played by threat in such generalization phenomena. Of course, other methods could be employed in future research, such as the implicit association test (Greenwald, McGhee, & Schwartz, 1998). Emotional processes underlying outgroup-to-outgroup generalization The present studies went beyond previous evidence by exploring the process underlying the effects of threat in leading to outgroup-to-group generalization by showing that repulsion emotions (i.e., disgust/contempt) elicited by the minority outgroup acted as the mediating variable on the generalization of Roma over Romanians and of Islamic terrorists over Arabs. In this vein, these studies provided further evidence of the interplay between cognition (here generalization outcomes) and emotions triggered by groups in 69
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
shaping intergroup relations (Brewer & Alexander, 2002; Cottrell & Neuberg, 2005). Such results suggest that these negative emotions elicited by outgroups may lead the perceiver to generalize among particular groups and enhance confounding between the groups. In this vein, our findings are in line with evidence that negative intergroup emotions such as disgust/repulsion instigate defence mechanisms and restriction of intergroup boundaries (Hodson & Costello, 2007; Rozin et al., 2000), thus underlying the defensive motivation driving outgroup-to-outgroup enhanced generalization under intergroup threat (Haselton & Nettle, 2006). It should be emphasized that evidence of the mediational role of repulsion emotions was detected on the explicit measure of generalization in both studies employing two different pairs of outgroups. Even more interestingly, repulsion emotions also mediated the effect of threat on the less blatant measure of linguistic generalization relying on adjectives. The fact that the mediational role of repulsion emotions was not replicated on verbs should not be conceived as problematic, since these terms provide generalization outcomes only to a very limited extent, as predicted by the LCM (Semin & Fiedler, 1988). Specificity of outgroup-to-outgroup generalization processes Overall, this contribution helped to specify the features and appearance conditions of the outgroup-to-outgroup generalization tendency, which can be also defined as outgroup projection bias (Albarello & Rubini, 2011). This is a peculiar phenomenon which cannot simply be encompassed under the label “outgroup homogeneity” (given, for instance, that it does not apply to positive characteristics and requires a more negative evaluation of the minority outgroup than the superordinate one and a sub/superinclusion relation between the groups) nor should it be confused with the “ingroup projection” tendency highlighted by Mummendey and Wenzel (1999). First of all, as underlined above, not all outgroups are generalized onto another group (as shown, for instance, by the fact that Romanians are not perceived as prototypes of Roma not even under threat, and that Roma are not generalized onto Turks). Second, the ingroup projection (IP) effect (Mummendey & Wenzel, 1999) and the outgroup projection bias (OP) rely on different core features and processes. Indeed, the IP involves an ingroup member’s perspective and the fact that one’s ingroup is projected onto the superinclusive category encompassing both the ingroup and the outgroup such that the ingroup is perceived as possessing the normative characteristics of the superinclusive common category to a higher extent than the outgroup. In contrast, the OP bias involves only outgroups. It consists of generalization of a negative outgroup prototype (i.e., ideal type members of a category that best represent its identity; Oakes, Haslam, & Turner, 1998) onto a larger partially-inclusive outgroup category. A core feature of this process refers to the fact that it relies on the perceiver’s view, that is, someone who is not a member of the projected or the superinclusive group. It also relies on assimilation of the superinclusive partially overlapping outgroup to the negative characteristics of a negatively perceived subgroup. Moreover, it is due to a defensive motive (protecting the ingroup from potential damages from negative outgroups; cf. Haselton & Nettle, 2006) rather than highlighting the higher normative adherence of the ingroup to the superinclusive group, thus leading to ingroup positive distinctiveness in contrast to the outgroup as the IP does. In this vein, the OP shares some similarities with the ingroup overexclusion effect (IOE; Leyens & Yzerbyt, 1992), which is observed when group members are motivated to protect their ingroup from contamination by