Can School Diversity Policies Reduce Belonging and Achievement Gaps Between Minority and Majority Youth? Multiculturalism,Colorblindness,and Assimilationism Assessed

Abstract

European societies and schools face the challenge of accommodating immigrant minorities from increasingly diverse cultural backgrounds. In view of significant belonging and achievement gaps between minority and majority groups in school, we examine which diversity approaches are communicated by actual school policies and which approaches predict smaller ethnic gaps in student outcomes over time. To derive diversity approaches, we content-analyzed diversity policies from (n = 66) randomly sampled Belgian middle schools. Cluster analysis yielded different approaches valuing, ignoring, or rejecting cultural diversity in line with multiculturalism, colorblindness, and assimilationism, respectively. We estimated multilevel path models that longitudinally related diversity approaches to (N = 1,747) minority and (N = 1,384) majority students’ school belonging and achievement (self-reported grades) 1 year later. Multiculturalism predicted smaller belonging and achievement gaps over time; colorblindness and assimilationism were related to wider achievement and belonging gaps, respectively. Longitudinal effects of colorblindness on achievement were mediated by (less) prior school belonging.

Keywords

diversity policies multiculturalism colorblindness assimilationism achievement

In North America and Europe, immigrant minorities often lag behind their majority peers in school (Organisation for Economic Co-operation and Development, 2015), with far-reaching consequences for their psychological development and future life chances (Motti-Stefanidi & Masten, 2013). The overlap of migration-related diversity with persistent and often dramatic educational inequalities poses a major challenge to schools and societies at large. Compared with their majority peers, disadvantaged minority students show lower academic achievement (Dimitrova, Chasiotis, & van de Vijver, 2016). Moreover, their school belonging is less stable and more contingent on a supportive environment (Cook, Purdie-Vaughns, Garcia, & Cohen, 2012). Lack of belonging is not only a critical developmental risk factor in minority youth (Eccles, Wong, & Peck, 2006), but school belonging is also a key process connecting the school environment to individual achievement outcomes (Roeser, Midgley, & Urdan, 1996). The present study examines belonging and achievement gaps between minority and majority adolescents in the same schools—with an explanatory focus on the school environment.

As migration-related diversity is on the rise in schools and workplaces, organizations can dampen or perpetuate related inequalities, depending on how they handle diversity (Banks, 2015; Stevens, Plaut, & Sanchez-Burks, 2008). In educational contexts, school policies specifically dealing with cultural differences communicate different diversity approaches, which value, ignore, or reject cultural differences (Guimond, de la Sablonnière, & Nugier, 2014). Against this background, the present study investigates (a) which diversity approaches are communicated by actual school policies in a European educational context and (b) how school diversity policies relate to educational inequalities and to minority belonging and achievement in particular. To this end, we draw on large-scale longitudinal surveys following over 3,000 minority and majority adolescents in 66 secondary schools in Flanders, Belgium (Children of Immigrants Longitudinal Study [CILS] Belgium, 2014).

There is some empirical evidence associating specific diversity approaches with school adjustment in minority youth (e.g., Hoti, Heinzmann, Muller, & Buholzer, 2017; Schachner, Noack, van de Vijver, & Eckstein, 2016). One limitation of previous studies is their reliance on student or teacher perceptions of the prevailing diversity climate. Congruence between perceived school diversity norms by majority and especially minority students (Civitillo et al., 2017) and teachers (Fine-Davies & Faas, 2014) is generally low. Therefore, rather than diversity perceptions of students or teachers, we analyzed the actual contents of school policy documents (rules and mission statements) to assess how schools deal with diversity. To test the outcomes related to different diversity policies across schools, we applied multilevel modeling of school climate effects (Marsh et al., 2012).

Another limitation of most previous studies is their main focus on the school adjustment of minority youth (Aronson & Laughter, 2016). In the absence of direct achievement measures and without majority comparison samples, we cannot know whether and how school diversity approaches predict educational inequalities between minority and majority groups, nor do we know how they affect majority students as distinct from their minority peers. It has been suggested that valuing cultural diversity can be costly on majorities who may feel excluded or alienated (Stevens et al., 2008). Furthermore, value in diversity might boost minorities’ belonging in school, yet fail to reduce real ethnic inequality in their school achievement. To assess educational inequality, we estimated the gap between minority and majority students’ academic outcomes. As a measure of academic achievement, we made use of their self-reported Dutch language grades. Dutch language grades are very important in the school careers of Flemish-Belgian students because educational policies have a strong focus on Dutch proficiency as essential for academic success (Pulinx, Van Avermaet, & Agirdag, 2017). Next, we tested whether school diversity policies at the beginning of middle school predicted the achievement gap up to 1 year later. To elucidate the processes connecting unequal achievements to the school environment, we also tested related gaps in the school belonging of minority and majority youth.

Finally, most research on organizational diversity approaches is informed by North American contexts of migration and race relations (Plaut, Thomas, Kyneshawau, & Romano, 2018). In European societies, however, different vocabularies reflect distinct histories and understandings of migration-related diversity (Guimond et al., 2014). Given such differences, the prevalence and the specific contents of diversity approaches vary across societal contexts. Moreover, schools are local institutional contexts with distinct diversity patterns and norms (Celeste, Meeussen, Verschueren, & Phalet, 2016). We do not know to what extent school contexts reflect societal diversity approaches or what specifically constitutes diversity policies in schools. Rather than imposing general diversity approaches on these school policies, therefore, we used a mixed-methods design: Thematic content coding of actual diversity policies was combined with subsequent cluster analysis to optimally capture the meaning of different diversity approaches (Namey, Guest, Thairu, & Johnson, 2008).

Diversity Approaches: Multiculturalism, Colorblindness, and Assimilationism

North American research has contrasted two main approaches: multiculturalism and colorblindness. While both approaches seek to include minorities, multiculturalism embraces cultural diversity as added value and colorblindness ignores diversity, emphasizing instead individual merits or equal treatment (Plaut et al., 2018). Assimilationism represents a third distinct approach that prioritizes majority culture adoption (Guimond et al., 2014). Below, we discuss multiculturalism, colorblindness, and assimilationism as three commonly distinguished diversity approaches.

Multiculturalism policies acknowledge and value cultural diversity (Rosenthal & Levy, 2010). From a social identity perspective, a multicultural approach seeks to include minorities through affirming their distinct cultural identities (Derks, Van Laar, & Ellemers, 2007; Dovidio, Gaertner, & Saguy, 2007). In practice, multiculturalism includes school policies such as designated “diversity days” to learn about each other’s cultural heritage and educational practices aimed to improve intercultural understanding (Verkuyten & Thijs, 2013). Although multicultural education may contain other elements such as combating racism and developing caring relations between teachers and students (Zirkel, 2008), the main focus of a multiculturalist approach is on value in diversity (Apfelbaum, Stephens, & Reagans, 2016; Civitillo et al., 2017; Schachner et al., 2016).

In contrast, colorblindness tends to ignore cultural diversity. It draws on Enlightenment individualism, valuing uniqueness along with meritocratic ideals and equal treatment regardless of one’s cultural background (Rosenthal & Levy, 2010; Stevens et al., 2008). In line with value in individuality, colorblindness seeks to neutralize prejudice and discrimination by de-emphasizing group categories and ignoring group differences (Park & Judd, 2005). Recent comparative research highlights different meanings of colorblindness across intergroup contexts, however, depending on which component of the approach is foregrounded in the context (Guimond et al., 2014). Thus, stressing individual uniqueness versus ensuring equal treatment may represent conflicting ideas depending on the context (Apfelbaum et al., 2016). Along those lines, a distinct egalitarian approach of diversity in German schools combined equal treatment with protection from discrimination (Civitillo et al., 2017; Schachner et al., 2016). Similarly in France, an egalitarian approach labeled “colorblind equality” related to decreased anti-immigrant prejudice in secondary schools (Roebroeck & Guimond, 2015). It is a matter of ongoing debate when an egalitarian approach may or may not overlap with colorblindness (Apfelbaum et al., 2016; Schachner et al., 2016).

Assimilationism has been put forward as a distinct approach that rejects cultural diversity in Europe (Civitillo et al., 2017; Fine-Davies & Faas, 2014; Guimond et al., 2014) and in the United States (Levin et al., 2012). The approach is akin to (American-style) colorblindness to the extent that both approaches de-emphasize cultural diversity (Rosenthal & Levy, 2010). Under assimilationism, minorities are expected to adopt the majority culture and to relinquish their distinct minority cultures or any markers of it (Hornsey & Hogg, 2000). The aim is to foster a common identity for both majorities and minorities, like the melting pot idea of a mainstream culture in the United States. Conceptually, assimilationism has been associated with the common ingroup identity model, which represents a majority group perspective on prejudice reduction (Dovidio et al., 2007). However, contrary to expectations from common ingroup identity, assimilationism was revealed as a hierarchy-enhancing ideology. Thus, assimilationist majority attitudes were related to a stronger social dominance orientation and predicted more prejudice against immigrants in the United States (Levin et al., 2012). For minorities, assimilationism entails conformity pressure to adopt majority cultural customs, norms, and values while relinquishing the heritage culture (Van Acker & Vanbeselaere, 2012).

