Reaping the Benefits of Cultural Diversity: Classroom Cultural Diversity Climate and Students’ Intercultural Competence PDF

Received: 4 January 2019 | Accepted: 17 July 2019 DOI: 10.1002/ejsp.2617 RESEARCH ARTICLE Reaping the benefits of cultural diversity: Classroom cultural diversity climate and students’ intercultural competence Miriam Schwarzenthal1 | Maja K. Schachner1 | Linda...

Received: 4 January 2019 | Accepted: 17 July 2019 DOI: 10.1002/ejsp.2617 RESEARCH ARTICLE Reaping the benefits of cultural diversity: Classroom cultural diversity climate and students’ intercultural competence Miriam Schwarzenthal1 | Maja K. Schachner1 | Linda P. Juang1 | Fons J. R. van de Vijver2,3,4,5* 1 University of Potsdam, Potsdam, Germany 2 Abstract Tilburg University, Tilburg, The Netherlands 3 North‐West University, Potchefstroom, Culturally diverse schools may constitute natural arenas for training crucial intercul‐ South Africa tural skills. We hypothesized that a classroom cultural diversity climate fostering con‐ 4 University of Queensland, St Lucia, Qld, tact and cooperation and multiculturalism, but not a climate fostering color‐evasion, Australia 5 would be positively related to adolescents’ intercultural competence. Adolescents Higher School of Economics, Moscow, Russia in North Rhine‐Westphalia (N = 631, Mage = 13.69 years, 49% of immigrant back‐ ground) and Berlin (N = 1,335, Mage = 14.69 years, 52% of immigrant background) in Correspondence Miriam Schwarzenthal, University of Germany reported their perceptions of the classroom cultural diversity climate and Potsdam, Potsdam, Germany. completed quantitative and qualitative measures assessing their intercultural compe‐ Email: miriam.schwarzenthal@uni-potsdam. de tence. Multilevel structural equation models indicate that contact and cooperation, multiculturalism, and, surprisingly, also color‐evasion (as in emphasizing a common humanity), were positively related to the intercultural competence of immigrant and non‐immigrant background students. We conclude that all three aspects of the class‐ room climate are uniquely related to aspects of adolescents’ intercultural compe‐ tence and that none of them may be sufficient on their own. KEYWORDS color‐evasion, contact, intercultural competence, multiculturalism, school climate 1 | I NTRO D U C TI O N as well as the behavioral flexibility to deal with these. While plenty of social psychological contact research has examined intergroup Adolescents growing up in culturally diverse societies encounter attitudes, little research has empirically examined intercultural com‐ a multitude of worldviews, perspectives, and lifestyles on a daily petence in the context of culturally diverse societies. basis. In order to ensure that the human rights of individuals from Intercultural competence may be acquired by engaging in pro‐ all cultural backgrounds are acknowledged and respected, tolerance longed intercultural contact and by actively reflecting on different and understanding of different worldviews and lifestyles needs to worldviews (e.g., Allport, 1954; Gurin, Dey, Hurtado, & Gurin, 2002). be promoted from an early age (Barrett, Byram, Lázár, Mompoint‐ While the positive social learning outcomes of these experiences Gaillard, & Philippou, 2013). These skills are usually labeled intercul‐ have been well documented among students studying abroad, at‐ tural competence (Deardorff, 2004). tending intercultural trainings or international colleges (American Intercultural competence goes beyond intergroup attitudes by Field Service [AFS], 2012; Busse, Riedesel, & Krause, 2017; Gurin also comprising awareness and knowledge of different worldviews, et al., 2002), the potential of culturally diverse schools to foster *Deceased June 1, 2019 This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made. © 2019 The Authors. European Journal of Social Psychology published by John Wiley & Sons Ltd Eur J Soc Psychol. 2020;50:323–346.  wileyonlinelibrary.com/journal/ejsp | 323  | 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 324 SCHWARZENTHAL et al. intercultural competence has largely been neglected by the pub‐ However, awareness of cultural influences does not mean that all lic debate and by psychological research. Instead, in countries like behavior should be explained with cultural factors. Preconceived Germany, media and research often focus on the challenges that attributions to culture can also foster stereotypes and lead to es‐ go along with cultural diversity in schools (Erdmann, 2017; Stanat, sentialism (Barrett et al., 2013). Therefore, in order to develop high 2006). metacognitive CQ, students should also learn to ask questions, sus‐ However, culturally diverse schools may constitute a natural pend judgments, and consider alternative explanations for behavior arena for training intercultural skills. Adolescence is a crucial devel‐ (Sieck, Smith, & Rasmussen, 2013). Completely neglecting culture, opmental period for acquiring intergroup attitudes and intercultural on the other hand, may result in a “color‐evasive” perspective where skills (Quintana, 1998; Raabe & Beelmann, 2011) and schools are a important information that might aid interpretation of behavior is major context for cross‐cultural encounters (Sam & Oppedal, 2003). neglected (Schofield, 2010). How schools approach cultural diversity may determine whether it Intercultural competence is often assessed using self‐report mainly poses a threat and a challenge or an opportunity for exchange measures, such as the cultural intelligence scale (CQS) (Van Dyne and personal growth. Therefore, we will investigate how three types et al., 2012). However, these measures are not tailored to adoles‐ of classroom cultural diversity climate (fostering contact and co‐ cents and have been criticized because people might not be able to operation, color‐evasion, and multiculturalism) are related to the accurately report their own competences (Klafehn, Li, & Chiu, 2013). intercultural competence of students attending culturally diverse In the face of calls for a multimodal assessment of intercultural com‐ secondary schools in Germany. petence (Deardorff, 2011), situational judgment tests (SJTs) have been used as an alternative assessment method. These are based on a short description of an intercultural situation followed by ques‐ 1.1 | What is intercultural competence? tions assessing the participants’ response to or interpretation of the Defining intercultural competence requires clarifying the underlying situation (Rockstuhl, Ang, Ng, Lievens, & Van Dyne, 2015). In our term “culture” first. Culture consists of different layers, for example, study we use a version of the CQS that was adapted for adolescents, of material culture (e.g., food or dress), social culture (e.g., social rules and SJTs set in the school and peer context. or language), and subjective culture (e.g., attitudes or values) (Chiu & Hong, 2013). Culture is shared by a collective of people, such as 1.2 | How may intercultural competence be a nation, a generation, or a social class. People usually only adopt fostered in schools? their collective's culture to a certain degree, and often adopt ele‐ ments of more than one culture. An individual's compilation of cul‐ The presence of a high proportion of outgroup members in the en‐ tural influences determines what is perceived as normative and thus vironment alone does not automatically promote intercultural learn‐ influences how that person interprets, judges, and reacts to others’ ing. A large proportion of outgroup members may induce feelings of thinking and behavior (Barrett et al., 2013; Spencer‐Oatey, 2008). threat and thus lead to more negative intergroup attitudes (Schmid, Hundreds of characteristics have been brought forward to de‐ Al Ramiah, & Hewstone, 2014; Taylor, 1998) or may increase