negatively-valued outgroupers (Rubin & Paolini, 2014). Similarly, the OP would imply overexclusion (Albarello & Rubini, 2011) and discrimination of the larger inclusive outgroupers as if they belonged to the negatively-valued subgroup. Such an effect might have the same motivational roots as the IOE. In other words, the OP might likely be due to the need to avoid the risks of not recognizing negatively-valued outgroupers, considering them as belonging to a less negatively-valued outgroup. The mediational role of repulsion emotions—which are related to avoidance and protecting (ingroup) motives (Rozin et al., 2000)—highlighted in these two studies seems indeed to confirm such an interpretation. Future work is nonetheless needed to provide stronger evidence on this claim. Limitations and future research This contribution addressed boundary conditions as well as emotional foundations of outgroup-to-outgroup generalization processes with reference to two highly stigmatized minority outgroups, Roma and Islamic terrorists, using samples of Italian undergraduate participants. These might be interpreted as leading to some limitations in generalization of findings. First, it might be argued that undergraduate students are not representative of the entire population. However, the evidence that threat enhances this tendency in these samples might suggest that it is a deeply rooted bias, which is not affected by actual knowledge of the groups at stake. This sample is also better educated than the overall population, and hence more likely to give socially desirable responses. We might therefore expect to find even stronger evidence with less-educated participants, who might be less aware, less motivated, or less able to control socially undesirability bias against outgroupers. As for the method we employed, it should be noted, as a further limitation, that data collection of Study 1 took place at two different times, in order to complete the design by adding the realistic and symbolic threat conditions to the Romanians condition. This means that participants could not be allocated at random to all the conditions, but we checked and noted the similarity of the sample that was collected at a later time to the sample collected earlier (see Footnote 3). Another issue that might be considered as a limitation is related to the Italian sample. However, Roma and Muslims are two very stigmatized outgroups in Italy (FRA, 2018; PEW Research Centre, 2017). Currently, in Italy Romanians and Moroccans (who belong to the Arab League) are the most numerous foreign population, representing respectively 23.0% and 8.12% of the foreigner population in Italy (ISTAT, 2018). Considering these data, we believe that our contribution captures a very relevant phenomenon in the realm of intergroup relations with salient negatively perceived outgroups. Future studies should consider whether this generalization tendency varies depending on differences in the presence of the involved outgroups in respondents’ neighborhood and on actual contact (Allport, 1954) with them. 70
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
There is no reason to assume that outgroup-to-outgroup generalization will not apply to other groups (assuming the necessary preconditions) and that it will not be found using other samples (assuming the negative evaluation of the groups and their threatinducing characteristic for the sample). However, future studies should also employ cross-cultural samples in order to test the generalizability of the phenomenon. Since both Roma and Muslims are very salient, negatively stigmatized outgroups in Europe (FRA, 2017, 2018), but also in other nations (e.g., North America; Malik, 2009; Varjú & Plaut, 2017), one first step could be a crosscultural comparison using these two groups as targets. Another issue that could be raised concerns the exclusive attention on repulsion emotions in the present work. The reason for this focus lies in the fact that Roma and Romanians are targets of contemptuous prejudice (Albarello & Rubini, 2011; see also Albarello & Rubini, 2008, 2012; Rubini et al., 2017). We are aware that this addresses only a partial picture of the role of intergroup emotions in intergroup context. For instance, other contributions (e.g., Cottrell, Richards, & Nichols, 2010) suggested that different outgroups are perceived as threatening and elicit different types of threat that can be associated with different emotions, such as fear (towards outgroups that are threatening for the ingroup’s safety) or disgust (towards outgroups that are perceived as threats to the ingroup’s values). Future studies, especially if focusing on very