Introduction

Integration between culturally diverse groups is one of the greatest challenges of modern societies. While governments and educational systems worldwide have developed social strategies to mitigate these antagonisms, none has proven to be successful in all situations. Licata et al. (2011) claim that prejudice is rooted in a form of interaction where a majority group does not recognize another minority group as deserving of social esteem, that is, as worthy of access to the public sphere. Thus, although the formal rights of foreign citizens are being progressively recognized, as long as negative emotions and stereotypes continue replicating, their integration in societies will remain a distant ideal.

In this sense, according to Rueda and Navas (1996), in a contemporary social climate where the majorities value democratic and egalitarian ideals, there is a social rejection towards open discriminatory behavior, based on race, religion or ethnicity. On the other hand, there are implicit forms of discrimination against certain type of social groups, which would be accepted by the social environment. Thus, it is common to subtly discriminate against certain groups based on cultural differences, economic competence (for example, when they are seen as competitors for labor supply) or the fact that they benefit from state resources. In certain social circumstances, this expression can turn into blatant racist manifestations (Rueda & Navas, 1996).

As an example, research has found a relationship between the unfavorable perception of the economic situation and the prejudice towards immigrants (Cosby, Aanstoos, Matta, Porter & James, 2013; Fischer, Hanke & Sibley, 2012; Kessler & Freeman, 2005; Sánchez, 2014). Furthermore, Domenech and Magliano (2008) explain that in periods of economic crisis, immigrant groups tend to be considered as "unacceptable" and tend to be blamed for the social ills of the time. In Argentina, especially in these critical circumstances, mass media -socially defined as conservatives- have spread discourses aimed to promote xenophobic expressions against certain minorities coming from other countries (Pizarro, 2012; Sar, 2016; Valverde, 2015).

Thus, ethnic prejudice uses particular traits of minorities, connoting them negatively and using them as legitimizing beliefs of discriminatory practices. In addition, different contextual aspects -such as migratory flows that make certain groups of migrants visible, legislative frameworks of regulation and integration / exclusion policies, economic and social crises and hegemonic discourses- contribute to the exacerbation of these manifestations. During the last decades, Argentina has develop less restrictive legal frameworks aimed at the inclusion of immigrant groups with an emphasis on the human rights perspective (INADI, 2016a; López Rita, 2017; Velez & Maluf, 2017). However, equalitarian rights have not been really conquered in the exercise of citizenship (López Rita, 2017). In this sense, true integration would imply the cessation of discriminatory behavior and representations deeply rooted in Argentine society (INADI, 2016b; Muller, Ungaretti & Etchezahar 2017; Sar, 2016). In this framework, the study of ethnic prejudice continues to be a relevant issue.

Traditionally, the empirical study of ethnic prejudice focused on the most blatant forms of the phenomenon, capturing extreme and intolerant expressions of racism based, for example, on the belief in the genetic superiority of the ingroup (Rueda & Navas, 1996). Progressively, the measurement instruments developed in the initial stages of prejudice studies began to show evidence of decreasing levels of prejudice. However, social inequalities and discriminatory practices continued to be recorded (Gaertner & Dovidio, 1986). This raised the suspicion that the instruments were not adequately accounting for the phenomenon they wanted to evaluate. In this vein, Frey and Gaertner (1986) showed results from numerous studies in different contexts that stated that racism had not decreased. In fact, old forms of ethnic prejudice had turned into more subtle, complex and maybe more insidious forms of intolerance. Therefrom, new theoretical approaches were developed that analyzed these “new” and more indirect forms of rejection in terms of "modern racism" (Campo-Arias, Oviedo, & Herazo, 2014; Chambers, Schlenker & Collisson, 2013; McConahay, Hardee & Batts, 1981), "symbolic racism" (Berg, 2013; Chambers, Schlenker & Collisson, 2013; McConahay & Hough, 1976), "aversive racism" (Dovidio, Gaertner, & Pearson, 2017; Gaertner & Dovidio, 1986; Minero & Espinoza, 2016), and "subtle prejudice" (Pettigrew & Meertens, 1992; 1995).