Looking beyond core conceptual distinctions between general diversity approaches that value, ignore, or reject cultural differences, there is also contextual variation in the way these approaches are construed in specific societies and organizational settings. Rather than imposing the prevalent societal diversity approaches on the school context, we take a bottom-up approach. To clarify how schools approach diversity in a European educational context and to balance generic and context-specific meanings of each diversity approach, our study starts from a qualitative analysis of school diversity policies.

Relationship of Diversity Approaches to School Belonging and Achievement

Existing research on diversity approaches has mostly assessed individual diversity attitudes of majority group members in relation to aspects of intergroup relations such as bias, prejudice, and stereotyping (Guimond et al., 2014; Rattan & Ambady, 2013; Rosenthal & Levy, 2010). There is less research on actual diversity policies in organizational contexts and in relation to academic outcomes. We take a social identity perspective to explain why and how diversity approaches may affect academic gaps between minority and majority students. From a social identity perspective, culturally diverse schools as intergroup contexts make minority identities salient. When minority identities are valued in the context, minority students will feel more included and perform better (Derks et al., 2007). Conversely, when the context is identity-threatening, experiences of exclusion, discrimination, or negative stereotyping signal to minority students that their identities are devalued. In such contexts, they will feel less belonging (Walton & Cohen, 2007) and may underperform (Baysu, Celeste, Brown, Verschueren, & Phalet, 2016).

Along those lines, a multiculturalist diversity approach is expected to be related to smaller majority–minority gaps (Hypothesis 1) because it values cultural differences and thus affirms minority identities (Derks et al., 2007; Verkuyten, 2005). In support of this expectation, there is evidence of the benefits of a multiculturalist approach for minorities’ sense of belonging (Meeussen, Otten, & Phalet, 2014) and achievement (Aronson & Laughter, 2016). Similarly, diversity education interventions in schools increased minority motivation and performance (Celeste, Baysu, Brown, & Phalet, 2019; Hoti et al., 2017).

On the contrary, a colorblind approach is expected to predict a persistent majority–minority gap, to the extent that it ignores cultural differences and related disadvantage (Hypothesis 2). From a social identity perspective, colorblindness disregards a distinct cultural identity of minority students, making them feel like they do not belong (Meeussen et al., 2014). It can undermine their performance by turning a blind eye on minority disadvantage (Holoien & Shelton, 2012). When equality is part of the definition of colorblindness, however, positive effects were also found (Apfelbaum et al., 2016). Thus, student perceptions of equal treatment in school buffered identity threat and protected school achievement for minority students (Baysu et al., 2016). The latter findings are in line with evidence of minority inclusion through positive intergroup contact and cooperation (Schachner et al., 2016). As distinct individualism-focused versus equality-focused variants of colorblindness predict opposite outcomes, our bottom-up analysis of school policies explores the specific meanings of colorblindness in a European educational context.

Finally, an assimilationist approach is expected to predict a larger majority–minority gap because it rejects cultural differences and thus threatens minority identities (Hypothesis 3). Accordingly, assimilationism was related to majority prejudice (Van Acker & Vanbeselaere, 2012), and minority students experienced more peer rejection when classroom norms stressed assimilationism (Celeste et al., 2016). Similarly, when schools enforced assimilation by keeping minority students from speaking their native language, they left them feeling alienated (Pulinx et al., 2017).

As schools and organizations are not made up of only minorities, a distinctive contribution of this study is the comparison across minority and majority outcomes. Multiculturalism can be beneficial for majority outcomes to the extent that they feel included as in all-inclusive multiculturalism (Meeussen et al., 2014; Stevens et al., 2008). It may also backfire, however, when majorities feel excluded from this approach (Jansen, Otten, & Van der Zee, 2015; Plaut et al., 2018). We would not expect colorblind or assimilationist approaches to affect majority academic outcomes as majority identities are not threatened. Yet, these approaches might also benefit majority outcomes through affirming their majority identities. Hence, we have no specific hypotheses for majority outcomes.

In addition, we examined whether ethnic school composition moderated the consequences of diversity policies for majority–minority gaps. In today’s highly culturally diverse and segregated schools, immigrant minority students often make up a numerical majority in their classrooms (Baysu & De Valk, 2012; Schachner et al., 2016). Possibly, policies valuing cultural diversity are more relevant and effective in highly diverse schools, where students from many different cultural backgrounds interact on a daily basis, as compared with less diverse schools.

Finally, school belonging has been revealed as a psychological process that connects diversity approaches to school achievement (Cook et al., 2012; Inzlicht & Good, 2006). Belonging is longitudinally associated with sustained school achievement across majority and minority adolescents (Gillen-O’Neel & Fuligni, 2013). Moreover, a lack of belonging was shown to impair minority achievement (Walton & Cohen, 2007), while interventions supporting minority belonging improved their achievement (Shnabel, Purdie-Vaughns, Cook, Garcia, & Cohen, 2013). Furthermore, school belonging mediated between minority experiences of cultural diversity and school outcomes (Schachner, He, Heizmann, & van de Vijver, 2017). Similarly, workplace belonging mediated between diversity approaches of multiculturalism (for minorities) or colorblindness (for majorities) and workplace outcomes (Jansen, Vos, Otten, Podsiadlowski, & Van der Zee, 2016). Together, these findings consistently and causally relate diversity approaches to achievement through belonging. Our study replicates the mediating role of belonging in earlier studies longitudinally, with both minority and majority samples and with actual diversity policies. Specifically, we expect the longitudinal effect of school diversity approaches on achievement to be mediated by prior school belonging (Hypothesis 4).

Method

Participants and Procedure

We surveyed 5,336 students at Time 1 and 4,152 students 1 year later at Time 2; 3,446 participated in both waves (64% of the initial sample). Of those who participated in both waves, 103 changed schools and another 212 were missing essential data and were not included in our sample. Our final sample thus consisted of N = 3,131 students (52% female) who stayed on in the same 66 Flemish secondary schools in both waves (see attrition analysis under the “Results” section). Using self-reported own country of birth and parentage (i.e., one or both foreign-born parents and/or grandparents), we categorized first-, second-, and third-generation immigrant-origin adolescents as having an ethnic minority status (N = 1,747); the rest were categorized as majority youth (N = 1,384). Minority youth were mostly second-generation (first N = 614, second N = 986, and third N = 133) from Moroccan (N = 544), Turkish (N = 423), and European backgrounds (mainly Southern European Union [EU] N = 365; Western EU N = 75, other regions Ns <30; unspecified N = 259). At Time 1, participants were in Year 1 (32%), Year 2 (26%), or Year 3 (42%) of secondary school with an average age of 14.74 (SD = 1.18) and attending academic (41%), vocational (29%), or technical (30%) training.

The timeline of data collection is provided as supplemental material. Data were collected as part of a large-scale longitudinal study in randomly sampled secondary schools (CILS Belgium, 2014). Classes were randomly sampled within schools with varying ethnic composition, using the percentage of students speaking a foreign language at home (n = 902 in schools with >10% minorities, n = 841 in schools with 10%-30% minorities, n = 681 in schools with 30%-60% minorities, n = 707 in schools with >60% minorities). We obtained initial informed consent from school principals and teachers and informed participants and their parents of their right to opt out. Participants filled out Dutch language questionnaires during class time under the supervision of trained research assistants and a teacher. Teachers and research assistants had a majority-Belgian background (96% of teachers were of Belgian origin). School policy documents consisted of the mission statements and the rules and regulations of the 66 schools, which were downloaded from school websites or obtained from administrative staff. Research was conducted in line with American Psychological Association (APA) ethical guidelines.

Measures

School belonging

School belonging was measured with four items (Wang, Willett, & Eccles, 2011) rated from 1 (strongly disagree) to 5 (strongly agree). Sample items are “I am proud to be a student of this school; I feel at home at this school.” The measure had high reliabilities in both waves (Time 1 α = .85, Time 2 α = .85). For the main analysis, we focus on school belonging at Time 2 to estimate longitudinal effects of diversity policies. In addition, prior belonging was also measured at Time 1 to test possible mediation from Time 1 belonging to Time 2 achievement. Note that students had already been exposed to the school policies when reporting their belonging at Time 1 as there is no baseline measure of school belonging preceding school entry.

School achievement

As a measure of school achievement, we asked participants to report their Dutch grades.¹ For our analysis, we focus on Time 2 language grades as a key outcome measure. Language achievement is at the same time most unequal between minority and majority students (Heath & Brinbaum, 2014) and most predictive of minority future success (Bleakley & Chin, 2004; Dustmann & Fabbri, 2003). Students indicated retrospectively their school-report grade for Dutch language at the end of the Fall term (recoded from 0 to 100 across schools with different grading systems). As schools were surveyed during the Spring term, Time 2 grades precede our Time 2 belonging measure by at least 1 month.