negatively perceived outgroups such as Islamic terrorists, should also consider a larger pattern of emotional reactions to groups (e.g., fear due to groups) in order to provide a more nuanced picture of the emotional roots of outgroup-to-outgroup generalization. With respect to the emotional underpinning of the effect of threat on outgroup-to-outgroup generalization, it should be noted that we did not find support for the mediational role of fear due to threat as leading to activation of defensive reactions of the organism (Öhman, 2000). This could have been due to the fact that we manipulated threat through a scenario, whereas the neurophysiological evidence we relied on to formulate our expectation was obtained in more controlled settings and with neurophysiological and objective indexes of fear activation in the brain. Future studies should integrate such measures of elicited fear in order to analyse the role of fear due to threat on outgroup-to-outgroup generalization. Besides these issues, future studies should also address more thoroughly the necessary conditions for outgroup-to-outgroup generalization, for instance directly testing the extent to which intergroup threat enhances salience of intergroup differentiation and of negatively perceived outgroups. Conclusions This contribution aimed at addressing the outgroup-to-outgroup generalization tendency, highlighting the necessary conditions of its appearance and exploring variables that mediate the effect of threat on this bias. We were able to show that the generalization tendency targeting very negatively perceived group exemplars is not only an intra-group effect (Rothbart et al., 1978), but can sometimes be extended to outgroup-to-outgroup generalization ending in an outgroup projection phenomenon. Further research is needed to better understand why people project very negatively-stereotyped outgroups onto other ones in order to develop interventions aimed at hindering these tendencies (e.g., Albarello et al., 2018; Albarello, Crocetti, & Rubini, 2019). Funding The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research reported in this article was supported by funds (Reg. 14 Prot. 139) from the Department of Psychology University of Bologna to the first author. Declaration of Competing Interest The authors declare that there are no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. References Albarello, F., & Rubini, M. (2008). Relazioni intergruppi e fenomeni di deumanizzazione [Intergroup relations and dehumanization phenomena]. Psicologia Sociale, 3, 67–94. https://doi.org/10.1482/2760. Albarello, F., & Rubini, M. (2012). Se sei una persona umana allora mi piaci di più. Gli effetti della ridotta umanità degli altri sulla percezione sociale [If you are human, I like you more: The effects of others reduced humanity on social perception]. Psicologia Sociale, 7, 443–452. https://doi.org/10.1482/38444. Albarello, F., Crisp, R. J., & Rubini, M. (2018). Promoting beliefs in the inalienability of human rights by attributing uniquely human emotions. The Journal of Social Psychology, 158, 309–321. https://doi.org/10.1080/00224545.2017.1346581. Albarello, F., Foroni, F., Hewstone, M., & Rubini, M. (2017). Generalization of Roma onto Romanians: Evidence of the outgroup projection effect. Psicologia Sociale, 11, 239–249. https://doi.org/10.1482/87249. Albarello, F., & Rubini, M. (2011). Outgroup projection: Il caso degli stereotipi negativi attribuiti a Rom e Rumeni [Outgroup projection: The case of negative stereotypes attributed to Roma and Romanians]. Psicologia Sociale, 6, 355–365. https://doi.org/10.1482/35788. Albarello, F., & Rubini, M. (2015). The role of reduced humanity in producing linguistic discrimination. Personality & Social Psychology Bulletin, 41, 224–236. https:// doi.org/10.1177/0146167214561195. Albarello, F., & Rubini, M. (2018). Linguistic discrimination towards Roma: Can intergroup threat enhance bias? Journal of Language and Social Psychology, 37, 350–364. https://doi.org/10.1177/0261927X17725880. Albarello, F., Crocetti, E., & Rubini, M. (2019). Prejudice and inclusiveness in adolescence: The role of multiple categorization and social dominance orientation. Child Development Advance on line publication retrieved from https://doi.org/10.1111/cdev.13295. Albarello, F., Prati, F., Sangiorgi, L., Tremosini, M., Menegatti, M., Depolo, M., & Rubini, M. (2019). Does Hub-and-Spoke organization of healthcare system promote workers’ satisfaction? Journal of Applied Social Psychology Advance on line publication retrieved from https://doi.org/10.1111/jasp.12623. Allport, G. (1954). The nature of prejudice. Reading, MA: Addison-Wesley.