oppor‐ scribe intercultural competence (Deardorff, 2004). The concept of tunities for intergroup contact (Blau, 1977) and thus lead to more Cultural Intelligence (CQ) (Ang et al., 2007; Van Dyne et al., 2012) positive intergroup attitudes (Allport, 1954; Pettigrew & Tropp, was chosen for the present research because it focuses on charac‐ 2008) and intercultural learning (Gurin et al., 2002). Studies exam‐ teristics that can be learned, rather than stable personality traits, ining relations between the ethnic composition of college or high and its measurement has been deemed one of the most reliable and school classrooms and intergroup attitudes and intercultural learn‐ valid to date (Matsumoto & Hwang, 2013). Individuals possessing ing have come to mixed conclusions—level of classroom diversity high CQ enjoy interacting with others who have different cultural was sometimes unrelated, and sometimes positively related to inter‐ affiliations (motivational CQ), know about norms, values, and behav‐ cultural learning outcomes (e.g., Denson & Chang, 2008; Rothman, iors in different cultures (cognitive CQ), are aware that culture may 2003; Schwarzenthal, Schachner, Van de Vijver, & Juang, 2018; influence their own and others’ behavior and thinking (metacogni‐ Terenzini, Cabrera, Colbeck, Bjorklund, & Parente, 2001). It has been tive CQ), and use appropriate behavior in culturally diverse situa‐ argued that the way schools or classrooms approach cultural diver‐ tions (behavioral CQ). sity is crucial in determining intergroup attitudes and intercultural Even before entering school, children are aware of social group skills (Schwarzenthal et al., 2018; Thijs & Verkuyten, 2014). identities and become increasingly aware of which attitudes or Three types of approaches to cultural diversity in schools may be behaviors are considered to be normative in a group (Rutland & especially relevant for students’ intercultural learning: fostering pos‐ Killen, 2015). Adolescence is a crucial time for the development of itive interactions between students of diverse cultural affiliations intercultural skills as intergroup attitudes are strongly influenced (i.e., fostering an approach of contact and cooperation), promoting by the social context and adolescents gradually learn to take oth‐ the idea that cultural differences are not important (i.e., fostering an er's perspectives (Van der Graaff et al., 2014), and to understand approach of color‐evasion), and/or actively promoting engagement more subtle cultural influences on behavior (Quintana, 1998; with students’ heritage cultures and different perspectives (i.e., fos‐ Schwarzenthal, Juang, Schachner, van de Vijver, & Handrick, 2017). tering an approach of multiculturalism).  | 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License DIVERSITY CLIMATE & INTERCULTURAL COMPETENCE 325 evasive approaches were originally based on the assumption that 1.2.1 | Contact and cooperation prejudice and discrimination result from an emphasis on group cat‐ Fostering contact and cooperation was proposed in the 1950s as a egories and that therefore group membership should be deempha‐ strategy to foster positive interracial attitudes among students at‐ sized and category boundaries eliminated (Rosenthal & Levy, tending recently desegregated schools (Allport, 1954). Since then, a 2010). These approaches not only informed social psychological large body of research has provided support for the claim that more interventions (Gaertner & Dovidio, 2000), but were also adopted intergroup contact goes along with more positive intergroup attitudes, by schools (Schofield, 2010). Schools adopting color‐evasive ap‐ especially if it occurs under certain conditions (the groups share equal proaches emphasize that racial or cultural categories are irrelevant status and common goals, they cooperate, and the contact is sup‐ and treat social interactions as interpersonal and not intergroup ported by an authority) (Allport, 1954; Levy & Killen, 2010; Pettigrew processes. These approaches have been strongly criticized be‐ & Tropp, 2006). In addition to direct personal contact, social norms that cause they tend to go along with ignorance of existing inequalities support positive contact are crucial in determining people's attitudes and structural discrimination (i.e., with “power‐evasion”, Neville, (Christ et al., 2014). Schools may constitute an ideal context for setting Lilly, Duran, Lee, & Browne, 2000). In conceptualizations of class‐ positive contact norms, for example when students help each other room cultural diversity climate, the respective constructs are la‐ with homework or work cooperatively in class (Dejaeghere, Hooghe, beled “colorblind socialization” (Byrd, 2017) or “color‐evasion” & Claes, 2012; Schachner, Noack, Van de Vijver, & Eckstein, 2016). (Schachner et al., 2019). Promotion of positive contact and cooperation is a key dimension of While power‐evasion has an inherently negative connotation, most conceptualizations of classroom cultural diversity climate, and research on color‐evasion (as in emphasizing a common humanity), is subsumed under broader labels such as “interpersonal interactions” suggests that it has positive as well as negative effects. Color‐eva‐ (Byrd, 2017), or “equality and inclusion” (Schachner, Schwarzenthal, sive approaches predict positive intergroup attitudes and behav‐ Moffitt, Civitillo, & Juang, 2019; Schachner et al., 2016). iors such as helping (for a review, see Dovidio, Gaertner, & Saguy, In classrooms in which positive contact and cooperation between 2010), and may reduce intergroup anxiety (Schofield, 2010). students of different cultural affiliations is supported, students tend However, relinquishing group identities altogether may also com‐ to display more positive outgroup orientations (Schwarzenthal et al., promise people's need for distinctiveness, and its effects on pos‐ 2018), more intercultural friendships (Schachner, Brenick, Noack, van itive intergroup attitudes are less stable than expected (Dovidio de Vijver, & Heizmann, 2015), and less prejudice (Molina & Wittig, et al., 2010). Color‐evasion promotes suppression of negative 2006). However, research examining relations between intercultural thoughts in the short term, but preexisting prejudice and nega‐ contact in culturally diverse schools and intercultural competence is tive behavior can rebound in the long term (Correll, Park, & Smith, scarce. The few existing studies found that the length of time stu‐ 2008; Sasaki & Vorauer, 2013). Moreover, researchers criticiz‐ dents had attended an international high school was related to their ing this approach have argued that a complete neglect of racial level of intercultural sensitivity (Straffon, 2003), and that students’ and cultural categories implies that opportunities to learn about individually reported intercultural contact at school was related to cultural diversity are not being used (Schofield, 2010), existing their CQ (Schwarzenthal et al., 2017). Research conducted on cul‐ discrimination is not recognized (Apfelbaum, Pauker, Sommers, turally diverse college campuses lends further support to the claim & Ambady, 2010), and appropriate behavioral strategies for in‐ that intercultural contact is not only related to attitudes, but also to terracial or intercultural interactions are not developed (Sasaki & intercultural knowledge, understanding, and interculturally compe‐ Vorauer, 2013). After 4‐year‐long observations at a middle school tent behavior (Bowman, 2010; Gurin et al., 2002; Jon, 2013). Based in the northeastern U.S., Schofield (2010, p. 287) concludes that on these research findings, we formulate the following hypothesis: “such a tendency, while undeniably a low‐risk one, failed to take advantage of the diversity of experiences and perspectives […] as Hypothesis 1: A