Control variables

At the school level, we controlled for ethnic school composition as indicated by administrative data on students speaking a foreign language at home (three dummies: 10%-30%, 30%-60%, and >60% minority students, with <10% as a reference category). At the individual level, we controlled for the school track of each student. The Belgian educational tracking system sets students up for different career paths: Vocational tracks prepare students directly for the labor market, while technical and academic tracks prepare for higher professional and academic education, respectively (Baysu & De Valk, 2012). School track was recoded as two dummy variables: vocational and technical with academic track as a reference category. As a proxy for students’ socioeconomic status, we accounted for parental education (based on the parent with the highest qualification: 1 = primary school, 2 = secondary school, 3 = university or higher). Given the wide age range, we controlled for age. While there was a significant gender difference in Dutch grades, t(2852) = −2.74, p = .006; girls reported higher grades, M_diff = 2.10, when adding all other control variables to our model gender did not show any significant effects. At the individual level, neither gender nor year in school had any significant effects and were hence dropped from the analysis.

Results

Results are presented in two main sections in line with our double research aim. To explore school diversity approaches, we report the content analysis of school policy documents and the cluster analysis of the frequency scores from the coding. To investigate the effects of school policies on belonging and achievement gaps, we present (a) attrition analysis; (b) multilevel path models testing Hypotheses 1 to 3, including additional analyses of interactions with ethnic school composition and associations between diversity approaches; and (c) mediation analysis testing Hypothesis 4.

Diversity Approaches

Content analysis and cluster analysis of school policy documents

In a first step, we conducted qualitative coding of each school’s rules (stating explicit regulations) and mission and vision statements (stating values/principles) for all 66 schools that participated in both waves. Conceptually, the coding of subthemes was organized around three general diversity approaches: multiculturalism, colorblindness, and assimilationism. We developed a thematic coding scheme, drawing on U.S. and European examples of cultural diversity vocabularies (Stevens et al., 2008; Verkuyten, 2005) and using conceptual distinctions between approaches that value, ignore, or reject diversity as an organizing framework (Guimond et al., 2014). To ensure that our codes fully covered and accurately rendered context-specific meanings, we inductively fine-tuned our thematic coding scheme on the basis of initial reads of the school policy texts.

Technically, coding units were sentences and each document was coded by two separate coders. Initially, coders independently coded the first few documents. Together, we discussed any discrepancies, amended the coding scheme, and recoded if necessary (Guest & McLellan, 2003), and then coders continued coding the remaining documents using NVivo 11 (2015). Documents containing school mission statements, school rules, or both were analyzed together. Interrater reliability was high (κ = .95, κ² = .90). More information about the coding scheme development and coding procedure can be found in the supplemental material.

In a next step, we removed subthemes occurring in less than 15% of the documents in the frequency report, so that all selected subthemes covered a broad sample of schools (Guest & McLellan, 2003). The frequencies for the final 15 coding subthemes (e.g., must speak only Dutch, occurrences N = 92) are indicated in Table 1. We then conducted cluster analyses on the frequency data to validate our thematic coding of the school documents. To select the best cluster solution, we used Ward’s Linkage method for hierarchical cluster analysis (Field, 2000). We tested two- to five-cluster solutions and decided that a four-cluster solution fit our data best (see supplemental material; Field, 2000; Norušis, 2011). The four-cluster solution distinguished three general diversity approaches (Multiculturalism, Colorblindness, Assimilationism) and one separate approach focusing on Equality. Table 1 shows the specific subthemes included in the final four clusters, which define different diversity approaches in the Belgian school context. We provide a table with specific examples of diversity language for each subtheme as supplemental material.

Table 1.

School Diversity Approaches in Final Four-Cluster Solution: Subthemes and Frequencies.

Cluster (frequency range per school)	Specific subthemes	N	Total N
Multiculturalism (0-12)	• Learning about different cultures as a value	37	85
Multiculturalism (0-12)	• Inclusion of diversity in curriculum/instruction	48	85
Colorblindness (0-13)	• Emphasis on individual talent	89	273
	• Emphasis on individual counseling	48
	• Antidiscrimination bodies available	12
	• Zero-tolerance toward racism and discrimination	24
	• Can’t wear religious symbols	27
	• Religious symbols only in religion class	13
	• Religion class not compulsory	27
	• Special classes offered for newcomers	13
	• Respect individual rights and liberties	20
Assimilationism (0-7)	• Must speak only Dutch (classroom/playground)	92	143
Assimilationism (0-7)	• Can’t wear headscarves	51	143
Equality (0-5)	• Awareness of difference	49	67
Equality (0-5)	• Equality as a value	18	67
	Total codes		568

Note. N indicates frequency of each code across all documents.

“Multiculturalism” consisted of two subthemes valuing cultural diversity, focusing on the value of learning about different cultures and how diversity is included in the curriculum. “Colorblindness” was a broad cluster, coupling principled religious neutrality with individualistic values. In the Belgian school context, ignoring diversity refers narrowly to religious difference. Secularist policies are common in nondenominational schools and ensure religious neutrality. Specifically, Belgian state schools offer religious (including Islamic) classes in line with a European understanding of freedom of religion, restricting religious expression to these designated classes and imposing strict neutrality outside of those classes. In addition, colorblind policies value individual merits and rights; they see disadvantage as individual challenge or deficit, and they protect individual students from discrimination. “Assimilationism” was a more narrow cluster rejecting cultural diversity. In the Belgian school context, it refers specifically to restrictive policies targeting linguistic and religious differences. Finally, a separate cluster “Equality” was induced from our data. This egalitarian approach valued social equality while recognizing difference. As distinct from multiculturalism, its focus was on valuing equality rather than diversity. This approach also differed from colorblindness in that differences were acknowledged rather than ignored.

In a final step, we standardized frequency scores to create four z scores on multiculturalism, colorblindness, assimilationism, and equality for each school so that the same school could endorse different approaches to varying degrees. Across all documents, instances of colorblindness covered most subthemes (nine) and were most frequent (273 times); assimilationism, multiculturalism, and equality clusters were more narrowly defined (two subthemes each), and assimilationism was more frequent (143 times) than multiculturalism (85 times) and equality (67 times) (see Table 1). Additional analyses regressing the clusters on school-level student perceptions of the school diversity climate provided some support for the psychological validity of the different diversity approaches (see supplemental material).

Effects on Belonging and Achievement Gaps

Attrition analyses

While this study reached a large and representative sample of both minority and majority students across Flemish secondary schools, panel dropout at Time 2 was not entirely random. Comparison of participants who dropped out (i.e., only completed our survey at Time 1) with those who stayed on (i.e., participated at Times 1 and 2) revealed more attrition among ethnic minority (vs. majority) students; of those who dropped out, 69% had ethnic minority status against 56% in the final sample; t(4,135.56) = −8.66, p < .001, 95% confidence interval (CI) = [−0.145, −0.092]; for older (vs. younger) age groups, M = 15.03, SD = 1.23 versus M = 14.74, SD = 1.20; t(4,967) = 8.18, p < .001, 95% CI = [0.223, 0.363]; and at lower levels of Time 1 school belonging, M = 3.35, SD = 1.00 versus M = 3.74, SD = 0.83; t(3,218.95) = −14.22, p < .001, 95% CI = [−0.446, −0.338]. The most commonly reported reasons for panel dropout were being absent or changing schools. Selective attrition reflects the reality that ethnic minority students are more likely to change schools or leave school (Kalmijn & Kraaykamp, 2003).

Multilevel path analyses

Contextual effects of school diversity approaches on minority and majority students’ academic outcomes at Time 2 were estimated by way of multilevel path analyses in Mplus 7 on the pooled minority and majority samples (Muthén & Muthén, 1998-2012). Separate analyses for majority and minority samples largely replicated the pattern of findings in the pooled analyses (see supplemental material). We tested cross-level interaction effects of ethnic minority status with school policy clusters. As the least frequent equality cluster had no significant effects on academic outcomes, it was removed from further analyses. Specifically, analyses were conducted with school belonging and achievement at Time 2 as dependent measures, with ethnic minority status (individual level), diversity approaches (school level) (multiculturalism, colorblindness, assimilationism), and their cross-level interaction as predictors, along with (individual- and school-level) control variables defined at Time 1 (see Table 2 for descriptive statistics of study variables).

Table 2.

Means, Standard Deviations, and Correlations of Study Variables at the Individual Level and School Level (N = 3,131, n = 66).

Individual level		M/%	SD	1	2	3	4
1.	Age T1	14.74	1.18
2.	Ethnic minority status T1	44%	0.50	−.17***
3.	Parental education T1	2.34	0.55	−.11***	.24***
4.	School belonging T2	3.57	0.90	−.14***	.15***	.06***
5.	Dutch grades T2	62.56	22.77	−.11***	.16***	.12***	.10***
School level		M	SD	6	7	8	9	10
6.	Ethnic school composition (10%-30%)	0.27	0.44
7.	Ethnic school composition (30%-60%)	0.22	0.41	−.32***
8.	Ethnic school composition (>60%)	0.23	0.42	−.33***	−.29***
9.	Multiculturalism	−0.001	0.99	−.12***	.06****	.23***
10.	Colorblindness	−0.032	0.98	.08***	.13***	.13***	.18***
11.	Assimilation	−0.007	1.01	−.05**	.11***	.02	.50***	.22***

Note. Percentages are presented for categorical variables instead of means. Ethnic minority status is coded as 0 = ethnic minority background and 1 = majority background.

p < .05. **p < .01. ***p < .001.

