Social Inclusion of Refugee and Native Peers Among Adolescents: It is the Language that Matters!

This study investigated the role of refugee status and language skills for adolescents' inclusion or exclusion decisions in hypothetical intergroup scenarios. 100 German adolescents (Mage = 13.65 years, SD = 1.93) were presented three scenarios in which groups of adolescents are planning leisure time activities, and peers from their own country (Germany) versus another country (Syrian refugees) with either good or bad German skills want to join them. Whereas adolescents' inclusion decisions did not differ between the German protagonist and the Syrian one with good German skills, the Syrian protagonist with bad German skills was less likely to be included than either of the other two. These findings have implications for understanding the role of language in adolescents' inclusion decisions.

Since the outbreak of the war in Syria in 2011, refugees have increasingly sought shelter across Europe, Asia, and the United States (UNHCR, 2017). During 2015 and 2016, more than a million refugees arrived in Europe, with Germany attracting the highest number of all the European countries. For instance, in 2017 alone, 186,644 refugees entered Germany with the majority coming from Syria, Afghanistan, and Iraq (BAMF, 2018; UNHCR, 2018). Official statistics have shown that more than 29% of the refugees in Germany are children and youth (Huddleston, Tánczos, & Wolffhardt, 2016). These huge refugee movements might present challenges related to promoting social integration in educational settings. Hence, investigating factors that influence the healthy development and social integration of this young refugee population into the host country Germany is important.

While there has been focused attention ensuring that refugees have access to shelter, resources, and education, there have also been calls for additional research on how to best support the psychological health of refugee youth (Hodes et al., 2018; Persaud, 2017). In this context, additional attention must be paid to the attitudes of the German people toward refugees.

John Berry's model of acculturation has been one of the most influential models aiming to explore acculturation. Within this model, Berry theorizes four acculturation categories—integration (adopts the host culture and retains the heritage culture), assimilation (discards the heritage culture and adopts the host culture), separation (retains the heritage culture and rejects the host culture), and marginalization (rejects both the heritage and host cultures) (Berry, 1999, 2001). In the current study, we focused on the conceptualization of attitudes toward integration. The integration process is recognized as a process of mutual accommodation between relative newcomers to a country and the members of the host society (Berry, Phinney, Sam, & Vedder, 2006). As a reciprocal process, it requires the host society to be open to integration and welcoming toward the immigrants and refugees (Berry, 2011). Thus, the attitudes of Germans toward refugees are crucial for integration. If we are interested in the experiences of young refugees, we need to find out more about the attitudes of their German peers. Thus, it is necessary to investigate how open adolescents are toward including refugees in their social and peer interactions. This is an especially important area for research as evidence documents how detrimental social exclusion can be for one's health and well‐being (Buhs, 2005; Buhs & Ladd, 2001; Juvonen & Gross, 2005; Rutland & Killen, 2015).

However, little is known about which factors, in particular, might have an impact on peer inclusion or exclusion, especially in the context intergroup processes including locals and refugee youth. In the current study, we examined the role of language skills for decisions about the inclusion or exclusion of refugee adolescents. Prior research has documented that language skills are a key element for integration in terms of academic achievement (Duong, Badaly, Liu, Schwartz, & McCarty, 2016; Stanat & Christensen, 2006) and labor market entry (Auer, 2018; Esser, 2006). Further, language is an important intergroup category for social categorization (Kinzler, 2013; Mulvey, Boswell, & Niehaus, 2018). Prior research with children has demonstrated that children expect to be quite inclusive of language out‐group members (Mulvey, Boswell et al., 2018), but new research is needed that focuses on adolescents. Moreover, while prior work has examined inclusion of language out‐group members (Mulvey, Boswell et al., 2018), no prior work has examined inclusion of refugee youth or considered differences in language skill level as a possible important factor in decisions to include. Drawing on these findings, the current study investigates the role of German language skills for adolescents’ inclusion or exclusion decisions in hypothetical intergroup scenarios including refugee youth.