From this framework, Pettigrew and Meertens (1995) pointed out the inability of the existing prejudice measurement to capture these new forms of discrimination. Consequently, they developed a more complex scale that aimed to capture both subtle and blatant prejudice. Thereby, they contribute to bring this distinction from the theoretical level to an empirical one (Coenders, Scheepers, Sniderman & Verberk, 2001) developing a validated measurement that could -at least partially- overcome the influence of factors such as social desirability (Cárdenas et al., 2007; Gaertner & Dovidio, 1986). Additionally, that empirical proposal not only differentiates between those two general aspects of prejudice, but also proposes a multidimensional structure within them. On the one hand, they posit two dimensions accounting for blatant prejudice. The first one was named as Threat and Rejection and assess racist beliefs based on the genetic inferiority of the outgroup, used as a justification of the unfavorable position of the latter in society, also denying the existence of discrimination towards these groups. The second component of blatant prejudice is named Intimacy and it refers to an emotional resistance to maintain close relationships with the outgroup. On the other hand, subtle prejudice comprises three dimensions that express prejudice in ways that are considered normative and acceptable in Western societies (Pettigrew & Meertens, 1995). The first dimension of subtle prejudice entails the defense of traditional values ​​inherent to the ingroup and used them as a parameter from which to determine which behaviors are acceptable and necessary to be successful in the society in question. In contrast, it assumes that members of the outgroup act in improper ways. The second component involves the exaggeration of cultural differences, considering them as the reason of the disadvantaged position of the outgroup. Although these cultural differences may be real, in this case they are exaggerated and become stereotypes. Finally, the third subtle prejudice dimension involves the disguised denial of positive emotional responses to the outgroup. This dimension is understood as the most novel contribution of this proposal because it includes an affective dimension (Rueda & Navas, 1996).

This instrument has been translated and adapted for the Spanish context by Rueda and Navas (1996), being that the first psychometric study for a Spanish version. In turn, it has been applied for other researchers to evaluate prejudice towards ethnic minorities with satisfactory results (Canto, Perles, & San Martín, 2012; Cárdenas et al., 2007, Muñiz, Serrano, Aguilera & Rodriguez, 2013; Rubalcaba & Quintero, 2013; Ramirez Barría, Estrada Goic, & Yzerbyt, 2017). This version of the scale showed a two-dimensional structure and satisfactory reliability coefficients that allow the differentiation between subtle and blatant prejudice. Additionally, both dimensions were positively but moderately related which evidences that they are two distinct expressions of the same phenomenon. Also, Rueda and Navas (1996) warn about the possibility that the evaluation of both dimensions is subject to the effect of social desirability, requiring new attempts of adjustment to the social context.

In Latin America, the adaptation of Cárdenas et al. (2007) to the Chilean context specifically addresses prejudice towards ethnic minorities in the region and presents psychometric evidence that supports its validity. These authors also replicated a two-dimensional structure that reflected the theoretical dimensions of the original construct, being consistent with the previous evidence. They also presented evidence of satisfactory reliability, as well as positive and significant correlations between the two sub-scales.

In the case of Argentina, up to now are not any local psychometric studies on the application of this scale for the analysis of prejudice towards immigrants. However, there is a local version of this instrument which assesses prejudice toward “villeros” (stereotypical representation about people of social disadvantaged groups who lives in “villas”) (Muller et al., 2017). There is also a theoretical review that presents evidence on the development of this instrument in Latin America (Ungaretti, 2017). This previous studies are relevant antecedents for our own research proposal, being relevant to develop specific instruments to assess attitudes toward immigrants.

Consequently, our objective is to adapt and test a local version of the Subtle and Blatant Prejudice Scale developed by Pettigrew and Meertens (1995), evaluating its psychometric properties and its validity for the study of prejudice towards immigrants in Argentina. In addition, we assess not only the differences between the subtle and blatant dimensions, but the sub-dimensions within them. We also seek to provide evidence of convergent validity analyzing it’s correlation with Assimilationism and Right Wing Authoritarianism (RWA), two variables that the literature has considered to be related to prejudice. This analysis will allow us to account for the adequacy of this instrument to measure the phenomenon we are interested in (Blacker & Endicott, 2002).

Assimilationism is a form of integration in which people recognized as culturally diverse must adapt to the dominant culture which concept is opposed to multiculturalism (Levin et al., 2012). Previous research has shown that those who hold attitudes that imply the desire for immigrants to abandon their customs and cultural values ​​in order to adapt to the dominant ones, tend to show more prejudice and are less willing to demonstrate positive emotions towards them (Berry, 2006; Guimond et al., 2010; Levin et al., 2012). This differentiation in-outgroup, which associate a strong negative connotation to the outgroup characteristics, accounts for a belief in the superiority of the ingroup, a typical expression of prejudice (Bizumic & Duckitt, 2012; Jost & Hunyady, 2003; Monsegur, Espinosa & Beramendi, 2014).