To conclude that there are real belonging and achievement gaps between majority and minority youth, the effect of minority status on belonging and achievement should remain significant even after controlling for individual and school characteristics, and to conclude that school diversity approaches attenuate or exacerbate the gaps, the cross-level interaction effects of ethnic minority status with diversity approaches on belonging and achievement should be significant. To interpret interaction effects, we report Wald chi-square tests to indicate significant simple slopes (Muthén & Muthén, 1998-2012). Estimating exact power for multilevel models using cross-level interactions is highly complex; therefore, a 30/30 rule of thumb is suggested for sufficient statistical power: 30 groups with 30 individuals per group for a sample of 900 (Scherbaum & Ferreter, 2009). The present research exceeds the 30/30 rule of thumb with 3,131 students across 66 schools.

The analyses were conducted in a stepwise fashion and each step significantly improved the model fit (see Table 3 for stepwise model fit statistics and explained variances). We started from a null model with a random intercept only. The residual variances of school belonging and Dutch grades were significant both at the individual level, 0.72 (0.03), p < .001, 95% CI = [0.667, 0.768]; 279.40 (36.57), p < .001, 95% CI = [207.733, 351.067], and at the school level, 0.06 (0.01), p < .001, 95% CI = [0.036, 0.086]; 180.64 (53.59), p < .001, 95% CI = [75.608, 285.664]. The intraclass correlations (ICCs) indicated that, respectively, 7.8% and 39.2% of the total variance in school belonging and Dutch grades are found between schools. In a second step, the model included only control variables, which explained significant variance at the individual level in both outcomes. In a third step, the main-effects-only model, we added minority status and diversity approaches (multiculturalism, colorblindness, assimilationism) as predictors. Adding diversity approaches explained significant variance at the school level in both outcomes. In a final step, we added cross-level interactions between minority status and each of the diversity approaches. For parsimony, we included only significant cross-level interactions in the final model.

Table 3.

Model Fit Statistics and Explained Variance for Stepwise Models.

	Covariates-only model	Main-effects-only model	Final model
Log likelihood	28,091.52	28,080.01	28,067.24
parameters	23	27	32
AIC	28,137.52	28,134.02	28,131.24
BIC	28,272.01	28,291.89	28,318.36
Sample size–adjusted BIC	28,198.93	28,206.11	28,216.69
Δχ², p value	1,440.81	13.89	13.07
Δχ², p value	p < .001	p = .008	p = .023
Explained variance
Individual level
School belonging	R² = .025	R² = .026
School belonging	p = .022	p = .022
Dutch grades	R² = .044	R² = .045
Dutch grades	p = .006	p = .005
School level
School belonging	R² = .206	R² = .265
School belonging	p = .094	p = .031
Dutch grades	R² = .128	R² = .204
Dutch grades	p = .084	p = .024

Note. R² values are not calculated for (final) model with cross-level interactions. AIC = Akaike information criterion; BIC = Bayesian information criterion.

In line with the expected ethnic gaps, minority students reported significantly less belonging (M = 3.52) and lower Dutch grades (M = 59.28) than majority peers (M = 3.70 and M = 63.14, respectively; both ps < .001). Significant ethnic gaps remained even in the third step when control variables and diversity policies were added to the model: Minority students still evinced significantly less belonging (M = 4.37) and lower Dutch grades (M = 72.86) than majority peers (M = 4.50, M = 75.70, p = .004, p = .001, respectively). Below we report detailed results from our final model with significant cross-level interactions (see Table 4 and Figure 1). When main effects differ between the main-effects-only model and the final model, we report effects from both models.

Table 4.

Multilevel Path Analysis Relating Ethnic Minority Status and School Diversity Policies and Their Interaction to Time 2 School Belonging and Dutch Grades.

Individual level	T2 school belonging	T2 Dutch grades
Individual level	B (SE)	B (SE)
Control variables
Age	−0.04 (0.02)*	−0.65 (0.42)
Technical track (vs. academic)	−0.16 (0.06)**	−5.51 (1.39)***
Vocational track (vs. academic)	−0.18 (0.08)*	−3.11 (2.52)
Parental education	ns	1.59 (0.77)*
Intercept of the random slope
Ethnic minority status	0.13 (0.04)**	2.93 (0.81)***
Residual variance	0.70 (0.03)***	258.35 (34.74)***
School level
Intercept	4.33 (0.24)***	73.37 (6.45)***
Control variables
Ethnic school composition (10%-30%)	−0.14 (0.08)*	−1.25 (2.74)
Ethnic school composition (30%-60%)	−0.21 (0.07)**	−7.86 (4.39)^{^}
Ethnic school composition (>60%)	−0.15 (0.09)^{^^}	−11.31 (4.55)**
Predictors
MC	0.04 (0.03)	3.13 (1.16)**
CB	ns	−3.42 (1.60)*
AS	−0.09 (0.03)***	ns
Random slope cross-level interactions
MC × Ethnic Minority Status	−0.10 (0.04)**	ns
CB × Ethnic Minority Status	ns	1.88 (1.14)^{^^}
AS × Ethnic Minority Status	0.10 (0.05)*	ns
Residual variance	0.02 (0.01)*	119.51 (34.54)***

Note. Model presents unstandardized regression results with standard errors in parentheses. ns indicates the “nonsignificant” effects that were set to be 0. Ethnic minority status is coded 0 = ethnic minority and 1 = majority. Academic Track is the reference category for Technical Track and Vocational Track dummy-coded variables. MC = multiculturalism; CB = colorblindness; AS = assimilationism.

^^p < .097. ^p < .07. *p < .05. **p < .01. ***p < .001.

Figure 1.

Multilevel path model relating ethnic minority status and school policies and their interaction to Time 2 school belonging and Dutch grades.

Longitudinal effects on school belonging

We found significant effects of school diversity approaches on the belonging gap at Time 2 (Hypotheses 1-3; see Table 4 and Figure 1). For multiculturalism, no main effect was found, but there was a significant cross-level interaction between minority status and multiculturalism (Figure 2). In low multiculturalism schools, minority students felt significantly less belonging than majority students, Wald χ²(1) = 12.90, p < .001. In high multiculturalism schools, on the contrary, the gap was not significant such that majority and minority students did not differ in their school belonging, Wald χ²(1) = 0.54, p = .463. The minority slope was not significant, p = .228. Moreover, majority students’ belonging was not significantly related to multiculturalism. In the pooled model, the majority slope neared significance, p = .088, but separate analyses for minority and majorities revealed no significant effect for multiculturalism on majority students’ belonging. As expected, the interaction pattern reveals that multiculturalism relates to a reduced majority–minority belonging gap without significantly relating to reduced majority belonging. For colorblindness, there were no significant effects on belonging. Assimilationism was related to lower school belonging, as evident from a significant main effect. As this main effect was not significant in the main-effects-only model, p = .106, the effect appears to be conditional on the significant interaction with minority status (see Figure 3). In low assimilationism schools, minority and majority students did not differ in belonging; thus, there was no significant majority–minority gap. Conversely, in high assimilationism schools, the gap was significant: Minority students felt significantly less belonging than majority students, Wald χ²(1) = 12.18, p < .001. The more assimilationist the school, the less belonging minority students felt, Wald χ²(1) = 10.12, p = .002. Assimilationism was not related to majority students’ belonging, however: The majority slope was not significant, p = .845.

Figure 2.

Cross-level interaction effect of ethnic minority status with school diversity policies of multiculturalism on Time 2 school belonging.

Figure 3.

Cross-level interaction effect of ethnic minority status with school diversity policies of assimilationism on Time 2 school belonging.

Longitudinal effects on achievement

We also found significant effects of diversity approaches on Dutch grades at Time 2 (Hypotheses 1-3; see Table 4 and Figure 1). Multiculturalism was related to better achievement across minority and majority students, as evident from a significant main effect. The more multiculturalist the school, the higher the students’ self-reported Dutch language grades. While the interaction with ethnic minority status was not significant, additional analyses in minority and majority samples separately suggest that the achievement benefits of multiculturalism in the pooled data were driven by a significant positive effect for minority students, whereas the grades of majority students were not significantly related to multiculturalism (see supplemental material). Colorblindness also had a significant main effect on grades—albeit in the opposite direction—such that the more colorblind a school was, the lower the self-reported Dutch grades were (Table 4). As this main effect was near significant in the main-effects-only model (B = −2.75, SE = 1.51, p = .069), the effect in the final model appears to be conditional on the trend-significant interaction with minority status (Table 4). This interaction shows that in low colorblindness schools, minority and majority students did not differ in grades (p = .346); thus, there was no significant majority–minority gap. In high colorblindness schools, on the contrary, the gap was significant: Minority students reported significantly lower grades than majority students, Wald χ²(1) = 9.48, p = .002. The more colorblind the school, the lower the minority students’ self-reported grades, Wald χ²(1) = 4.55, p = .033. Moreover, the majority slope was not significant, so majority students’ grades were not affected by a colorblind approach (p = .337) (see Figure 4). Assimilationism did not affect Dutch grades.

Figure 4.

Cross-level interaction effect of ethnic minority status with school diversity policies of colorblindness on Time 2 Dutch grades (B = 1.88 SE = 1.14, p = .097, 95% CI = [−0.344, 4.110]).