Social Exclusion of Refugees

Social exclusion based on group membership, such as ethnicity, is a serious problem faced by many immigrants and refugees in a new host country (Minority Rights Group International, 2010). Further, minority groups are especially likely to be confronted with stereotypical mindsets and behavior, which can result in the exclusion of immigrants and refugees (Killen, Mulvey, & Hitti, 2013). Social exclusion during the acculturation process is a significant acculturative stressor (Verkuyten & Thijs, 2002; Ward, Furnham, & Bochner, 2005). In this context, children and adolescents from refugee families deserve special attention because they face challenging stressors: They have had to leave their country, their homes, their friends, and often their families. They are also faced with the difficulty of adapting to a new environment and culture. Many refugees have endured traumatic experiences in their home countries or during their transit to the new host country and most have been exposed to numerous psychological stressors (Ruf, Schauer, & Elbert, 2010). As a result, youth from refugee families can be considered to be an at‐risk population (Gavranidou, Niemiec, Magg, & Rosner, 2008). Thus, social exclusion might have an even greater impact on them than on other groups and constitute a problematic stressor for this special group.

Generally, social exclusion can have severe consequences for health and well‐being. Several studies have demonstrated that children and adolescents who are exposed to social exclusion experience anxiety, low self‐esteem, health and behavioral problems, as well as difficulties in interpersonal relationships (Araújo & Borrell, 2006; Eisenberger, Lieberman, & Williams, 2003; Gazelle & Druhen, 2009; Murray‐Close & Ostrov, 2009; Sanders‐Phillips, Settles‐Reaves, Walker, & Brownlow, 2009; Tummala‐Narra, Alegria, & Chen, 2012). Further, facing social exclusion can have a negative impact on academic engagement and achievement (Buhs, Ladd, & Herald, 2006). Beyond this, intergroup social exclusion negatively influences the social and emotional development of other youth including those who exclude others and those who witness social exclusion (Hutchinson, 2012; Twenge, Baumeister, DeWall, Ciarocco, & Bartels, 2007). Children and adolescence who exclude someone or observe intergroup social exclusion might perpetuate prejudicial beliefs and discriminatory practices throughout adulthood (Abrams & Killen, 2014). Peer aggression, bullying and social exclusion also can influence the entire school climate by decreasing perceptions of school safety and by creating a climate of fear and disrespect (Garnett & Brion‐Meisels, 2017). Further, Park and Park (2015) examined whether observing social exclusion influences bystanders’ evaluations of the humanness of its victims and perpetrators. The results demonstrated that a victim of social exclusion can be perceived as less human by the bystanders. It is suggested that bystanders are more likely to blame victims and are more likely to have negative attitudes toward them. Thus, social exclusion has cascading negative outcomes for those who experience exclusion, who perpetuate exclusion, and who observe exclusion.

Social Exclusion from a Social Reasoning Developmental Perspective

Given the strong human need to belong (Leary & Baumeister, 2017) and the severe consequences of social exclusion, it is not surprising that adolescents generally reject social exclusion and describe it as morally unacceptable (Killen & Rutland, 2011). However, social exclusion is a central issue in the development of adolescents. In social situations, adolescents frequently experience exclusion or rejection from friends, peers, and peer groups or observe others engaging in exclusion (Killen & Rutland, 2011). Adolescents may experience unique challenges with exclusion, as peer relations become increasingly important during the adolescence (Hitti, Mulvey, & Killen, 2016).

In the current paper, we draw on the Social Reasoning Developmental (SRD) perspective (Rutland, Killen, & Abrams, 2010) that integrates elements from social identity development theories (Nesdale, 2004; Tajfel, 1979) and social domain theory (Turiel, 1983). This framework has been applied to social exclusion research in intergroup contexts by highlighting the use of justifications to understand how children and adolescents make sense of harmful social behaviors (such as social exclusion) within their social ecologies. Adolescents are faced with increasingly complex situations when they navigate their social world. These situations sometimes may lead them to make evaluations based on a desire to maintain the integrity of their group, and other times based on their moral considerations (Killen, Mulvey et al., 2013). The SRD perspective looks beyond the correlates of social exclusion to provide a deeper understanding of how individuals perceive and justify social exclusion. Those justifications and explanations are classified into subcategories from different social domains involving moral (concerns for fairness and equality), societal (group identity and group functioning), and psychological (personal choice) reasons (Hitti et al., 2016; Smetana, 2006).