We also analyzed the correlation between anti-immigrant prejudice and RWA. Previous studies posited that, among groups characterized by an orientation towards the preservation of norms, RWA was a strong predictor of negative attitudes towards diverse ethnic groups (Dru, 2007) or towards immigrants (Quinton, Cowan & Watson, 1996). According to some studies, the activation of threat perception could explain this correlation between RWA and Prejudice (Cohrs & Asbrock, 2009; Dru, 2007; Duckitt & Sibley, 2009).

Method

Results

Table 1 shows the descriptive indexes for each of the items on the prejudice scale. According to the criteria established by George and Mallery (2011) to account for univariate normality, asymmetry and kurtosis values ± 2.00 are considered to be excellent and values ​​between ± 1.00 adequate. Consequently, item 5 ("Los inmigrantes proceden de razas menos capaces y esto explica por qué no les va tan bien como a muchos de los argentines” - immigrants came from less competent races which explains why they live in worse conditions than Argentines-) and item 9 ("No me importaría si un inmigrante adecuadamente preparado fuera mi profesor o jefe” -I would not mind if a properly prepared immigrant was my teacher or boss-) were considered to be inadequate and were eliminated from subsequent analyzes.

Including the 18 items that presented an adequate adjustment, we proceeded to study the construct validity through a confirmatory factor analysis (CFA). Thus, we specified the multidimensional model of 5 components distributed according to the original proposal (Pettigrew & Meertens, 1995). Table 2 shows the model global adjustment indexes. Although the chi-square value yielded an unexpected result as it was statistically significant (p≤ .001), previous literature suggest that this is frequent and may be a product of the sample size (Kline, 2011). Meanwhile, the rest of the fit indexes are within the parameters established by the specialized literature, thus providing evidence of their adequacy.

Figure 1 shows the model structure and the standardized beta coefficients that account for the load of each item on the dimension with their corresponding error terms. In all cases, the statistical significance was p≤.001.

From this structure, we proceeded to the internal consistency analysis for each sub-dimensions. In addition, in order to verify the existence of items that reduce the reliability of the subscales, we estimated the Cronbach's Alpha statistic considering its variation if each element is eliminated (Table 3). As can be observed, the dimensions of blatant prejudice (threat and rejection and intimacy) obtained adequate levels of internal consistency (Maroco & García Marques, 2013). In the case of the intimacy dimension, the exclusion of the item “Tendría relaciones sexuales con un/a inmigrante” (I Would have sex with an immigrant) would increase internal consistency.

Table 1: Descriptive statistics of Subtle and Blatant Prejudice items 

Table 2: Global Fit Indexes of the Subtle and Blatant prejudice five factor model 

Figure 1: Confirmatory Factor Analysis of Subtle and Blatant Prejudice Scale 

Table 3: Internal Consistency of Subtle and Blatant prejudice dimensions 

In respect of subtle prejudice, cultural differences dimension showed a satisfactory internal consistency (α = .75), while traditional values ​​dimension was barely below the reference criterion (α = .59). In the latter case, the suppression of the item “Los inmigrantes que viven en Argentina no deberían esforzarse por hacerse un lugar donde no son queridos” (Immigrants who live in Argentina should not strive to obtain a place where they are not wanted) would increase the dimension reliability, although this increase is small (α = .61). Lastly, we could not estimate internal consistency for the positive emotions dimensions since it has only two items. Consequently, we estimated the inter-item correlation which was positive and statistically significant (r = .53; p≤ .001).

To test convergent validy, we estimated the Pearson correlation between each dimension of the prejudice scale, assimilationism and the submission / aggression dimension of RWA. The results exhibit a positive, moderate and statistically significant correlations (p≤ .01) between these variables, as expected (Table 4). This evidence indicates the adequacy of this construct for the evaluation of prejudice.

Table 4: Pearson Correlation Coefficient between Prejudice, Assimilationism and Aggression / Submission dimension of RWA 

Discussion

The present study represents a significant contribution to the study of prejudice in Argentina as it provides an empirically validated instrument for addressing this phenomenon, not only in its most evident forms, but also in its more subtle aspects. More specifically, the results above presented provide evidence about the adequacy of the construct, both with respect to its structure and internal consistency and its external validity. It is important to note that most of the Spanish versions only replicated the differentiation between subtle and blatant prejudice, not being able to distinguish between different attitudinal conglomerates within each of those mentioned (Cárdenas et al., 2007; Rueda & Navas 1996). On our part, we were able to account for the multidimensional structure of the prejudice consistently with the original proposal developed by Pettigrew and Meertens (1995). This is particularly relevant since enables a more comprehensive analysis of this phenomenon.