Interaction with ethnic school composition

We tested whether the effects of different diversity approaches on majority–minority gaps held in schools with varying ethnic composition. Specifically, we differentiated between schools with low versus high minority presence, that is, schools where minority students make up the local numerical minority versus majority (>60% minority students). We found only one significant interaction effect: Ethnic school composition moderated the association of multiculturalism with Dutch grades (B = 4.56, SE = 2.135, p = .033, 95% CI = [0.370, 8.740]) so that multiculturalism was beneficial for students’ self-reported grades in schools with larger shares of minority students, Wald χ²(1) = 10.68, p = .001 (Figure in supplemental material).

Associations between diversity approaches.²

Diversity approaches were positively correlated. Colorblindness correlated similarly with multiculturalism (r = .18, p < .001) and assimilation (r = .22, p < .001), but the strongest correlation was between multiculturalism and assimilationism (r = .50, p < .001). These associations suggest that schools endorsed multiple diversity approaches (Civitillo et al., 2017). Additional analyses testing interaction effects between diversity approaches on grades did not reach statistical significance (for the results, see supplemental material).

Mediation analyses

We tested whether the effects of school diversity approaches on the achievement gap at Time 2 were mediated through prior school belonging (Hypothesis 4). To ensure that belonging was indeed measured before achievement, in these mediational analyses we used the measure of school belonging at Time 1 (1 year earlier) as a predictor of school belonging and Dutch grades at Time 2. To investigate the expected mediation, we retested our final model adding effects of controls and school diversity approaches on Time 1 belonging to the model, and testing the effects of Time 1 belonging on Time 2 belonging and Time 2 Dutch grades. This model also included all previously tested associations of school diversity approaches, ethnic minority status, and their interaction with belonging and achievement at Time 2 (see supplemental material for the complete mediation model). Our final model in Table 4 held up when adding Time 1 belonging as a mediator, except for minor changes in line with the mediation: Colorblindness predicted Time 1 school belonging (B = −0.07, SE = 0.03, p = .008, 95% CI = [−0.119, −0.018]); Time 1 school belonging significantly predicted Time 2 belonging and Time 2 Dutch grades (B = 0.68, SE = 0.19, p < .001, 95% CI = [0.297, 1.059]; B = 30.89, SE = 11.44, p = .007, 95% CI = [8.470, 53.317], respectively); and the main effect of colorblindness on Time 2 Dutch grades became nonsignificant. Using the Goodman test which gives an unbiased estimate of the variance of the indirect effect (Preacher & Leonardelli, 2015), the indirect effect of colorblindness (through Time 1 belonging) on Time 2 Dutch grades was near significant, z = −1.84, p = .066, and the indirect effect of colorblindness (through Time 1 belonging) on Time 2 belonging was significant z = −2.19, p = .029. Thus, for both minority and majority students, colorblindness predicted lower school belonging at Time 1, which in turn predicted less belonging and (near significantly) lower achievement 1 year later. The findings provide partial support for Hypothesis 4 for colorblindness only.

Discussion

Minority adolescents often lag behind their majority peers in terms of belonging and achievement in today’s increasingly diverse schools. One way for schools to reduce ethnic inequalities is to develop institutional policies and practices to maximize the benefits of cultural diversity and to minimize its costs. We examined how schools approach cultural diversity in a European educational context by analyzing actual school policies. Thus, we inductively derived meaningful and distinct diversity approaches. In a next step, we could relate these approaches to the gaps between minority and majority students’ school belonging and achievement 1 year later. Overall, approaches that value cultural diversity predicted smaller gaps, whereas approaches ignoring or rejecting diversity were related to persistent gaps in student outcomes. Below we first discuss the distinct diversity approaches that emerged from our analysis of actual school policies. Next, we discuss how these contextual approaches relate to the individual school belonging and achievement of minority and majority students over time.

Diversity Approaches: Multiculturalism, Colorblindness, and Assimilationism

Our study distinguished multiculturalism, colorblindness, assimilationism, and equality as common diversity approaches in our Belgian school panel. The defining aspects of each approach mirror conceptual distinctions between alternate policies valuing, ignoring, or rejecting cultural diversity: for instance, including diversity in the curriculum for multiculturalism, emphasizing individual talent for colorblindness, and only being allowed to speak Dutch for assimilationism.

Appreciating cultural diversity is a defining aspect of multiculturalism (Rattan & Ambady, 2013; Rosenthal & Levy, 2010), which resonates with similar notions of “value in diversity/difference” or “cultural pluralism” in organizational or school contexts (Apfelbaum et al., 2016; Civitillo et al., 2017; Schachner et al., 2016). However, it is more narrow than multicultural education, which includes a range of related practices such as combating racism, developing caring relations between teachers and students, and endorsing various pedagogical tools (Banks, 2015; Zirkel, 2008).

Unexpectedly, colorblindness was the most frequent approach in Belgian schools, rather than assimilationism, which has been proposed as the most common diversity approach in European diversity research (Civitillo et al., 2017; Fine-Davies & Faas, 2014). Along with individualistic aspects, a closer look at the meaning of colorblindness revealed aspects of antidiscrimination and secularism that reflect historically rooted and widely shared principles in European societies (Guimond et al., 2014). Assimilationist policies, in contrast, referred narrowly to restrictions on minority language use and religious expression as defining aspects and were hence less broad and less common than anticipated.

In the Belgian school context, colorblindness covers three main themes: individualism, antidiscrimination, and secularism. Stressing individual talents, merits, rights, or needs is a core aspect of colorblindness and reflects ideals of individualism (Guimond et al., 2014). Combating racism and discrimination, however, is sometimes associated with multiculturalism, as in multicultural education (Banks, 2015; Verkuyten & Thijs, 2013), and at other times with colorblindness (Rosenthal & Levy, 2010; Schwarzenthal, Schachner, van de Vijver, & Juang, 2018), as is the case here. The difference probably lies in the focus: The former—multicultural education—challenges discrimination to expose issues of racial or cultural difference, whereas the latter highlights equal treatment. Secularism is also part of colorblindness and denotes the so-called neutrality principle that restricts religious expression or symbols in the public domain. Religious neutrality is the institutionalized approach of religious diversity in state schools as distinct from Catholic schools in Flanders, Belgium. It is not surprising that institutionalized secularism fits with colorblindness because of its purpose to prevent differential treatment of (non)religious minorities. While colorblindness is prevalent in the North American context (Stevens et al., 2008), we show that this is also the case in the European context, though what constitutes colorblindness differs on both sides of the Atlantic (Guimond et al., 2014).

The assimilationist approach is theoretically an extension of the colorblind approach (Rosenthal & Levy, 2010), yet it is also distinct. Assimilationism requires that minorities adopt the majority culture (as in German schools, Civitillo et al., 2017) by distancing themselves from their distinct minority culture or identity. Accordingly, assimilationist policies tend to restrict the maintenance of the heritage culture, for instance, banning headscarves or penalizing the use of one’s mother tongue in school. These markers of minority identity—headscarf and language—are hot topics in public debates over national identity and cultural diversity targeting Muslim minorities in Western Europe (Fleischmann & Phalet, 2018).

In addition, equality emerged as a separate diversity approach in this study. This finding resonates with the argument by Apfelbaum and colleagues (2016) that ignoring diversity and equal treatment are separate components of colorblindness. It is also in line with recent research highlighting potential benefits of a focus on equality as a value in the European context (Apfelbaum et al., 2016; Roebroeck & Guimond, 2015; Schachner et al., 2016). Interestingly, egalitarian values were associated with the awareness—rather than the denial—of difference in our study. Still, this approach was less prevalent than other approaches in Belgian school policies. Thanks to our bottom-up method, we were able to separate out equality from colorblindness proper, whereas confounding these aspects could have resulted in mixed or zero effects of colorblindness.

The bottom-up definition of diversity approaches also revealed unanticipated context-relevant meanings, which enriched our initial coding scheme. For example, the phrases “can’t wear religious symbols” and “can’t wear headscarves” came under different approaches of colorblindness and assimilationism, respectively. This distinction reflects distinct policies in the Belgian school context: “can’t wear religious symbols” is a blanket assertion of principled neutrality toward all religions (colorblindness) in Belgian state schools, whereas “can’t wear headscarves” targets Muslims as a devalued religious minority group in Belgian society (assimilationism). Our bottom-up analysis of actual policies thus captured some more subtle differences between colorblind and assimilationist language.

It is noteworthy that schools typically endorsed more than one diversity approach, in line with previous research (Civitillo et al., 2017; Plaut et al., 2018), and that different approaches were positively correlated (Ryan, Casas, & Thompson, 2010). One reason may be a distinction between diversity principles and practices. Our initial coding scheme anticipated both behaviorally oriented subthemes (about rules and regulations) and value-oriented subthemes (about principles and ideals) for each diversity approach. Yet, assimilationism relied narrowly on behavioral restrictions, regulating language use and religious dress code, while both multiculturalism and colorblindness included more value-oriented subthemes. Thus, schools may communicate principled multiculturalism, yet implement some assimilationist policies that maintain inequalities in a practical manner, in line with a possible principle–implementation gap (Dixon, Durrheim, & Tredoux, 2007).