71
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
Brandstätter, V., Ellemers, N., Gaviria, E., Giosue, F., Huguet, P., Kroon, M., & Perez, J. A. (1991). Indirect majority and minority influence: An exploratory study. European Journal of Social Psychology, 21, 199–211. https://doi.org/10.1002/ejsp.2420210303. Brewer, M. B., & Alexander, M. G. (2002). Intergroup emotions and images. In D. M. Mackie, & E. R. Smith (Eds.). From prejudice to intergroup relations: Differentiated reactions to social groups (pp. 209–225). New York, NY: Psychology Press. Brown, R. J. (2011). Prejudice: Its social psychology. Oxford, UK: Wiley-Blackwell. Bruneau, E., Kteily, N., & Laustsen, L. (2018). The unique effects of blatant dehumanization on attitudes and behavior towards Muslim refugees during the European ‘refugee crisis’ across four countries. European Journal of Social Psychology, 48, 645–662. https://doi.org/10.1002/ejsp.2357. Claps, E., & Vitale, T. (2011). Not always the same old story: Spatial segregation and feelings of dislike against Roma and Sinti in large cities and medium-size towns in Italy. In M. Stewart, & M. Rovid (Eds.). Multi-disciplinary approaches to Romany studies (pp. 228–253). Budapest, H: Central European University Press. Chang, L. W., Krosch, A. R., & Cikara, M. (2016). Effects of intergroup threat on mind, brain, and behavior. Current Opinion in Psychology, 11, 69–73. https://doi.org/ 10.1016/j.copsyc.2016.06.004. Cottrell, C., & Neuberg, S. L. (2005). Different emotional reactions to different groups: A sociofunctional threat-based approach to “prejudice”. Journal of Personality and Social Psychology, 88, 770–789. https://doi.org/10.1037/0022-3514.888.5.770. Cottrell, C., Richards, D. A. R., & Nichols, A. L. (2010). Predicting policy attitudes from general prejudice versus specific intergroup emotions. Journal of Experimental Social Psychology, 46, 247–254. https://doi.org/10.1016/j.jesp.2009.10.008. Council of Europe (2018). Estimates on Roma population in European countries. Retrieved fromhttps://www.coe.int/en/web/portal/roma. Cuddy, A. J. C., Fiske, S. T., & Glick, P. (2007). The BIAS map: Behaviors from intergroup affect and stereotypes. Journal of Personality and Social Psychology, 92, 631–648. https://doi.org/10.1037/0022-3514.92.4.631. Doosje, B., Zimmermann, A., Küpper, B., Zick, A., & Meertens, R. (2009). Terrorist threat and perceived Islamic support for terrorist attacks as predictors of personal and institutional out-group discrimination and support for anti-immigration policies–Evidence from 9 European countries. Revue Internationale de Psychologie Sociale, 22, 203–233. European Commission (2014). Roma heath report: Health status of the Roma population. Data collection in the Member States of the European UnionBrussels, B: European Unionhttps://doi.org/10.2772/33515. Fino, E., Menegatti, M., Avenanti, A., & Rubini, M. (2016). Enjoying vs. smiling: Facial muscular activation in response to emotional language. Biological Psychology, 118, 126–135. https://doi.org/10.1016/j.biopsycho.2016.04.069. Fiske, S. T., Cuddy, A. C., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82, 878–902. https://doi.org/10.1037/0022-3514.82.6.878. FRA-Fundamental Rights Agency (2008). Incident report: Violent attacks against Roma in the Ponticelli district of Naples, ItalyVienna, A: EUFRA. FRA-Fundamental Rights Agency (2017). Second European Union minorities and discrimination survey Muslims – Selected findings. Luxembourg, L: Publications Office of the European Unionhttps://doi.org/10.2811/072254. FRA-Fundamental Rights Agency (2018). A persisting concern: Anti-Gypsyism as a barrier to Roma inclusion. Luxembourg, L: Publications Office of the European Unionhttps://doi.org/10.2811/423901. Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal of Personality and Social Psychology, 74, 1464–1480. https://doi.org/10.1037/0022-3514.74.6.1464. Haselton, M. G., & Nettle, D. (2006). The paranoid optimist: An integrative evolutionary model of cognitive biases. Personality and Social Psychology Review, 10, 47–66. https://doi.org/10.1207/s15327957pspr1001_3. Haslam, S. A., Oakes, P. J., Turner, J. C., & McGarty, C. (1995). Social categorization and group homogeneity: Changes in the perceived applicability of stereotype content as a function of comparative context and trait favourableness. The British Journal of Social Psychology, 34, 139–160. https://doi.org/10.1111/j.2044-8309. 1995.tb01054.x. Hayes, A. F., & Preacher, K. J. (2014). Statistical mediation analysis with a multicategorical independent variable. The British Journal of Mathematical and Statistical Psychology, 67, 451–470. https://doi.org/10.1111/bmsp.12028. Ho, A. K., Sidanius, J., Cuddy, A. J. C., & Banaji, M. R. (2013). Status boundary enforcement and the categorization of Black–White biracials. Journal of Experimental Social Psychology, 49, 940–943. https://doi.org/10.1016/j.jesp.2013.04.010. Hodson, G., & Costello, K. (2007). Interpersonal disgust, ideological orientation, and dehumanization as predictors of intergroup attitudes. Psychological Science, 18, 691–698. https://doi.org/10.1111/j.1467-9280.2007.01962.x. Ion, G. (2011). Romanian pupils at the Spanish primary schools: Continuities and discontinuities between former and current educational experiences. International Journal of Instruction, 4, 193–210. ISTAT-Istituto Nazionale di Statistica (2018). Foreigner population in Italy and in Italian regions updated at the 1st of January 2018. Retrieved fromhttp://dati.istat.it/ Index.aspx?DataSetCode=DCIS_POPSTRRES1. Judd, C. M., & Park, B. (1988). Out-group homogeneity: Judgments of variability at the individual and group levels. Journal of Personality and Social Psychology, 54, 778–788. https://doi.org/10.1037/0022-3514.54.5.778. Kende, A., Hadarics, M., & Láštiková, B. (2017). Anti-Roma attitudes as expressions of dominant social norms in Eastern Europe. International Journal of Intercultural Relations, 60, 12–27. https://doi.org/10.1016/j.ijintrel.2007.06.002. Leyens, J. P., & Yzerbyt, V. Y. (1992). The ingroup overexclusion effect: Impact of valence and confirmation on stereotypical information search. European Journal of Social Psychology, 22, 549–569. https://doi.org/10.1002/ejsp.2420220604. Maass, A. (1999). Linguistic intergroup bias: Stereotypes-perpetuation through language. In M. P. Zanna (Vol. Ed.), Advances in experimental social psychology: Vol. 31, (pp. 79–121). San Diego, CA: Academic Press. https://doi.org/10.1016/S0065-2601(08)60272-5. Maass, A., Castelli, L., & Arcuri, L. (2000). Measuring prejudice: Implicit versus explicit techniques. In D. Capozza, & R. Brown (Eds.). Social identity processes: Trends in theory and research (pp. 96–116). Thousand Oaks, CA: Sage Publications Ltd. https://doi.org/10.4135/9781446218617.n7. Maass, A., Salvi, D., Arcuri, L., & Semin, G. R. (1989). Language use in intergroup contexts: The linguistic intergroup bias. Journal of Personality and Social Psychology, 57, 981–993. https://doi.org/10.1037/0022-3514.57.6.981. Mackie, D. M., Devos, T., & Smith, E. R. (2000). Intergroup emotions: Explaining offensive action tendencies in an intergroup context. Journal of Personality and Social Psychology, 79, 602–616. https://doi.org/10.1037/0022-3514.79.4.602. Malik, M. (2009). Anti-Muslim prejudice in the West, past and present: An introduction. Patterns of Prejudice, 43, 207–212. https://doi.org/10.1080/ 00313220903109144. Mobbs, D., Marchant, J. L., Hassabis, D., Seymour, B., Tan, G., Gray, M., ... Frith, C. D. (2009). From