Studies drawing on the SRD perspective have demonstrated that adolescents generally reject social exclusion as morally unacceptable because of the unfair nature of exclusion or the psychological harm that it can cause, but, when they do condone exclusion, it is often justified with reasons referring to smooth group functioning (Hitti, Mulvey, & Killen, 2011). In other words, they are also likely to evaluate exclusion as okay to protect the distinctiveness of their groups and groups’ norms (Mulvey, 2016). In line with this, researchers using the SRD perspective have shown how group processes and moral principles exist simultaneously when someone needs to decide about social inclusion in intergroup contexts (Killen & Rutland, 2011).

Interestingly, at times children and adolescents recognize that their own inclusion preferences may not align with those of their group. For instance, research has documented that, often, children and adolescents individually prefer to be inclusive, and think their group should be inclusive for moral reasons, but believe that their group will be more swayed by group loyalty and group functioning leading to greater expected exclusivity by groups (Mulvey, Hitti, Rutland, Abrams, & Killen, 2014; Mulvey & Killen, 2016). Importantly, expectations about group inclusivity are related to how participants think about if groups should be inclusive and if they, personally, will be inclusive (Mulvey, McMillian, Irvin, & Carlson, 2018; Mulvey, Boswell et al., 2018). Further, children's and adolescents’ social‐cognitive abilities, emotional understanding, awareness of moral and societal norms, and understanding of the information in the social context are factors that influence their inclusion decisions when balancing morality and group norms (Mulvey, 2016).

Whereas developmental science research has attended to the role of intergroup processes to understand roots of social exclusion, it remains unclear how other factors work together with group membership in explaining adolescents’ decisions and reasoning about social inclusion and exclusion.

Language as the Bridge to Inclusion

In intergroup contexts with immigrants or refugees, one possible factor that might intertwine with group membership is language. As many immigrants and refugees do not have host country language skills upon arrival, language barriers can lead to difficulties in social interactions and communication (McBrien, 2005) and might result in social exclusion. Language might matter for social inclusion and exclusion decisions in intergroup contexts for several reasons. It has been shown that language reflects a meaningful intergroup category (Kinzler, 2013; Kinzler, Shutts, Dejesus, & Spelke, 2009) and that it can be a basis for group‐based exclusion (Mulvey, Boswell et al., 2018). In turn, speaking the same language provides similarity to others, and research shows that youth tend to be rejected when they are perceived as marginally different from the others (Mendez, Bauman, & Guillory, 2012). Thus, language can increase one's likelihood of social acceptance by peers. Most theories of immigrant integration also propose that exposure to the host country's langauge facilitates integration through acculturation (Berry, 1997) as immigrants become more similar to the culture via language (Alba & Nee, 2009). And finally, language has a bidirectional relationship with intergroup contact (i.e., meaningful interaction between members of different social groups) and cross‐group friendship (Wright & Tropp, 2005). Intergroup contact theory argues that positive intergroup contact between in‐group and out‐group members can be fostered in settings which meet a few basic criteria: (1) Members of the two groups hold equal status during the social interaction; (2) it promotes cooperative interdependence between the members of the two groups; and (3) relevant authorities support the intergroup contact (Allport, 1954). Given that language is a primary marker of group membership, it is argued that language can significantly contribute to establishing basic conditions within social contexts (Wright & Tropp, 2005). More specifically, for immigrant and refugee youth, language is an important tool to foster positive contact and friendships with majority youths. Having better contact with peers, in turn, promotes engagement of minority youth in the host culture and supports the development of better language skills (Berry et al., 2006).

However, little is known about adolescents’ inclusion decisions and reasoning concerning refugees who do and do not speak the language of the host country. Given the serious consequences of social exclusion, it is a highly relevant topic in terms of youth's well‐being and social integration to investigate the role of language skills in decisions about social exclusion or inclusion in the context of intergroup processes. Further, this field of research is of great relevance because results may have important implications for educational practice and integration as well as for interventions in order to reduce intergroup conflicts in society.