Relationship of Diversity Approaches to School Belonging and Achievement

We now turn to the relationship of these diversity approaches to majority–minority gaps in academic outcomes. We focused on language grades, as opposed to other achievement outcomes, to assess how school policies related to an outcome that is the most unequal and most predictive of future success for minorities (Bleakley & Chin, 2004; Heath & Brinbaum, 2014). Each of the three main diversity approaches had unique effects, and multiculturalism was the most beneficial for both belonging and achievement. Multiculturalism attenuated the majority–minority gap in belonging. Specifically, while in schools scoring low on multiculturalism minority students felt significantly less belonging than majority students, in schools with high multiculturalism there was no gap in belonging. The more multiculturalist the school was, the higher the reported Dutch grades were (driven by the positive effect for minority students). This is in line with the social identity argument that multiculturalism values (rather than threatens) minorities’ cultural identities and thus enables their belonging and achievement. While the benefits of multiculturalism for minority outcomes are well-documented (Plaut et al., 2018; Rattan & Ambady, 2013; Stevens et al., 2008), the absence of significant negative effects for majority students’ achievement is an important nonfinding. In particular, highly diverse schools may thus be able to create a school climate of all-inclusive multiculturalism, affirming the distinct cultures and identities of minority students without excluding majority students (Jansen et al., 2015; Stevens et al., 2008). Potentially, in the “super-diverse” schools of present days (Vertovec, 2007) where majorities are no longer the numerical majority, a multiculturalist diversity definition may signal the value of majorities’ identities as much as minorities’ identities. If so, one would expect the benefits of multiculturalism for all students to be higher in more diverse schools. Additional analyses indeed supported this latter possibility by showing that multiculturalism policy was especially beneficial for achievement in schools with larger shares of ethnic minority students (>60% minorities).

A colorblind approach focusing on individualism and ignoring differences was detrimental for both belonging and achievement, particularly for minority students, and predicted persistent gaps in achievement. In less colorblind schools, minority and majority students did not differ in achievement. In more colorblind schools, on the contrary, the gap was significant: Minorities reported significantly lower grades than majorities. The more colorblind the school, the lower the minority students’ grades; yet it did not harm or benefit the majority students. This is also in line with social identity valuation perspective such that ignoring diversity through a colorblind approach undermines minority belonging and performance (Holoien & Shelton, 2012; Meeussen et al., 2014).

An assimilationist approach rejecting cultural diversity had negative effects on minorities’ sense of belonging and predicted persistent gaps in belonging. In less assimilationist schools, minority and majority students did not differ in belonging. Conversely, in more assimilationist schools, minority students felt significantly less belonging than majority students. The more assimilationist the school was, the less belonging minority students felt, whereas it did not affect majority students’ belonging. In line with social identity threat, the pressure to adopt the majority culture and to leave behind the minority culture in school appears to alienate minority students from the school as a majority context. Assimilationism, at least from the perspective of minorities, is thus not necessarily about majority and minority identities becoming part of a new overarching common identity (Guimond et al., 2014). Our findings suggest that assimilationist policies rejecting cultural diversity relate to the worst outcomes for minorities’ identity and belonging. An interesting nonfinding is that the minority–majority gap in Dutch language grades persists in assimilationist schools, whereas assimilationist policies are commonly justified as necessary to promote Dutch language mastery (Pulinx et al., 2017).

The additional cluster of equality did not have any effects on our outcome measures, possibly because it was infrequently mentioned in the school documents and had lower variance compared with other approaches. However, previous research has shown that student perceptions of equal treatment matter and are related to higher belonging and achievement (Baysu et al., 2016), psychological adjustment in school (Schachner et al., 2016), and lower prejudice (Roebroeck & Guimond, 2015).

We also found that the relationship between school diversity policies and achievement was mediated by prior school belonging: Minorities felt weaker belonging in more colorblind schools, and this lack of belonging in turn predicted later academic outcomes. This mediation suggests that colorblindness creates an unwelcoming environment that can harm minority achievement through undermining belonging (Cook et al., 2012). Plausibly, minority students who lack a sense of belonging in school will sooner disengage from their schoolwork and therefore fall behind (Baysu, Phalet, & Brown, 2011). Our findings differ from those of Jansen et al. (2016) who find that for majority employees, colorblindness positively affects workplace outcomes, mediated by enhanced belonging. Notably, their definition of colorblindness in the workplace-context focused on valuing competence (qualifications and job performance), which was less central in colorblind school policies. This further suggests that definitions and outcomes of colorblindness vary across contexts (Guimond et al., 2014).

The mediation hypothesis was not supported, however, for assimilationism and multiculturalism. As these policies were unrelated to the initial belonging of the students, they could not mediate policy effects on student outcomes over time. This nonfinding also implies that the longitudinal consequences of assimilation and multiculturalism for minority outcomes 1 year later cannot be due to initial differences in the belonging of minority students in those schools. This strengthens the causality argument that school policies affect school belonging and achievement, rather than the reverse.

The present research contributes to the current literature in several ways. First, the positive effect of a policy of multiculturalism in terms of reducing academic inequalities between minority and majority students extends a large amount of research associating multiculturalism with lower, and colorblindness and assimilationism with higher, levels of ethnic stereotyping and prejudice (Dovidio et al., 2007; Levin et al., 2012). Importantly, the fact that we did not find any evidence of the alleged costs of multicultural policies for majority outcomes is inconsistent with recent studies suggesting that multiculturalism is threatening for majorities (Plaut et al., 2018). These novel findings highlight the distinctive nature of our study. Methodologically, we combine qualitative and quantitative, multilevel and longitudinal forms of analyses to address our questions. Qualitative coding based our assessment of diversity policies in the details of actual diversity management in schools. Cluster analysis revealed how these details fit together to form distinct and coherent diversity approaches. Multilevel and longitudinal analyses related these approaches at the school level to individual outcomes over time for large random samples of minority and majority students. We showed that different diversity approaches are consequential, even after controlling for parental education, age, school track, and ethnic school composition, all of which affected academic outcomes. Another distinctive feature of our study is the inclusion of both minority and majority groups in the same schools, which enabled us to estimate real ethnic inequalities between both groups. We showed that outcomes for minority and majority students were differently related to diversity policies. Future research should further examine the influence of different school policies on students’ own attitudes toward diversity (Celeste et al., 2016; Schachner et al., 2016).

There are also limitations. First, the attrition of our sample was not entirely random, such that those who dropped out—around 1,800 students—were older, more often had a minority status, and felt less school belonging at Time 1. Although we could not assess how their belonging and achievement related to the school policies over time, it is likely that both harmful and beneficial effects would have been stronger for those who already felt marginalized in school. Another limitation is that many subthemes that we initially coded were not frequent enough to quantitatively analyze. This is partly due to the nature of a mixed-method approach, and an in-depth focus on the qualitative analysis might have uncovered more approaches. In terms of language, the questionnaire was presented in the Dutch language, which could reinforce the majority culture. However, we would expect any effects of the questionnaire language would be very small within the daily monolingual Dutch-speaking setting of the school. The self-reported nature of student grades as an achievement measure is another limitation. Although there is no obvious reason why minority and majority students would report their grades differently depending on diversity approaches, nevertheless, future studies may add actual student grades as external measures. In addition, the indirect effect of colorblindness on Dutch grades through prior belonging was only near significant, and mediation was not supported for multiculturalism or assimilationism. Future research can investigate alternate ways in which school policies may affect grades such as through school teachers as diversity managers (Meeussen et al., 2014): If policies affect teachers’ expectations (Pulinx et al., 2017) and behaviors in class (e.g., support or grading system), this could in turn affect student outcomes (Baysu & Phalet, 2012).

Finally, given that school policies were already in place when we collected data, we could not test reverse causation: whether schools may have implemented particular diversity approaches in response to achievement gaps. Therefore, we do not know how schools have developed different diversity approaches in response to diversity issues, nor how school diversity approaches and students’ diversity experiences may mutually reinforce each other over time. Future research could investigate such recursive cycles by following up the same schools over several years or by zooming in on naturally occurring policy changes, for instance, when schools transition to a new scale of diversity as a consequence of school mergers or, finally, by implementing new diversity policies and monitoring student achievement in school intervention studies (e.g., Civitillo et al., 2017).

To conclude, we examined real school policies and asked how schools approach diversity and how this relates to disparate belonging and achievement outcomes among majority and minority students. We found positive effects of an inclusive multiculturalism approach, which may attenuate the majority–minority gap in belonging and boost achievement across all students. We found these positive effects in spite of the fact that actual multicultural policies were mainly value-oriented and that some schools combined multiculturalism with colorblind or assimilationist policies. Our main findings suggest the potential of designing multiculturalist diversity policies as an effective way to reduce the gap between minority and majority school careers. Future applied studies should investigate ways to implement such policies, for instance, by training teachers to create identity-valuing school climates for all students and by raising awareness of the risks of often well-intentioned colorblind or assimilationist approaches to cultural diversity.