threat to fear: The neural organization of defensive fear systems in humans. Journal of Neuroscience, 29, 12236–12243. https://doi.org/10.1523/JNEUROSCI.2378-09.2009. Moscatelli, S., Albarello, F., Prati, F., & Rubini, M. (2014). Badly off or better off than them? The impact of relative deprivation and relative gratification on intergroup discrimination. Journal of Personality and Social Psychology, 107, 248–264 10/1037/a0036704. Moscatelli, S., Albarello, F., & Rubini, M. (2008). Linguistic discrimination in minimal groups: The impact of status differentials. Journal of Language and Social Psychology, 27, 140–154. https://doi.org/10.1177/0261927X07313652. Moscatelli, S., Menegatti, M., Albarello, F., Pratto, F., & Rubini, M. (2019). Can we identify with a nation low in morality? The heavy weight of (im) morality in international comparison. Political Psychology, 40, 93–110. https://doi.org/10.1111/pops.12504. Mummendey, A., & Wenzel, M. (1999). Social discrimination and tolerance in intergroup relations: Reactions to intergroup difference. Personality and Social Psychology Review, 3, 158–174. https://doi.org/10.1207/s15327957pspr0302_4. Nadotti, C. (2008, September 25). “Noi Romeni brava gente”. Bucarest compra spot [“Us Romanians, good people”. Bucarest buys advertisement]. La Repubblica. Retrieved from http://www.repubblica.it/2008/09/sezioni/politica/immigrazione/spot-romeni/spot-romeni.html. Oakes, P., Haslam, S. A., & Turner, J. C. (1998). The role of prototypicality in group influence and cohesion: Contextual variation in the graded structure of social categories. In S. Worchel, J. F. Morales, D. Páez, & J.-C. Deschamps (Eds.). Social identity: International perspectives (pp. 75–92). Thousand Oaks, CA: Sage
72
International Journal of Intercultural Relations 73 (2019) 59–73
F. Albarello, et al.
Publications, Inc. https://doi.org/10.4135/9781446279205.n6. Ogan, C., Willnat, L., Pennington, R., & Bashir, M. (2014). The rise of anti-Muslim prejudice: Media and Islamphobia in Europe and in the United States. International Communication Gazette, 76, 27–46. https://doi.org/10.1177/1748048513504048. Öhman, A. (2000). The role of the amygdala in human fear: Automatic detection of threat. Psychoneuroendocrinology, 30, 953–958. https://doi.org/10.1016/j. psyneuen.2005.03.019. Orosz, G., Bruneau, E., Tropp, L. R., Sebestyén, N., Tóth-Király, I., & Böthe, B. (2018). What predictions anti-Roma prejudice? Qualitative and quantitative analysis of everyday sentiments about the Roma. Journal of Applied Social Psychology, 48, 318–328. https://doi.org/10.1111/jasp.12513. Park, B., & Rothbart, M. (1982). Perception of out-group homogeneity and levels of social categorization: Memory for the subordinate attributes of in-group and outgroup members. Journal of Personality and Social Psychology, 42, 1051–1068. https://doi.org/10.1037/0022-3514.42.6.1051. Pereira, C., Vala, J., & Costa‐Lopes, R. (2010). From prejudice to discrimination: The legitimizing role of perceived threat in discrimination against immigrants. European Journal of Social Psychology, 40, 1231–1250. https://doi.org/10.1002/ejsp.718. PEW Research Centre (2017). Five facts about the Muslim population in Europe. Retrieved fromhttp://www.pewresearch.org/fact-tank/2017/11/29/5-facts-about-themuslim-population-in-europe/. Povoledo, E., & Pianigiani, G. (2018). Italian Minister moves to count and expel Roma, drawing outrage (2018, June 19) Retrieved fromThe New York Timeshttps://www. nytimes.com/2018/06/19/world/europe/italy-roma-matteo-salvini.html. Preacher, K. J., & Hayes, A. F. (2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods Instruments & Computers, 36, 