classroom climate promoting contact and co‐ a resource for the educational process.” The concept of color‐eva‐ operation between students of diverse cultural affiliations is sion bears similarities to the concept of minimization in Bennett's related to higher motivational, cognitive, metacognitive, and developmental model of intercultural sensitivity (Bennett, 1993). behavioral CQ among students. People in the minimization stage believe that all humans are simi‐ lar, which is part of the lower, or “ethnocentric” stages of intercul‐ tural sensitivity, as opposed to the “ethnorelative” stages in which 1.2.2 | Color‐evasion people start to accept cultural variations. Systematic efforts to foster contact and cooperation between stu‐ The color‐evasion scale used for this research (Schachner et al., dents can go along with neglecting cultural variations—and thus 2019) captures an emphasis on a common humanity, but does not pursuing a color‐evasive approach (Schachner et al., 2016).1 Color‐ directly assess ignorance of discrimination and structural inequal‐ ities (i.e., “power‐evasion”). Since color‐evasion has some potential 1 The concept of color‐evasion is comparable to what is usually captured by the term to reduce prejudice in the short term, but not in the long term, and “colorblindness.” However, in order to avoid a deficit view on people with disabilities, and to acknowledge that not seeing race/ethnicity/culture is a conscious choice, we prefer entails that intercultural learning opportunities are not being used, the term “color‐evasion” in this research. we formulate the following hypothesis:  | 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 326 SCHWARZENTHAL et al. Hypothesis 2: While a color‐evasive classroom climate may be related to a range of cognitive, metacognitive, and behavioral intercul‐ related positively to motivational CQ, we expect that it is tural learning outcomes (Bowman, 2010; Gurin et al., 2002; Maddux, unrelated to cognitive, metacognitive, and behavioral as‐ Bivolaru, Hafenbrack, Tadmor, & Galinsky, 2013). Thus, we formulate pects of CQ among students. the following hypothesis: Hypothesis 3: A multicultural classroom climate is related to higher motivational, cognitive, metacognitive, and behav‐ 1.2.3 | Multiculturalism ioral CQ among students. According to the multicultural approach, race, ethnicity, and culture should be paid attention to, group differences should be valued, and 1.2.4 | Relations between the three approaches individuals should learn about the perspectives of various groups in society (Sasaki & Vorauer, 2013). Supporters of the multicultural The diversity approaches described above are not mutually exclusive approach argue that in order to reduce prejudice, it is not necessary (Rosenthal & Levy, 2010). It is possible that positive contact is essen‐ to remove intergroup categories. Instead, it is assumed that learning tial for anxiety to be reduced before people can undertake the cog‐ about and critically reflecting on diversity can reduce prejudice (Park nitive effort to engage in perspective‐taking and gaining knowledge & Judd, 2005; Rosenthal & Levy, 2010), and foster intercultural com‐ about the outgroup (i.e., before they can profit from multicultural ap‐ petence (Barrett, 2018). In education, a multicultural approach has proaches) (Richeson & Shelton, 2003). A focus on a common humanity become increasingly popular since the 1970s and 1980s (Schachner, (as promoted by color‐evasive approaches) may help to lessen risks of 2017). For example, Banks (2015), one of the leading scholars in mul‐ stereotyping and “othering”, while learning about cultural variations ticultural education, proposes that schools should not only focus on (as in multicultural approaches) can increase outgroup knowledge and prejudice reduction, but should also include cultural content in the perspective‐taking. Despite these theoretical considerations, relatively curriculum, and foster understanding of implicit cultural assumptions few studies have directly compared the different types of classroom and perspectives. In conceptualizations of classroom cultural diversity cultural diversity climate in one study (Rosenthal & Levy, 2010), and climate, multicultural approaches are often subsumed under labels none have looked at relations with students’ intercultural competence. such as “cultural socialization”, “promotion of cultural competence” (Byrd, 2017), or “cultural pluralism” (Schachner et al., 2016, 2019). 1.3 | Individually perceived and classroom‐ Multiculturalism is associated with positive intergroup attitudes aggregated cultural diversity climate and lower ethnocentrism (Rosenthal & Levy, 2010; Schwarzenthal et al., 2018; Verkuyten & Thijs, 2013). However, when a multicultural In order to assess classroom climate, most researchers ask students approach puts too much emphasis on differences (and essentializes for their perceptions, using the classroom or the teacher as a refer‐ these differences), it can also increase stereotyping and reduce ent (Lam, Ruzek, Schenke, Conley, & Karabenick, 2015; Wang & Degol, perceived similarity and liking (Rosenthal & Levy, 2010; Schofield, 2016). The individual student ratings are then aggregated at the class‐ 2010). Indeed, multiculturalism in schools is also related to more per‐ room level to form measures of classroom climate (e.g., see Marsh ceived discrimination (Schwarzenthal et al., 2018). et al., 2012). However, in many studies students’ perceptions of their Even though empirical research assessing the impact of multicul‐ classroom's climate are quite heterogeneous (Lam et al., 2015; Miller tural education at school on intercultural knowledge, understanding, & Murdock, 2007). This may be due to differential treatment that stu‐ and interculturally competent behavior is scarce, theories and previous dents receive in the same classroom, and to students’ idiosyncratic research from other areas suggest that multicultural education may interpretations of experiences (Wang & Degol, 2016). For example, positively affect these aspects of intercultural competence. A multi‐ cultural minority members tend to perceive a more negative climate of cultural ideology encourages active engagement with and learning contact and cooperation in the classroom than cultural majority mem‐ about diversity (Crisp & Turner, 2011; Sasaki & Vorauer, 2013). Active bers (Byrd, 2014; Schwarzenthal et al., 2018). Consequently, effects of engagement with diversity does not only increase cultural knowledge, individually perceived classroom diversity climate on intergroup and but can also shape the structure of our thoughts, as engagement with psychological outcomes are usually stronger than the effects of the ag‐ new information that does not fit into one's schema drives the learning gregated diversity climate (Schachner et al., 2016; Schwarzenthal et al., process and fosters cognitive development (Piaget, 1977). This idea in‐ 2018). In this research, we will therefore investigate effects of the indi‐ formed researchers in acculturation psychology and higher education vidually perceived as well as of the classroom‐aggregated climate. who assume that engaging with diverse perspectives results in a higher complexity of cultural representations (Tadmor & Tetlock, 2006), and 1.4 | Differential effects between students of promotes students’ intercultural learning and cognitive growth (Gurin immigrant and non‐immigrant background et al., 2002). Indeed, norms around multiculturalism (as compared to color‐evasion) are related to increased perspective‐taking (Todd & The effect of cultural diversity climate on CQ may differ between dif‐ Galinsky, 2012) and more positive other‐directed remarks (Vorauer, ferent groups of students. In Germany, about one‐third of the student Gagnon, & Sasaki, 2009). Learning about cultural