Supplemental Material

Celeste_OnlineAppendix – Supplemental material for Can School Diversity Policies Reduce Belonging and Achievement Gaps Between Minority and Majority Youth? Multiculturalism, Colorblindness, and Assimilationism Assessed

Supplemental material, Celeste_OnlineAppendix for Can School Diversity Policies Reduce Belonging and Achievement Gaps Between Minority and Majority Youth? Multiculturalism, Colorblindness, and Assimilationism Assessed by Laura Celeste, Gülseli Baysu, Karen Phalet, Loes Meeussen and Judit Kende in Personality and Social Psychology Bulletin

Supplemental Material

Supplemental_Materials – Supplemental material for Can School Diversity Policies Reduce Belonging and Achievement Gaps Between Minority and Majority Youth? Multiculturalism, Colorblindness, and Assimilationism Assessed

Supplemental material, Supplemental_Materials for Can School Diversity Policies Reduce Belonging and Achievement Gaps Between Minority and Majority Youth? Multiculturalism, Colorblindness, and Assimilationism Assessed by Laura Celeste, Gülseli Baysu, Karen Phalet, Loes Meeussen and Judit Kende in Personality and Social Psychology Bulletin

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research is supported by grants from the Research Foundation-Flanders (Fonds Wetenschappelijk Onderzoek-FWO G.0747.13), and University of Leuven Research Council (Interdisciplinaire Onderzoeksprogramma-IDO 10.005).

Notes

Supplemental Material

Supplemental material is available online with this article.

ORCID iD

Laura Celeste

References

Apfelbaum

E. P.

Stephens

N. M.

Reagans

R. E.

(2016). Beyond one-size-fits-all: Tailoring diversity approaches to the representation of social groups. Journal of Personality and Social Psychology, 111, 547-566. doi:10.1037/pspi0000071

Aronson

Laughter

(2016). The theory and practice of culturally relevant education: A synthesis of research across content areas. Review of Educational Research, 86, 237-276. doi:10.3102/0034654315582066

Banks

J. A.

(2015). Cultural diversity and education: Foundations, curriculum, and teaching (6th ed.). New York, NY: Routledge.

Baysu

G. D.

Celeste

Brown

R. J.

Verschueren

Phalet

(2016). Minority adolescents in ethnically diverse schools: Perceptions of equal treatment buffer threat effects. Child Development, 87, 1352-1366. doi:10.1111/cdev.12609

Baysu

G. D.

Phalet

(2012). Staying on or dropping out: The role of the school environment in minority and non-minority school careers. Teachers College Record, 114, 1-25.

Baysu

G. D.

Phalet

Brown

(2011). Dual identity as a two edged sword: Identity threat and minority school performance. Social Psychology Quarterly, 74, 121-143. doi:10.1177/0190272511407619

Baysu

G. D.

De Valk

(2012). Navigating the school system in Sweden, Belgium, Austria and Germany: School segregation and second generation school trajectories. Ethnicities, 12, 776-799. doi:10.1177/1468796812450857

Bleakley

Chin

(2004). Language skills and earnings: Evidence from childhood immigrants. The Review of Economics and Statistics, 86, 481-496. doi:10.1162/003465304323031067

Braun

Clarke

(2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3, 77-101. doi:10.1191/1478088706qp063oa

10.

Celeste

Baysu

Brown

R. J.

Phalet

(2019). Self-affirmation and test performance in ethnically diverse schools: A new dual-identity affirmation intervention. Manuscript submitted for publication.

11.

Celeste

Meeussen

Verschueren

Phalet

(2016). Minority acculturation and peer rejection: Costs of acculturation misfit with peer-group norms. British Journal of Social Psychology, 55, 544-563. doi:10.1111/bjso.12147

12.

Children of Immigrants Longitudinal Study Belgium. (2014). Technical report: Children of Immigrants Longitudinal Study (CILS) Belgium. Leuven, Belgium: CeSo/CSCP, University of Leuven.

13.

Civitillo

Schachner

Juang

van de Vijver

F. R. J.

Handrick

Noack

(2017). Towards a better understanding of cultural diversity approaches at school: A multi-informant and mixed-methods study. Learning, Culture and Social Interaction, 12, 1-14. doi:10.1016/j.lcsi.2016.09.002

14.

Cook

J. E.

Purdie-Vaughns

Garcia

Cohen

G. L.

(2012). Chronic threat and contingent belonging: Protective benefits of values affirmation on identity development. Journal of Social and Personality Psychology, 102, 479-496. doi:10.1037/a0026312

15.

Derks

Van Laar

Ellemers

(2007). The beneficial effects of social identity protection on the performance motivation of members of devalued groups. Social Issues and Policy Review, 1, 217-256. doi:10.1111/j.1751-2409.2007.00008.x

16.

Dimitrova

Chasiotis

van de Vijver

F. J. R.

(2016). Adjustment outcomes of immigrant children and youth in Europe: A meta-analysis. European Psychologist, 21, 150-162. doi:10.1027/1016-9040/a000246

17.

Dixon

Durrheim

Tredoux

(2007). Intergroup contact and attitudes toward the principle and practice of racial equality. Psychological Science, 18, 867-872. doi:10.1111/j.1467-9280.2007.01993.x

18.

Dovidio

J. F.

Gaertner

S. L.

Saguy

(2007). Another view of “we”: Majority and minority group perspectives on a common ingroup identity. European Review of Social Psychology, 18, 296-330. doi:10.1080/10463280701726132

19.

Dustmann

Fabbri

(2003). Language proficiency and labour market performance of immigrants in the U.K.. The Economic Journal, 113, 695-717.

20.

Eccles

J. S.

Wong

C. A.

Peck

S. C.

(2006). Ethnicity as a social context for the development of African-American adolescents. Journal of School Psychology, 44, 407-426. http://dx.doi.org/10.1016/j.jsp.2006.04.001

21.

Field

(2000). Postgraduate statistics: Cluster analysis [Handout]. Retrieved from http://www.statisticshell.com/docs/cluster.pdf

22.

Fine-Davies

Faas

(2014). Equality and diversity in the classroom: A comparison of students’ and teachers’ attitudes in six European countries. Social Indicators Research, 119, 1319-1334. doi:10.1007/s11205-013-0547-9

23.

Fleischmann

Phalet

(2018). Religion and national identification in Europe: Comparing Muslim youth in Belgium, England, Germany, the Netherlands, and Sweden. Journal of Cross-Cultural Psychology, 49, 44-61. doi:10.1177/0022022117741988

24.

Gillen-O’Neel

Fuligni

(2013). A longitudinal study of school belonging and academic motivation across high school. Child Development, 84, 678-692. doi:10.1111/j.1467-8624.2012.01862.x

25.

Gregory

Cornell

Fan

(2011). The relationship of school structure and support to suspension rates for Black and White high school students. American Educational Research Journal, 48, 904-934. doi:10.3102/0002831211398531

26.

Guest

McLellan

(2003). Distinguishing the trees from the forest: Applying cluster analysis to thematic qualitative data. Field Methods, 15, 186-201. doi:10.1177/1525822X03251188

27.

Guimond

de la Sablonnière

Nugier

(2014). Living in a multicultural world: Intergroup ideologies and the societal context of intergroup relations. European Review of Social Psychology, 25, 142-188. doi:10.1080/10463283.2014.957578

28.

Heath

A. F.

Brinbaum

(Eds.). (2014). Unequal attainments: Ethnic educational inequalities in ten Western countries (Proceedings of the British Academy, Vol. 196). Oxford, UK: Oxford University Press.

29.

Holoien

D. S.

Shelton

N. J.

(2012). You deplete me: The cognitive costs of colorblindness on ethnic minorities. Journal of Experimental Social Psychology, 48, 562-565. doi:10.1016/j.jesp.2011.09.010

30.

Hornsey

M. J.

Hogg

M. A.

(2000). Assimilation and diversity: An integrative model of subgroup relations. Personality and Social Psychology Review, 4, 143-156. doi:10.1207/S15327957PSPR0402_03

31.

Hoti

H. A.

Heinzmann

Muller

Buholzer

(2017). Psychological adaptation and school success of Italian, Portuguese and Albanian students in Switzerland: Disentangling migration background, acculturation and the school context. Journal of International Migration and Integration, 18, 85-106. doi:10.1007/s12134-015-0461-x

32.

Inzlicht

Good

(2006). How environments can threaten academic performance, self-knowledge, and sense of belonging. In Levin

Van Laar

(Eds.), Stigma and group inequality: Social psychological approaches (pp. 129-150). Mahwah, NJ: Lawrence Erlbaum.

33.

Jansen

W. S.

Otten

Van der Zee

K. I.

(2015). Being part of diversity: The effects of an all-inclusive multicultural diversity approach on majority members’ perceived inclusion and support for organizational diversity efforts. Group Processes & Intergroup Relations, 18, 817-832. doi:10.1177/1368430214566892

34.

Jansen

W. S.

Vos

M. W.

Otten

Podsiadlowski

Van der Zee

K. I.

(2016). Colorblind or colorful? How diversity approaches affect cultural majority and minority employees. Journal of Applied Social Psychology, 46, 81-93. doi:10.1111/jasp.12332

35.