The empirical evidence obtained regarding the internal consistency of each of these sub-dimensions indicates that the subtle aspects -especially those related to social values- ​​seem to be more difficult to empirically assess. It is possible that, depending on the particularities of the migratory processes in each region, some of these items are ambiguous for many respondents. However, this is a hypothetical assumption and we need complementary evidence to test its adequacy (for example, application of the scale in conjunction with a cognitive interview).

It is relevant to highlight another original contribution of this study which refers to the convergent validity test. We were able to establish the relationship between our version of the subtle and blatant prejudice scale and two documented related variables (Berry, 2006; Dru, 2007; Guimond et al., 2010; Quinton, Cowan & Watson, 1996; Levin et al., 2012): Assimilationism and RWA. The correlation coefficients were more robust for some dimensions (threat and rejection, traditional values ​​and cultural differences), while they were lower for the dimensions of intimacy and positive emotions. That is probably due to the fact that they correspond to the attitudinal aspects that emphasize less on cultural asymmetries, which are those that are more specifically evaluated trough the two convergent variables (especially the Asymilization index). On the other hand, the positive emotions index consists of only two items that can be interpreted in an ambiguous way, as they can refer (positively) to a certain empathy due to the unfavorable situation of immigrants in our country, or (negatively) to demonstrations of condescension towards those groups. Thus, it is possible that this index is not an exact approximation to the affective dimension of subtle prejudice insofar it may be biased by social desirability. However, it would be necessary to collect more empirical evidence to support this hypothesis and it would be desirable to make this measurement more complex, avoiding the evaluation of ambiguous affective aspects.

To sum up, the main findings of our study are the identification of the prevalence of prejudicial attitudes -even among young people- and the relevance of being able to evaluate the subtle dimensions of prejudice in order to achieve a better understanding of this phenomenon. Also, it allowed us to provide a locally validated measure that allows us to dialogue with the previous literature that has applied the same construct in different contexts. Also, as noted by Rueda and Navas (1996), it is possible that these measurements do not avoid the effect of social desirability, which should be controlled in future studies. For this reason, we emphasize the need to update and evaluate the adequacy of our empirical approaches in the study of complex social phenomena that are in permanent transformation.

Evidence of this is the very existence of subtle prejudice, which deepens - to the detriment of blatant expressions- as new rules are established that proscribe open expressions of prejudice and discrimination (Pettigrew & Meertens, 2001). Thus, in the last decades the indirect forms of prejudice represented new ways of preserving hierarchies based on racial, ethnic and religious dominance (Gómez Berrocal & Moya 1999). Consequently, the social function of prejudice is not only maintained, but it is reinvented in ways that -because they are less evident- become increasingly difficult to perceive, challenge and counteract. Furthermore, the use of instruments that are incapable of capturing the refined expression of prejudice could be functional to the masking of the consequences of prejudice. Thus, the illusion is created that it is an increasingly banished phenomenon of our social practices while inequality between groups is attributed to intrinsical and static attributes. Altogether, this allows to rationalize, justify and perpetuate existing hierarchies (Cárdenas et al., 2007). In this framework, the Pettigrew and Meertens (1995) scale and its adaptations, represent a valid contribution for the detection and study of new forms of prejudice in different contexts.

In addition, it is necessary to point out some methodological limitations of our research that must be taken into account for complementary studies. First, data were collected through a non-probabilistic sampling method, which limits the possibility of its generalization to the population. In addition, this instrument assesses attitudes towards immigrants as a general category, although this is not a homogeneous group and it is possible that - in accordance with prevailing hegemonic discourses - people have different attitudes towards immigrants based on aspects such as their origin. Thus, it would be relevant to compare attitudes towards European immigrants and Latin American immigrants; or towards immigrant women, given their particular difficulty in obtaining employment, which has been the subject of previous research (i.e. Andrade-Rubio, 2016; Bruno, 2016; Magliano, Perissinotti & Zenklusen, 2014). Finally, we want to emphasize the relevance of including the perspective of the immigrants themselves, who are directly suffering the consequences of prejudice and discrimination. It is essential to take into account the complexity of these phenomena and its concrete consequences for immigrants in order to devise public policies aimed to reversing situations of inequality and the achieving of real intercultural integration.