diversity in college is population is of immigrant background, meaning that at least one of  | 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License DIVERSITY CLIMATE & INTERCULTURAL COMPETENCE 327 their parents did not acquire German citizenship at birth (Statistisches dataset allows disentangling effects of contact and cooperation and Bundesamt, 2018). Thus, these students are in the minority in the so‐ color‐evasion. Moreover, the larger number of classrooms enables ciety. They are more likely to experience discrimination (Frankenberg, better investigation of relations with the classroom‐aggregated cul‐ Kupper, Wagner, & Bongard, 2013) and typically have lower socioeco‐ tural diversity climate. nomic status than students of non‐immigrant background (Kristen & To test our hypotheses, we use a multiple indicators, multiple Granato, 2007). If for them intercultural interactions are marked by causes structural equation model (see for example van de Vijver, status differences and experiences of discrimination, it is less likely that 2002), which links input (i.e., classroom cultural diversity ap‐ they will lead to the development of CQ, because conflict can hinder proaches) and output (i.e., different facets of CQ or SJT) through a positive intercultural learning outcomes (Correll et al., 2008). latent variable (i.e., CQ or SJT). This approach also allows detecting Moreover, students of immigrant background encounter cultural direct relations between predictor variables and the different CQ or diversity on a daily basis. While the family and same‐ethnic peer SJT facets through modification indices. contexts are typically more influenced by their ethnic culture, the We combine this approach with a multilevel framework which school and different‐ethnic peer contexts are more influenced by allows taking the clustered nature of the data into account, and in‐ the mainstream culture (Motti‐Stefanidi, Berry, Chryssochoou, Sam, vestigating effects of the classroom‐aggregated cultural diversity & Phinney, 2012). For students of non‐immigrant background on climate. Thus, our approach is similar to the doubly latent framework the other hand, intercultural experiences may be less typical, im‐ suggested by Marsh et al. (2012) to study school context and climate plying that these experiences may have a stronger impact on their effects, with the exception that only the outcome variables, not the intergroup attitudes and intercultural skills if they occur (Denson & predictor variables, were modeled as latent. We decided to only Zhang, 2010; Loes, Pascarella, & Umbach, 2012; Tropp & Pettigrew, model the outcome variables as latent variables because in multilevel 2005). Thus, we formulate the following hypothesis: models, the sample size at the classroom level is determined by the number of clusters. If all variables are modeled as latent variables, the Hypothesis 4: Relations between contact and cooperation and number of parameters to be estimated at the classroom level quickly multiculturalism on the one hand and CQ on the other hand exceeds the numbers of clusters, and the models get too complex. are stronger among students of non‐immigrant background We included gender, immigrant background, grade average (i.e., than among those of immigrant background. average of grades in Math, English, and German, recoded so that higher scores reflect better grades), number of books in the house‐ hold (5‐point Likert scale from (1) none or very few to (5) more than 200 books, Bos et al., 2003), and the proportion of students of immi‐ 1.5 | The present research grant background in a classroom as control variables in both studies, In this research, we assess how three types of classroom cultural since they are related to intercultural competence among adoles‐ diversity climate (contact and cooperation, color‐evasion, and cents (Schwarzenthal et al., 2017; Terenzini et al., 2001).3 multiculturalism) are related to CQ among adolescents of immi‐ grant and non‐immigrant background attending culturally diverse 2 | S T U DY 1 schools in Germany. 2 In Germany, the notion that cultural diversity can be beneficial was only officially recognized in the 1990s, when 2.1 | Method new guidelines by the Conference of the Ministers of Education of the German Federal States compelled schools to incorporate in‐ 2.1.1 | Participants and procedure tercultural learning into the curriculum for all children (Faas, 2008). However, until today German schools focus more on preju‐ After obtaining permission from school principals, parental ap‐ dice reduction and equality than on multiculturalism and intercul‐ proval, and students’ assent or consent, we administered a ques‐ tural learning (Civitillo et al., 2016; Schachner et al., 2016). tionnaire in 29 6th, 8th, and 10th grade classrooms in seven schools We base our analyses on data collected in two culturally di‐ in North Rhine‐Westphalia, Germany in the winter of 2015/2016 verse regions in Germany: Study 1 is based on data from 631 6th as part of a larger cross‐national study on inclusive identity. A total to 10th graders in 29 classrooms in North Rhine‐Westphalia. Using of 631 students (48.2% female, M age = 13.69 years, SD age = 1.83, this sample, we investigate relations between two cultural diversity approaches in the classroom (contact and cooperation and multi‐ 3 Re‐running the analyses without any control variables did not substantially alter the results. We used the number of books in the household as a control variable for several culturalism) and students’ CQ. In Study 2, we try to replicate and reasons: Direct information from the parents on their income or education was not expand the results from Study 1 with a sample of 1,335 9th grad‐ available in our study, and children are often unable to report their parent's educational background or occupation correctly (for a review, see Currie et al., 2008). Even though in ers in 66 classrooms in Berlin. Since these students were also asked Study 2, we had also included the Family Affluence scale (Boyce et al., 2006; German to report their perceptions of color‐evasion in the classroom, this version by Richter & Leppin, 2007), the number of books was consistently more strongly related to CQ than the Family Affluence Scale. Therefore, the number of books was 2 We conducted additional analyses looking at the students’ cultural self‐identification deemed to be the most relevant control variable reflecting cultural capital in the family. (with the cultural majority vs. with a cultural minority) as a moderator. However, cultural However, future studies should take into account that more and more people are reading self‐identification did not moderate any of the relations in our models. books online (which was less the case in 2015/2016 when our data were collected).  | 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 328 SCHWARZENTHAL et al. range age = 11–18 years) completed the survey. Of the participat‐ (2015) and a study by Sieck et al. (2013), but also drawing on the ing students, 49% were of immigrant background, meaning that data at hand to develop appropriate coding categories (for the full at least one parent had immigrated to Germany. Most of these coding manual, see Schwarzenthal et al., 2019). Two independent students (86%) were born in Germany. They represented 63 her‐ researchers coded the answers. The students’ situational judgment itage countries, with the largest group from Turkey. The propor‐ was coded with regard to consideration of cultural influences (3‐ tion of students of immigrant background in the classroom ranged point‐scale ranging from low to high) and suspension of judg‐ between 13% and 89%. All students attended the most compre‐ ment (3‐point‐scale ranging from low to high). Their response hensive school type in North Rhine‐Westphalia (the Gesamtschule) judgment was coded with regard to interculturally competent be‐ that offers various school leaving certificates. havior (5‐point‐scale ranging from not at all effective to very effective). Please note that there was not only “one” right response, but that instead any response by the students that reflected inte‐ 2.2 | Measures gration of different cultural interests was rated as interculturally competent (e.g., if the students actively mediated between dif‐ 2.2.1 | Cultural intelligence ferent parties, or if they proposed a solution that fit both parties’ We used a CQ measure that was specifically developed for adoles‐ interests). Disagreements between coders were resolved via con‐ cents in culturally diverse societies, and that contains a self‐report sensus. Two‐way‐random intraclass‐correlations (Shrout & Fleiss, questionnaire as well as SJTs (Schwarzenthal, Juang, Schachner, & 1979) confirmed high intercoder reliability (from.86 to.96). van de Vijver, 2019) 2.2.2 | Cultural diversity climate Self-reported CQ The original self‐report CQ scale (Ang et al., 2007; Van Dyne et al., The response scale for both subscales of cultural diversity climate 2012) contains four subscales assessing motivational, cognitive, ranged from (1) no, that's not right to (5) yes, that's right. For both metacognitive, and behavioral CQ, and was related to a range of subscales, we included students’ individual perceptions, as well as positive outcomes, such as cultural adaptation and cross‐cultural the classroom‐level aggregates of these perceptions in our analyses. negotiation effectiveness, in previous research (for reviews, see Ott Since students in Germany spend most of their time at school in one & Michailova, 2018; Sharma & Hussain, 2017). Maintaining the sub‐ classroom, we chose to aggregate at the classroom level, and not at scales of the original scale, items were developed that were under‐ the school level. standable and relevant for adolescents growing up in multicultural contexts. CQ motivation (e.g., “It's fun for me to interact with people Contact and cooperation from other cultures”, α =.88), CQ cognition (e.g., “I can describe what Based on Beaton, Dovidio, and Léger (2008), four self‐developed is expected of men and women in various cultures”, α =.88), CQ meta‐ items measured positive interactions between students of different cognition (e.g., “If I don't understand the behavior of people from an‐ cultural backgrounds in the classroom, such as “Students from my other culture, I try to find out why they might have acted the way cultural group have good relationships with students from other cul‐ they did”, α =.86), and CQ behavior (e.g., “When I talk to people from tural groups” (α =.75). Since this measure was used for the first time another culture, I am considerate regarding their traditions and ways in this study, we ran a CFA in MPlus (Muthén & Muthén, 1998–2011) of living”, α =.82) were measured with six items each. The response to confirm the one‐dimensional factor structure. The CFA revealed scale ranged from (1) no, that's not right to (5) yes, that's right. Items an adequate fit of the one‐factorial model (with one correlated error from each subscale were averaged to represent the four CQ facets. because of similar item wording), χ2/df (N = 591) = 5.22, p =.02, RMSEA =.09, CFI =.99, SRMR =.02. Situational judgment tests In order to assess intercultural competence a multimodal assessment Multiculturalism is desirable (Deardorff, 2011). Therefore, Schwarzenthal et al. (2019) Multiculturalism was assessed with the cultural pluralism scale de‐ also developed SJTs for adolescents in culturally diverse societies, veloped by Schachner et al. (2016), which contains three subscales : based on the example set by Rockstuhl et al. (2015). The scores that (a) the perceived interest of students and teachers in students’ cul‐ participants received in the original SJTs predicted task performance tural background (six items; e.g., “At school students are interested and organizational citizenship behavior in culturally diverse teams in how people from different cultures and countries live”); (b) learn‐ (Rockstuhl et al., 2015). Two SJTs were developed that were situ‐ ing about intercultural relations (three items; e.g., “At school we learn ated in the peer (SJTa) and school (SJTb) context. Brief descriptions to respect each other even though we are from different cultures”); of intercultural situations were followed by questions assessing situ‐ and (c) learning about multicultural topics (five items; e.g., “At school ational judgment (“Why does person A/B behave this way?”) as well we talk about the celebrations and traditions of other cultures”). The as response judgment (“What would you do next in this situation?”). response scale ranged from (1) no, that's not right to (5) yes, that's Three intercultural researchers developed a coding manual right. The three subscales were averaged to create an overall score (following Syed & Nelson, 2015), partly based on Rockstuhl et al. for multiculturalism. Cronbach's α was.92.  | 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License DIVERSITY CLIMATE & INTERCULTURAL COMPETENCE 329 2.2.3 | Control variables 2.3.4 | Correlations Since the students in Study 1 were between 11 and 18 years old, and As a first step, we estimated correlations between the predictor vari‐ age is related to CQ (Schwarzenthal et al., 2017), we included age as ables and the self‐report CQ subscales and SJT ratings separately for an additional control variable in Study 1. students of immigrant and non‐immigrant background (see Table 1). Contact and cooperation and multiculturalism were positively re‐ 2.3 | Results lated to the self‐reported CQ subscales in both groups, and to some of the SJT ratings among students of non‐immigrant background. 2.3.1 | Preliminary analyses We first conducted preliminary analyses. In total, 3% of data were 2.3.5 | Multilevel SEM missing from the variables that were used for analyses. We used To test our hypotheses, we ran two separate models for the self‐re‐ full information maximum likelihood (FIML) in Mplus 7.3 (Muthén & port CQ factor and the SJT factor.4 We built the models step by Muthén, 1998–2011), which is considered best practice for dealing step. All continuous predictors that were strictly at the individual with missing data (Enders, 2010). level were grand‐mean centered prior to the analyses. Correlations between predictors were allowed. First, control variables (age, gen‐ 2.3.2 | Measurement models at the individual level der, immigrant background, grade average, and number of books in As a first step, we specified the measurement models at the individ‐ the household) were introduced as predictors at the individual‐level. ual level. The latent CQ variable was measured by the four means of Students of immigrant background (compared to students of non‐ the self‐report CQ motivation, cognition, metacognition, and behav‐ immigrant background) and students with more books reported ior subscales, and the latent SJT variable by the six ratings that the higher CQ. Older students, females, and students with higher grade students received for their responses in the two SJTs. Both meas‐ average showed higher scores in the SJTs. The model fit was good urement models fit well (see Appendix 1). Tests of measurement for the self‐reported CQ model, χ2/df(N = 631) = 2.49, p <.001, equivalence between students of immigrant and non‐immigrant RMSEA =.05, CFI =.97, SRMRwithin =.03, AIC = 12,812.95, and ad‐ background confirmed metric invariance (i.e., equivalence of factor equate for the SJT model, χ2/df(N = 631) = 2.47, p <.001, loadings) for the self‐report CQ factor model and scalar invariance RMSEA =.05, CFI =.91, SRMRwithin =.04, AIC = 15,187.82. (i.e., equivalence of factor loadings and intercepts) for