Current Study

The current study focuses on adolescents’ social inclusion or exclusion decisions as a function of refugee status and language skills of hypothetical peers in order to better understand the conditions of social inclusion and integration of refugee youth. To understand whether and under which circumstances adolescents support inclusion in the context of intergroup processes, adolescents evaluated and reasoned about three hypothetical scenarios in which groups of adolescents are planning leisure time activities, and peers from their own country (Germany) or another country (Syrian refugees) want to join them. The protagonists from Syria were introduced as refugees who have come to Germany with their families. Additionally, the level of German skills of the Syrian protagonists was varied (poor versus good) to examine the role of language in the context of intergroup processes. The stories were similar except for the origin and language skills of the protagonist. In all scenarios, participants were asked whether they would let the protagonist join (own inclusion decision). Additionally, they were asked what they think, what their group would do (descriptive group inclusion evaluation), and what they think the group should do (prescriptive group inclusion evaluation). For each measure, participants were also asked to provide reasoning about their choice (Why?).

We focused on adolescence because it is a period in which intergroup tension is prevalent between majority and minority youth. During this period, youth are more likely to have sophisticated knowledge about intergroup processes, and their social horizons and knowledge about group dynamics widen. They also come to understand cultural similarity and dissimilarity, which might lead to the formation of in‐group and out‐group biases (Devine & Hughes, 2016). Further, with age adolescents are less likely to be supportive of humanitarian values, egalitarianism, and a personal sense of fairness when evaluating the rights of minority groups compared to adults or younger children (Ruck & Tenenbaum, 2014; Tenenbaum & Ruck, 2012). Moreover, examining reasons for social inclusion and exclusion during this period is vital because peer affiliation and belonging is an essential need especially for youth, and as stated in the introduction, social exclusion can seriously undermine refugee youths’ mental well‐being and school adjustment (Correa‐Velez, Gifford, & Barnett, 2010; Wolke, Copeland, Angold, & Costello, 2013).

We chose Syrian refugees as our targeted out‐group because migration from Syria is very salient in Germany as Syrians represent the biggest group of incoming refugees (BAMF, 2018). Also in the educational system, Syrian students are increasingly prevalent. In 2016/2017, for instance, approximately 12% of all students in Germany were Syrian (Statistisches Bundesamt, 2017). In particular, against the backdrop of conflicting societal discourses in Germany, that is, “culture of welcome” vs. rising nationalism and increasing xenophobia (Holmes & Castañeda, 2016), it is necessary to examine factors that might influence social inclusion, especially in a country hosting large numbers of refugees.

Hypotheses

First, for participants’ decisions about inclusion or exclusion of peers, we expected generally high rates of inclusion (over 50%) in all three measures (own inclusion decision, descriptive group inclusion evaluation, and prescriptive group inclusion evaluation) (H1). This assumption is based on prior findings that adolescents generally evaluate social exclusion as wrong (Killen & Rutland, 2011) and that they reject exclusion based on race or ethnicity (Killen, Lee‐Kim, McGlothlin, Stangor, & Helwig, 2002).

Further, based on research on social categorization, we hypothesized that participants’ responses to our three measures would differ based on the origin of the protagonist (German vs. Syrian). More specifically, we expected participants to be less inclusive concerning out‐group members (i.e., the two Syrian protagonists) (H2). Considering prior findings of language being a meaningful intergroup category, we additionally hypothesized that their decisions would differ based on the levels of German skills of the Syrian protagonists in the stories (Syrian with good German skills vs. Syrian with poor German skills). More specifically, we expected that participants would be less inclusive concerning the Syrian protagonist with poor German skills in comparison with the Syrian with good German skills (H3).

Further, we also expected that participants may expect their group to be less inclusive (descriptive group inclusion evaluation) than they believe the group should be (prescriptive group inclusion evaluation) or then they expect they, individually would be (H4). This is based on prior research which has documented that adolescents often assume that their peers will be less inclusive than they should be or than the participants expect they will be (Mulvey et al., 2014; Mulvey, Boswell et al., 2018).

As some prior studies have documented, gender differences in inclusion (Killen & Stangor, 2001), we expected girls to be more inclusive than boys (H5).