Kalmijn

Kraaykamp

(2003). Dropout and downward mobility in the educational career: An event-history analysis of ethnic schooling differences in the Netherlands. Educational Research and Evaluation, 9, 265-287. doi:10.1076/edre.9.3.265.15572

36.

Levin

Matthews

Guimond

Sidanius

Pratto

Kteily

. . . Dover

(2012). Assimilation, multiculturalism, and colorblindness: Mediated and moderated relationships between social dominance orientation and prejudice. Journal of Experimental Social Psychology, 48, 207-212. doi:10.1016/j.jesp.2011.06.019

37.

Marsh

H. W.

Ludtke

Nagengast

Trautwein

Morin

A. J. S.

Abduljabbar

A. S.

Koller

(2012). Classroom climate and contextual effects: Conceptual and methodological issues in the evaluation of group-level effects. Educational Psychologist, 47, 106-124. doi:10.1080/00461520.2012.670488

38.

Meeussen

Otten

Phalet

(2014). Managing diversity: How leaders’ multiculturalism and colorblindness affect work group functioning. Group Processes & Intergroup Relations, 17, 629-644. doi:10.1177/1368430214525809

39.

Motti-Stefanidi

Masten

A. S.

(2013). School success and school engagement of immigrant children and adolescents: A risk and resilience developmental perspective. European Psychologist, 18, 126-135. doi:10.1027/1016-9040/a000139

40.

Muthén

L. K.

Muthén

B. O.

(1998-2012). Mplus user’s guide (7th ed.). Los Angeles, CA: Author.

41.

Namey

Guest

Thairu

Johnson

(2008). Data reduction techniques for large qualitative data sets. In Guest

MacQueen

K. M.

(Eds.), Handbook for team-based qualitative research (pp. 137-161). Lanham, MD: AltaMira Press.

42.

Norušis

(2011). Chapter 17: Cluster analysis. IBM SPSS statistics guides (pp. 375-404). Boston, MA: Addison-Wesley.

43.

NVivo 11. (2015). NVivo qualitative data analysis [Computer software]. Melbourne, Australia: QSR International Pty Ltd.

44.

Organisation for Economic Co-operation and Development. (2015). Immigrant students at school: Easing the journey towards integration. Paris, France: OECD Reviews of Migrant Education. doi:10.1787/9789264249509-en

45.

Park

Judd

C. M.

(2005). Rethinking the link between categorization and prejudice within a social cognition perspective. Personality and Social Psychology Review, 9, 106-130. doi:10.1207/s15327957pspr0902_2

46.

Plaut

V. C.

Thomas

K. M.

Kyneshawau

Romano

C. A.

(2018). Do color blindness and multiculturalism remedy or foster discrimination and racism? Psychological Science, 27, 200-206. doi:10.1177/0963721418766068

47.

Preacher

K. J.

Leonardelli

G. J.

(2015). Calculation for the Sobel test: An interactive calculation tool for mediation tests. Retrieved from http://quantpsy.org/sobel/sobel.htm

48.

Pulinx

Van Avermaet

Agirdag

(2017). Silencing linguistic diversity: The extent, the determinants and consequences of the monolingual beliefs of Flemish teachers. International Journal of Bilingual Education and Bilingualism, 20, 542-556. doi:10.1080/13670050.2015.1102860

49.

Rattan

Ambady

(2013). Diversity ideologies and intergroup relations: An examination of colorblindness and multiculturalism. European Journal of Social Psychology, 43, 12-21. doi:10.1002/ejsp.1892

50.

Roebroeck

Guimond

(2015). Schooling, citizen-making, and anti-immigrant prejudice in France. Journal of Social and Political Psychology, 3, 20-42. doi:10.5964/jspp.v3i2.391

51.

Roeser

R. W.

Midgley

Urdan

(1996). Perceptions of the school psychological environment and early adolescents’ psychological and behavioral functioning in school: The mediating role of goals and belonging. Journal of Educational Psychology, 88, 408-422. doi:10.1037/0022-0663.88.3.408

52.

Rosenthal

Levy

S. R.

(2010). The colorblind, multicultural, and polycultural ideological approaches to improving intergroup attitudes and relations. Social Issues and Policy Review, 4, 215-246. doi:10.1111/j.1751-2409.2010.01022.x

53.

Ryan

C. S.

Casas

J. F.

Thompson

B. K.

(2010). Interethnic ideology, intergroup perceptions, and cultural orientation. Journal of Social Issues, 66, 29-44. doi:10.1111/j.1540-4560.2009.01631.x

54.

Schachner

M. K.

Heizmann

van de Vijver

F. J. R.

(2017). Acculturation and school adjustment of immigrant youth in six European countries: Findings from the Programme for International Student Assessment (PISA). Frontiers in Psychology, 8, Article 649. doi:10.3389/fpsyg.2017.00649

55.

Schachner

M. K.

Noack

van de Vijver

F. J. R.

Eckstein

(2016). Cultural diversity climate and psychological adjustment at school: Equality and inclusion versus cultural pluralism. Child Development, 87, 1175-1191. doi:10.1111/cdev.12536

56.

Scherbaum

C. A.

Ferreter

J. M.

(2009). Estimating statistical power and required sample sizes for organizational research using multilevel modeling. Organizational Research Methods, 12, 347-367. doi:10.1177/1094428107308906

57.

Schwarzenthal

Schachner

M. K.

van de Vijver

F. J. R.

Juang

(2018). Equal but different: Effects of equality/inclusion and cultural pluralism on intergroup outcomes in multiethnic classrooms. Cultural Diversity and Ethnic Minority Psychology, 24, 260-271. doi:10.1037/cdp0000173

58.

Shnabel

Purdie-Vaughns

Cook

J. E.

Garcia

Cohen

G. L.

(2013). Demystifying values-affirmation interventions: Writing about social belonging is a key to buffering against identity threat. Personality and Social Psychology Bulletin, 39, 663-676. doi:10.1177/0146167213480816

59.

Stevens

F. G.

Plaut

V. C.

Sanchez-Burks

(2008). Unlocking the benefits of diversity: All-inclusive multiculturalism and positive organizational change. The Journal of Applied Behavioral Science, 44, 116-133. doi:10.1177/0021886308314460

60.

Van Acker

Vanbeselaere

. (2012). Heritage culture maintenance precludes host culture adoption and vice versa: Flemings’ perceptions of Turks’ acculturation behavior. Group Processes & Intergroup Relations, 15, 133-145. doi:10.1177/1368430211417263

61.

Verkuyten

(2005). Ethnic group identification and group evaluation among minority and majority groups: Testing the multiculturalism hypothesis. Journal of Personality and Social Psychology, 88, 121-138. doi:10.1037/0022-3514.88.1.121

62.

Verkuyten

Thijs

(2013). Multicultural education and inter-ethnic attitudes: An intergroup perspective. European Psychologist, 18, 179-190. doi:10.1027/1016-9040/a000152

63.

Vertovec

(2007). Super-diversity and its implications. Ethnic and Racial Studies, 30, 1024-1054. doi:10.1080/01419870701599465

64.

Walton

G. M.

Cohen

G. L.

(2007). A question of belonging: Race, social fit, and achievement. Journal of Personality and Social Psychology, 92, 82-96. doi:10.1037/0022-3514.92.1.82

65.

Wang

M.-T.

Willett

J. B.

Eccles

J. S.

(2011). The assessment of school engagement: Examining dimensionality and measurement invariance by gender and race/ethnicity. Journal of School Psychology, 49, 465-480. doi:10.1016/j.jsp.2011.04.001

66.

Zirkel

(2008). Creating more effective multiethnic schools. Social Issues and Policy Review, 2, 187-241. doi:10.1111/j.1751-2409.2008.00015.x

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.13 MB

0.25 MB

Can School Diversity Policies Reduce Belonging and Achievement Gaps Between Minority and Majority Youth? Multiculturalism,Colorblindness,and Assimilationism Assessed

Abstract

Keywords

Diversity Approaches: Multiculturalism, Colorblindness, and Assimilationism

Relationship of Diversity Approaches to School Belonging and Achievement

Method

Participants and Procedure

Measures

School belonging

School achievement

Control variables

Results

Diversity Approaches

Content analysis and cluster analysis of school policy documents

Effects on Belonging and Achievement Gaps

Attrition analyses

Multilevel path analyses

Longitudinal effects on school belonging

Longitudinal effects on achievement

Interaction with ethnic school composition

Associations between diversity approaches. 2

Mediation analyses

Discussion

Diversity Approaches: Multiculturalism, Colorblindness, and Assimilationism

Relationship of Diversity Approaches to School Belonging and Achievement

Supplemental Material

Celeste_OnlineAppendix – Supplemental material for Can School Diversity Policies Reduce Belonging and Achievement Gaps Between Minority and Majority Youth? Multiculturalism, Colorblindness, and Assimilationism Assessed

Supplemental Material

Supplemental_Materials – Supplemental material for Can School Diversity Policies Reduce Belonging and Achievement Gaps Between Minority and Majority Youth? Multiculturalism, Colorblindness, and Assimilationism Assessed

Footnotes

Declaration of Conflicting Interests

Funding

Notes

Supplemental Material

ORCID iD

References

Supplementary Material

Associations between diversity approaches.²