the SJT factor We then introduced individually perceived contact and coop‐ model (see Appendix 2), implying that for the CQ factor, only rela‐ eration and multiculturalism in the classroom as predictors at the tions between constructs, not means, should be compared between individual level (testing Hypotheses 1 and 3). Both positively pre‐ students of immigrant and non‐immigrant background. dicted self‐reported CQ. Only perceived multiculturalism positively predicted SJT scores. Finally, we introduced interaction terms one by one to test whether 2.3.3 | ICCs and measurement models at the the associations between the individually perceived classroom cultural classroom level diversity climate and the outcomes were stronger among students of As a next step, we estimated intraclass correlations (ICCs) for all non‐immigrant background (testing Hypothesis 4). Multiculturalism study variables (see Appendix 3). The proportion of total variance was less strongly related to the SJT factor among students of immigrant explained by student membership in different classrooms (i.e., ICC1) background (β = −.14, p =.03). None of the other interaction effects were of the outcome variables was mostly lower than.05, and the reli‐ significant. The fit of the final models was good for the self‐report CQ ability of the observed classroom aggregate (ICC2, as in Marsh et al., outcome and adequate for the SJT outcome. Standardized coefficients 2012) was low. Thus, multilevel modeling procedures may provide and model fit indices of the final models are presented in Figures 1 and few benefits and might run into estimation problems (Dyer, Hanges, 2. Overall, the models explained 38% of variance in the latent self‐re‐ & Hall, 2005). Specification of measurement models for the latent port CQ variable, and 56% of variance in the latent SJT variable. self‐report CQ and SJT factors at the classroom level resulted in bad model fit, indicating that these constructs could not be reliably 2.4 | Discussion modeled at the classroom level (see Appendix 1). Exploratory analy‐ ses revealed that none of the classroom level predictors significantly In line with Hypotheses 1 and 3, students who perceived a predicted any of the outcome variables. Therefore, we proceeded stronger classroom climate fostering contact and cooperation by only investigating relations at the individual level. However, since between students of diverse cultural backgrounds, or a stronger even very low ICCs of.05 or.01 can lead to a significant distor‐ climate fostering multiculturalism, reported higher CQ. However, tion of the results of significance tests in conventional regression only a perceived climate of multiculturalism, but not of contact analyses (Geiser, 2011; Hox, 2010), we used the two‐level option in Mplus 7.3 to take the clustered structure of the data into account 4 Additional analyses were run with both outcome variables in one model. However, this even though no predictors were introduced at the classroom level. did not alter the results.  330 TA B L E 1 Study 1 individual level bivariate correlations and descriptives | 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 1. Age – −.08 −.12* −.13* −.04.07.01.09† −.07 −.01.30***.24***.12*.09.17**.03 * * 2. Gender (0 = male, −.03 – −.01 −.02 −.04.08.11 −.02 −.00.01.08 −.05.13.04.07.02 1 = female) 3. Grade average −.17**.03 –.27***.13*.08.09.07.06.13*.05.12*.12*.02.02.20** 4. No. of books in −.11†.01.18** –.11*.04.14*.10†.04.02.06.06 −.01 −.01 −.08.09 the household 5. Contact and.01.02.21***.05 –.24***.35***.13†.17**.28***.12*.05.12†.04 −.05.18** cooperation 6. Multiculturalism.03.07.00 −.02.25*** –.44***.36***.30***.40***.12*.15**.20**.11†.12*.24*** 7. Self‐reported CQ.09†.17**.08.07.27***.34*** –.36***.49***.59***.23***.11*.29***.07 −.06.22*** motivation 8. Self‐reported CQ.10†.07 −.03.10.14*.31***.48*** –.42***.40***.18**.14*.06.06.06.18** cognition 9. Self‐reported CQ −.00.10†.08.07.16*.36***.41***.44*** –.70***.12*.11*.27***.06 −.03.23*** metacogn. 10. Self‐reported.05.11*.11†.11†.23**.37***.54***.50***.69*** –.12*.11*.28***.07.01.27*** CQ behavior 11. SJTa suspending.19** −.02.07 −.07.15*.01.02.05 −.05 −.03 –.48***.26***.09.03.10 judgment 12. SJTa considering.35***.01.03.11†.06 −.03.03.06.00 −.00.48*** –.28***.02.05.21*** cult. infl. 13. SJTa behavior.12*.10†.03.07.08.09.16**.06.12†.19**.12*.19*** –.07.12*.26*** 14. SJTb suspending.17**.05.14*.09 −.01.07.02.05 −.01.05.10.06.10† –.02.15** judgment 15. SJTb consider‐.24***.13*.05.01 −.05.04.13*.07 −.02.03.01.16**.21***.13* –.11† ing cult. infl. 16. SJTb behavior.07.14*.11.12†.07.07.12*.10.11†.19**.09.13*.20***.18**.44*** – Students of non‐immigrant background M (SD) 13.53 0.48 4.40 3.44 4.02 3.51 3.73 2.86 3.27 3.51 2.28 1.90 3.41 2.09 1.68 3.03 (1.82) (0.5) (0.77) (1.21) (0.71) (0.72) (0.83) (0.86) (0.9) (0.84) (0.75) (0.73) (0.9) (0.47) (0.77) (1.17) Students of immigrant background M (SD) 13.86 0.50 4.18 2.75 3.96 3.60 3.95 3.24 3.46 3.72 2.24 1.87 3.22 2.05 1.70 2.99 (1.83) (0.5) (0.76) (1.27) (0.67) (0.70) (0.73) (0.81) (0.85) (0.73) (0.76) (0.74) (0.93) (0.5) (0.81) (1.22) Note: N = 324 students of non‐immigrant background and N = 307 students of immigrant background. Correlations for students of non‐immigrant background above the diagonal, correlations for stu‐ dents of immigrant background below the diagonal. † p <.10, *p <.05, **p <.01, ***p <.001. SCHWARZENTHAL et al. 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License  | 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License DIVERSITY CLIMATE & INTERCULTURAL COMPETENCE 331 F I G U R E 1 Study 1: Model with CQ as outcome variable, standardized coefficients F I G U R E 2 Study 1: Model with SJT as outcome variable, standardized coefficients and cooperation, was related to students’ scores in the SJTs. the fact that students in one classroom usually perceive classroom Promoting contact and cooperation can sometimes go along climate quite differently (e.g., Schenke, Ruzek, Lam, Karabenick, & with pursuing a color‐evasive approach (Schachner et al., 2016) Eccles, 2017), and that these idiosyncratic experiences are more and neglecting opportunities to learn about cultural variations. In closely linked to learning outcomes than abstract classroom‐level ag‐ order to explore this possibility further, we explicitly distinguished gregates. However, it may also be a result of the rather small number between a classroom climate fostering contact and cooperation, of classrooms analyzed for this study (N = 29) and the fact that these and a classroom climate fostering color‐evasion in Study 2. were all from a single school type (the Gesamtschule), and thus may Not supporting Hypothesis 4, relations between classroom cul‐ not vary much with regard to their cultural diversity climate and the tural diversity climate and students’ CQ in Study 1 were largely sim‐ students’ CQ. We explored this possibility by analyzing a larger num‐ ilar between students of immigrant and non‐immigrant background. ber of classrooms from different school types in Study 2. Only the positive link between multiculturalism and performance in Conceptions of intercultural competence are sometimes criticized the SJTs was stronger among students of non‐immigrant background. for encouraging preconceived attributions to culture, which can also Due to low variance at the classroom level, the outcome variables foster stereotypes (Barrett et al., 2013). In order to address this cri‐ in this study could not be reliably modeled at this level and were tique, we introduce an additional coding category for the SJTs in Study unrelated to the classroom‐aggregated climate. This may be due to 2. Previous research has found that individuals with more intercultural  | 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 332 SCHWARZENTHAL et al. experience are not only more likely to consider cultural influences as scores for consideration of cultural influences, suspension of judg‐ explanations for behavior, but also to consider multiple alternative ment, and behavior. Two‐way‐random intraclass‐correlations (Shrout explanations for behavior, thereby reducing the likelihood of ste‐ & Fleiss, 1979) confirmed high intercoder reliability (from.76 to.93). reotyped interpretations of cultural differences (Sieck et al., 2013). Moreover, it was coded whether students considered multiple pos‐ Therefore, in Study 2, we will also code whether students consider sible explanations when trying to make sense of the behavior in the alternative explanations when interpreting the behavior in the SJTs. SJTs, in order to avoid stereotyped explanations of cultural differ‐ ences. The scale ranged from (1) student only considered one possible explanation for the behavior of the parties in the situation to (3) student 3 | S T U DY 2 considered alternative explanations for the behavior of both parties in the situation. Intercoder reliability was high (.82 and.83, respectively). 