In terms of social reasoning, we hypothesized that participants would use different forms of reasoning, depending on the target of inclusion (H6). More specifically, we expected that participants would be more likely to use group functioning, language‐related reasons, and positive attitudes toward foreigners as justifications for inclusion decisions concerning the Syrian protagonist with poor German skills compared to other two targets. Further, we expected that participants would be less likely to state that there is no reason not to include the protagonist as a justification for including the Syrian protagonist with poor German skills more so than in the two other stories.

Beyond these hypotheses, we considered that some of the participants’ personal background variables might be relevant for the aforementioned relations. For example, we controlled for intergroup contact, that is, contact between persons from the host society and immigrants, as this is an important impact factor that has been shown to reduce hostility and prejudices in general (Pettigrew & Tropp, 2006) and in the context of refugees (Becker, Ksenofontov, Love, & Borgert, 2018). Further, we controlled for family migration background and language spoken at home, as having a migration history in the family might influence the perception of others who have come into the country and may influence how participants perceived peers who do not speak German as they may be more familiar with language diversity through their families. We also controlled for age, as there may be developmental differences across adolescence, but we did not have an adequate sample for each age‐group to test for developmental differences. Prior research, however, has shown that, with age, adolescents are more likely to prioritize group functioning when reasoning about social exclusion (Mulvey, 2016).

Given these considerations, we controlled for participant's age, family migration background, the language spoken at home, and their intergroup contact1 in all analyses.

We had aimed to include intergroup contact as a factor in the analyses. But due to very unequal cell sizes (only 14% did not report intergroup contact), we decided to use it as a covariate.

Method

Participants

The study included 100 adolescents (M age = 13.65 years, SD = 1.93, range 10–17 years) attending grades 5 to 10 of a high school (Gymnasium) in a small city in Northern Germany. The sample was approximately evenly divided by gender (51 female, 49 male). 39% of the participants had a migration history in the family (i.e., at least one parent coming from another country than Germany). 20 families had their roots in Eastern European countries (such as Russia, Romania, Uzbekistan, Poland, Ukraine, Hungary, Lithuania). Nine families were from Middle Eastern countries (such as Iran, Syria, Lebanon, Palestine, Vietnam, Jordan), eight families from Middle and Southern European countries (such as Italy, Croatia, Turkey, France), one family came from China, and one from Vietnam. Almost three‐quarters (74%) of the participants indicated that they speak only German at home, 23% said that they speak German and another language at home, and 3% stated that they speak only another language than German at home. 86% of the participants stated that they know people who are refugees (intergroup contact).

Three participants were excluded from the analyses as their families were from Syria, and thus, the in‐group–out‐group manipulation would not have worked for them.

Informed parental consent was obtained for all participants.

Design and Procedures

Participants completed the paper–pencil assessment in class under the guidance of a trained research assistant. All participants were presented three scenarios including social situations. In all three stories, there was a group of adolescents planning a leisure time activity (playing basketball, meeting at the beach volleyball pit, going to the public pool). In all three stories, one additional adolescent (target person) wants to join the group. The origin and the German skills of the target person were varied across stories with the targets representing: (1) a German, (2) a Syrian with good German skills, and (3) a German with poor German skills. The order of the three stories was varied randomly. The names of protagonists in all stories matched the gender of the participant.

Measures

In all scenarios, three different measures were applied: (1) own inclusion decision (“Would you let xxx join?”), (2) descriptive group inclusion evaluation (“What do you think, will your group let xxx join?”), and (3) prescriptive group inclusion evaluation (“Do you think your group should let xxx join?”). Participants answered all three questions with a six‐point Likert‐type scale (1 = Not At All to 6 = Very Much). For each measure, participants were also asked to provide reasoning about their choice (Why?).