717–731. https://doi.org/10.3758/BF03206553. Rios, K., Sosa, N., & Osborn, H. (2019). An experimental approach to intergroup threat theory: Manipulations, moderators, and consequences of realistic vs. symbolic threat. European Review of Social Psychology, 29, 212–255. https://doi.org/10.1080/10463283.2018.1537049. Rothbart, M., Fulero, S., Jensen, C., Howard, J., & Birrell, P. (1978). From individual to group impressions: Availability heuristics in stereotype formation. Journal of Experimental Social Psychology, 14, 237–255. https://doi.org/10.1037/0022-3514.55.6.861. Rozin, P., Haidt, J., & McCauley, C. (2000). Disgust. In M. Lewis, & J. Haviland (Eds.). Handbook of emotions (pp. 637–652). (2nd ed.). New York, NY: Guilford Press. Rubin, M., & Paolini, S. (2014). Out-group flies in the in-group’s ointment. Social Psychology, 45, 265–273. https://doi.org/10.1027/1864-9335/a000171. Rubini, M., Menegatti, M., & Moscatelli (2014). Strategic use of language abstraction in achieving symbolic and practical goals: Evidence from intergroup, relational, organizational and political domains. In M. Hewstone, & A. Manstead (Vol. Eds.), European review of social psychology: Vol. 25, (pp. 263–313). London, UK: Psychology Press. https://doi.org/10.1080/10463283.2014.985501. Rubini, M., Moscatelli, S., Albarello, F., & Palmonari, A. (2007). Group power as a determinant of interdependence and intergroup discrimination. European Journal of Social Psychology, 37, 1203–1221. https://doi.org/10.1002/ejsp.391. Rubini, M., Roncarati, A., Ravenna, M., Albarello, F., Moscatelli, S., & Semin, G. (2017). Denying psychological properties of girls and prostitutes: The role of verbal insults. Journal of Language and Social Psychology, 36, 226–240. https://doi.org/10.1177/0261927X16645835. Semin, G. R., & Fiedler, K. (1988). The cognitive functions of linguistic categories in describing persons: Social cognition and language. Journal of Personality and Social Psychology, 54, 558–568. https://doi.org/10.1037/0022-3514.54.4.558. Semin, G. R., & Fiedler, K. (1991). The linguistic category model, its bases, applications and range. In W. Stroebe, & M. Hewstone (Vol. Eds.), European review of social psychology: Vol. 2, (pp. 1–30). London, UK: Psychology Press 1080/14792779143000006. Simon, B. (1992). The perception of ingroup and outgroup homogeneity: Reitroducing the intergroup context. In W. Stroebe, & M. Hewstone (Vol. Eds.), European review of social psychology: Vol. 3, (pp. 1–30). London, UK: Psychology Press. https://doi.org/10.1080/14792779243000005. Squires, N. (2018, June 12). Italy questions migrants who allegedly made death threats against crew of rescue ship. The Telegraph. Retrieved from https://www. telegraph.co.uk/news/2018/07/12/italy-questions-migrants-allegedly-made-death-threats-against/. Stephan, W. G., & Stephan, C. W. (2000). An integrated threat theory of prejudice. In S. Oskamp (Ed.). Reducing prejudice and discrimination (pp. 23–45). Mahwah, NJ: Lawrence Erlbaum Associates. Treccani (2017). Dizionario della lingua italiana [Dictionary of the Italian language]. Florence, I: Giunti. Varjú, V., & Plaut, S. (2017). Media mirrors? Framing Hungarian Romani migration to Canada in Hungarian and Canadian Press. Ethnic and Racial Studies, 40, 1096–1113. https://doi.org/10.1080/01419870.2017.1266007. Villano, P., Fontanella, L., Fontanella, S., & Di Donato, M. (2017). Stereotyping Roma people in Italy: IRT models for ambivalent prejudice measurement. International Journal of Intercultural Relations, 57, 30–41. https://doi.org/10.1016/j.ijintrel.2017.01.003. Woodcock, S. (2007). Romania and Europe: Roma, Roma and Țigani as sites for contestation of ethno-national identities. Patterns of Prejudice, 41, 493–515. https:// doi.org/10.1080/0031322070165729.
73