3.1 | Method 3.1.1 | Participants and procedure 3.2.2 | Cultural diversity climate We obtained permission from the Berlin Senate Committee for In order to assess cultural diversity approaches, we used subscales from Education, Youth, and Science and from 17 principals to conduct the the revised version of the Cultural Diversity Climate Scale (Schachner study in their schools. Following the guidelines of the Berlin Senate, pa‐ et al., 2019). The response scales ranged from (1) no, that's not right rental consent was only required for questions asking about the partici‐ to (5) yes, that's right. Validity and measurement equivalence between pants’ parents. After obtaining parental consent, and students’ assent students of immigrant and non‐immigrant background were confirmed or consent, we administered a questionnaire in 66 9th grade classrooms by Schachner et al. (2019) with the same sample that was used in the in 2016. The final sample comprised 1,335 adolescents (48% female, present research. We included both individual perceptions of these ap‐ Mage = 14.69 years, SDage = 0.74). Of these students, 52% were of im‐ proaches as well as classroom‐level aggregates in our analyses. migrant background, 37% of non‐immigrant background, and 11% did not receive permission from their parents to answer questions related Contact and cooperation to their parents’ place of birth. Most of the students of immigrant back‐ Contact and cooperation between students of diverse cultural back‐ ground (74%) were born in Germany. They represented 80 different grounds was assessed with three items (e.g., “Students in my class from heritage countries, with the largest group coming from Turkey. The pro‐ different heritage cultures get along well with one another”, α =.91). portion of students of immigrant background in a classroom ranged be‐ tween 9% and 100%. The students either attended a Gymnasium (32% Color-evasion of the sample), which is the academic track, or an Integrated Secondary Norms fostering a color‐evasive perspective were assessed with five School (68% of the sample), which combines the former vocational and items (e.g., “In class we learn that people of different backgrounds comprehensive school types, and offers all school leaving certificates.5 are all the same at heart”; α =.90). Multiculturalism 3.2 | Measures For multiculturalism, we used the “heritage and intercultural learn‐ ing” learning subscale, containing seven items (e.g., “During class we 3.2.1 | Cultural intelligence learn about the heritage cultures of fellow students”; α =.90). As in Study 1, CQ was assessed with an adapted version of the self‐report CQ questionnaire (Van Dyne et al., 2012), as well as 3.2.3 | Control variables with two SJTs (for a more detailed description of the measures, see Schwarzenthal et al., 2019). In order to ensure that the results from As the students in Study 2 attended two different school tracks, and Study 1 are not limited to the specific situations depicted in the two school track is related to intergroup outcomes (Schwarzenthal et al., SJTs, one SJT used for Study 2 was the same as in Study 1 (SJTb), 2018), we included school track (0 = Integrated Secondary School, and one was different (SJTc)—the new SJT focused specifically on the 1 = Gymnasium) as an additional control variable in Study 2. arrival of a refugee adolescent in school (this situation was chosen to acknowledge the high number of refugee students, largely from 3.3 | Results Syria, who recently entered the German school system). Cronbach's α of the CQ subscales ranged between.86 and.90. For the SJTs, 3.3.1 | Preliminary analyses the same coding procedure as in Study 1 was followed to obtain In total, 4% of data were missing on the variables that were used for 5 Each federal state in Germany has its own school system and its own school types. The analyses. Two students were excluded from analyses because clus‐ “Integrated Secondary School” in Berlin is not equivalent to the “Gesamtschule” in NRW. ter membership was unknown or because the student had missings The “Integrated Secondary School” combines the former school types of Haupt‐, Real‐ and Gesamtschule, while the “Gesamtschule” is an inclusive school form that exists on all outcome variables. As in Study 1, we employed multilevel SEM parallel to the other school forms of Haupt‐, Real‐, and Gesamtschule. with FIML to test our hypotheses.  | 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License DIVERSITY CLIMATE & INTERCULTURAL COMPETENCE 333 level, we introduced predictors step by step. First, control variables 3.3.2 | Measurement models at the individual level (gender, immigrant background, grade average, and number of The latent self‐report CQ variable was measured as in Study 1. Due books) were introduced as predictors at the individual level. Females to the introduction of an additional coding category, the latent SJT (compared to males), students of immigrant background (compared variable was now measured by eight ratings that the students re‐ to students of non‐immigrant background), students with higher ceived based on their responses to the SJTs. Both measurement grades, and students with more books in the household reported models fit well at the individual level (see Appendix 1). Tests of higher CQ. Female gender, grade average, and number of books measurement equivalence between students of immigrant and non‐ in the household were related to higher scores in the SJTs, while immigrant background confirmed metric invariance for both mod‐ immigrant background was related to lower scores (however, due to els. The model fit decreased slightly for the scalar invariance models lack of scalar measurement invariance, this mean difference should where intercepts were set to be equal across groups, implying that be treated with caution). Model fit was good for the self‐report CQ comparisons of means between the two groups should be treated model, χ2/df(N = 1,333) = 2.68, p <.001, RMSEA =.04, CFI =.98, with caution (see Appendix 2). TLI =.96, SRMRwithin =.02, SRMRbetweenv =.15, AIC = 22,829.54, as well as for the SJT model, χ2/df(N = 1,333) = 1.66, p <.001, RMSEA =.02, CFI =.96, TLI = 0.95, SRMRwithin =.03, SRMRbetween =.19, 3.3.3 | ICCs and measurement models at the AIC = 29,579.95. classroom level As a next step, we introduced the subscales of perceived ICC1s were all ≥.05 and thus higher than in Study 1. The reliabili‐ classroom cultural diversity climate as predictors at the individual ties of the classroom aggregates (ICC2) were all higher than.51 (see level (testing Hypotheses 1–3). Contact and cooperation as well Appendix 3). When the measurement models of the CQ and SJT as color‐evasion positively predicted both self‐reported CQ and variables were specified at the classroom level, the model fit was performance in the SJTs, while multiculturalism only positively good (see Appendix 1). Thus, we conducted multilevel CFAs and predicted self‐reported CQ. Since we were also interested in dif‐ tested whether factor loadings were invariant at the individual and ferential relations between the predictors and the CQ subscales, classroom level in order to verify if the structure of the constructs we inspected modification indices. These indicated additional was the same across levels (following Marsh et al., 2012). Since this direct positive relations between contact and cooperation and did not lead to a substantial decrease in model fit (see Appendix 1), CQ motivation and between multiculturalism and CQ metacogni‐ factor loadings were constrained to be invariant across levels in all tion, as well as a direct negative relation between multicultural‐ following models. Standardized factor loadings were all significant, ism and CQ motivation. Adding these relations improved model and, as is typically the case, were high at the classroom level (Marsh fit of the self‐report CQ model (before: χ2/df(N = 1,333) = 4.57, et al., 2012). p <.001, RMSEA =.05, CFI =.95, TLI =.90, SRMR within =.03, SRMR between =.17, AIC = 32,530.82, after: χ2/df(N = 1,333) = 1.90, p =.002, RMSEA =.03, CFI =.99, TLI =.98, SRMR within =.02, 3.3.4 | Correlations SRMR between =.17, AIC = 32,451.51). Since the measurement models fit well at both levels, we pro‐ We then tested one by one whether immigrant background ceeded by calculating correlations at the individual and classroom moderated the associations between individually perceived level (see Tables 2 and 3). At the individual level, perceptions of all classroom cultural diversity climate and the outcomes (testing three types of classroom cultural diversity climate were positively Hypothesis 4). A perceived climate of contact and cooperation was related to both the self‐report CQ subscales as well as to several more strongly related to CQ among students of non‐immigrant of the SJT ratings. At the classroom level, the three types of class‐ background (β = −.11, p =.04). None of the other interaction ef‐ room diversity climate were positively related to the self‐report fects were significant. Model fit was good for the model with the CQ subscales. Moreover, contact and cooperation was positively self‐report CQ factor as outcome variable, χ2/df(N = 1,333) = 4.92, related to the SJT ratings. p <.001, RMSEA =.05, CFI =.95, TLI =.90, SRMR within =.06, SRMRwithin =.17, AIC = 34,046.33, as well as for the model with the SJT factor as outcome variable, χ2/df(N = 1,333) = 1.60, 3.3.5 | Multilevel SEM p <.001, RMSEA =.02, CFI =.97, TLI =.96, SRMRwithin =.03, Multilevel random‐intercept models using the MLR estimator in SRMR between =.22, AIC = 39,416.77. Overall, the models explained Mplus 7.3 (Muthén & Muthén, 1998–2011) were run to test our 30% of variance in the latent self‐report CQ factor, and 31% of the hypotheses. All continuous predictors that were strictly at the variance in the latent SJT factor at the individual level. individual level were grand‐mean centered prior to the analyses. We proceeded by adding school track and proportion of stu‐ Correlations between predictors were allowed. As in Study 1, we ran dents of immigrant background in a class as control variables at the separate analyses with the latent self‐report CQ factor as outcome classroom level. Students attending Gymnasium (as compared to variable and with the latent SJT factor as outcome variable. After students attending the Integrated Secondary School) reported mar‐ specifying the measurement models at the individual and classroom ginally higher CQ and scored higher in the SJTs. The proportion of  TA B L E 2 Study 2 individual level bivariate correlations and descriptives |334 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 1. Gender –.24***.06.20***.10*.03.24****.14**.11*.17***.19***.21***.22***.28***.18***.21***.19***.22*** (0 = male, 1 = female) 2. Grade average.05 –.33***.26***.20***.12*.23***.23***.18***.19***.13*.19***.24***.19**.04.12*.17***.18*** 3. No. of books in −.03.32*** –.21***.12*.03.30***.28***.19***.22***.08†.02.11*.16***.03.08†.14**.20*** the household 4. Contact and.13**.20***.19*** –.40***.24***.40***.21***.27***.34***.13*.16**.15**.22***.09*.11*.21***.25*** cooperation 5. Color‐evasion.01.14**.19***.38*** –.50***.26***.17**.22***.29***.03.05.11*.12*.02.05.13***.16** 6..04.10*.12**.28***.52*** –.16*.23***.32***.29***.01.01.08†.04.02.03.11**.04 Multiculturalism 7. Self‐reported.15***.16**.21***.33***.32***.20*** –.42***.34***.49***.15**.12*.20***.33***.12**.12**.20***.43*** CQ motivation 8. Self‐reported.10**.16***.16***.17***.23***.24***.41*** –.37***.41***.07.07†.14**.15**.01.11**.08†.24*** CQ cognition 9. Self‐reported.08†.09*.07†.16**.26***.35***.34***.42*** –.63***.06.08†.15***.23***.06.13**.15**.22*** CQ metacogn. 10. Self‐reported.13**.14**.09*.17***.26***.27***.46***.47***.65*** –.15**.06.16***.28***.09*.13**.17***.31*** CQ behavior 11. SJTb.10**.10*.06.09†.07.02.13**.06.09*.11** –.31***.27***.25***.02.06.17***.28*** suspending judgment 12. SJTb consid‐.06.05.16***.12*.11* −.01.18***.06.02.06.29*** –.21***.25***.07.14**.09*.16*** ering cult. infl. 13. SJTb con‐.06.03.11*.06.11* −.01.04.04.01 −.00.23***.31*** –.25***.11*.23***.32***.25*** sidering altern. expl. 14. SJTb.16**.21***.20***.15***.19***.12*.24***.14**.14***.22***.25***.24***.18*** –.14**.26***.21***.31*** behavior 15. SJTc.02 −.03.08†.10*.07†.02.05.05.03 −.00.02.08 −.02.05 –.34***.10*.11* suspending judgment 16. SJTc consid‐.02.10*.11**.09*.02 −.01.09†.06.00.07*.10*.19***.06.14**.23*** –.46***.18*** ering cult. infl. 17. SJTc consid‐.07†.10*.11*.10*.10*.09†.06.02 −.03.04.07.16**.23***.14**.11**.30*** –.21*** ering altern. expl. SCHWARZENTHAL et al. (Continues) 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License  | 10990992, 2020, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ejsp.2617, Wiley Online Library on [24/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License DIVERSITY CLIMATE & INTERCULTURAL COMPETENCE 335 students of immigrant background in a class was unrelated to the (0.83) (0.91) outcome variables. Model fit was good for the self‐report CQ model, Note: N = 501 students of non‐immigrant background and N = 689 students of immigrant background, N = 145 students were excluded from this analysis because immigrant background was unknown. 3.22 3.31 18. – χ2/df(N = 1,333) = 4.86, p <.001, RMSEA =.05, CFI =.95, TLI =.90, SRMRwithin =.06, SRMRbetween =.17, AIC = 34,117.93, as well as for the (0.64) ** (0.74) 1.68 SJT model, χ2/df(N = 1,333) = 1.63, p <.001, RMSEA =.02, CFI =.97, 1.90.13 17. TLI =.96, SRMRwithin =.03, SRMRbetween =.22, AIC = 39,476.67. When all three subscales of classroom cultural diversity cli‐ *** (0.73) 2.08 (0.7) 1.94.16 16. mate were introduced as predictors at the classroom level (testing Hypotheses 1–3 at the classroom level), none of them were related (0.64) (0.66) to the outcome variables. Since the subscales of classroom cultural * 1.88.09 1.76 15. diversity climate were highly correlated at this level, the results may have been affected by multicollinearity. Therefore, we ran separate *** (1.22) (1.13) 3.04 3.22.31 analyses in which only one subscale of classroom cultural diversity 14. climate was introduced as a predictor at a time. In these analyses, color‐evasion and multiculturalism positively predicted self‐reported (0.53) (0.59) * 1.38 1.27.10 Correlations for students of non‐immigrant background above the diagonal, correlations for students of immigrant background below the diagonal. 13. CQ, and contact and cooperation positively predicted SJT. Color‐eva‐ sion and multiculturalism marginally positively predicted SJT. In order *** (0.82) (0.75)