Coding of Reasoning

A coding system was established drawing on prior research on inclusion based on language (Mulvey, Boswell et al., 2018) as well as on categories developed from the surveys themselves. For the present analyses, the five most commonly used categories were included: group functioning (e.g., “it's good for the group to have more players”), language (“she doesn't speak German”), positive attitudes toward foreigners (“we are happy to get in contact with new people”; “I want to help foreign people”), no reason for exclusion (“I can see no reason why he should not join us”), and psychological information about the target (“he might be a very good player”). Coders coded up to three relevant justifications for each statement. If the participant used only one code, this was assigned a value of 1. If they used two codes, each was given a value of .5. If three codes were used, each was given a value of .33. Coding was completed by two independent coders. On the basis of 25% of the interviews, interrater reliability was high, with Cohen's kappa = .83.

Demographic Variables

We assessed participants’ age, family migration background, language spoken at home, and intergroup contact as control variables. The family migration background was assessed by asking an open‐ended question regarding where the participants’ father and mother come from (separately for each parent). A participant was considered having a family migration background if he or she had at least one parent who came from another country than Germany (N = 39).

Further, to assess whether participants speak a language other than German in their families, we asked them “Which language do you speak at home?” Participants could indicate one or more languages. This variable was coded dichotomously so that participants indicating that they spoke only German at home received a 0 (N = 77), while participants who indicated that any other language was spoken at home received a 1 (N = 23).

In order to assess intergroup contact, we asked a dichotomous question whether they knew people who are refugees (yes, no) (86 participants indicated contact with refugees). Age was assessed by asking participants to report their birthday.

Results

Analyses on Inclusion and Exclusion

Data were analyzed using repeated measures ANCOVAs with participants’ intergroup contact, language spoken at home, family migration background, and age as covariates in all analyses. In order to test for differences in inclusion decisions for the three different target conditions across the three questions (H1‐5), a 2 (gender: male, female) × 3 (target person: German, Syrian with good German skills, Syrian with poor German skills) × 3 (measure: own inclusion decision, descriptive group inclusion evaluation, prescriptive group inclusion evaluation) was conducted with repeated measures on the last two factors. Results revealed a significant interaction between question type and target, F(4, 536) = 3.48, p = .018,  = .02. Pairwise comparisons revealed that for all three questions, participants’ responses differed significantly between the Syrian refugee with poor German skills and the other two targets (Syrian refugee with good German skills and German), all ps < .004. Specifically, participants were less inclusive, expected their group to be less inclusive, and thought the group should be less inclusive toward the Syrian refugee with poor skills than either of the other two targets. In all three measures, there were no differences between the German target and the Syrian refugee with good German skills. See Table 1 for all means and standard deviations, and Table 2 for correlations between the inclusion measures and the control variables.

Table 1

Means and Standard Deviations for All Three Target Persons on All Three Measures

Measure	German M (SD)	Syrian with good German skills M (SD)	Syrian with poor German skills M (SD)
Own Inclusion Decision	5.54 (.63)	5.51 (.86)	4.93 (1.26)
Descriptive Group Inclusion Evaluation	5.27 (.96)	5.08 (1.01)	4.40 (1.29)
Prescriptive Group Inclusion Evaluation	5.49 (.69)	5.54 (.86)	5.10 (1.20)

Table 2

Correlations of All Likert‐type Questions and Covariates

Variables	1	2	3	4	5	6	7	8	9	10	11	12
1. Age	–
2. Own Inclusion Decision (German)	.03	–
3. Descriptive Group Inclusion Evaluation (German)	−.06	.45***	–
4. Prescriptive Group Inclusion Evaluation (German)	.04	.82***	.47***	–
5. Own Inclusion Decision (Syrian with good German skills)	.09	.59***	.19	.49***	–
6. Descriptive Group Inclusion Evaluation (Syrian with good German skills)	.08	.40***	.68***	.43***	.36***	–
7. Prescriptive Group Inclusion Evaluation (Syrian with good German skills)	.05	.51***	.24*	.46***	.89***	.46***	–
8. Own Inclusion Decision (Syrian with poor German skills)	.15	.38***	.19	.35***	.60***	.35***	.64***	–
9. Descriptive Group Inclusion Evaluation (Syrian with poor German skills)	.30**	.30**	.43***	.29**	.32***	.67***	.38***	.60***	–
10. Prescriptive Group Inclusion Evaluation (Syrian with poor German skills)	.15	.38***	.21*	.36***	.69***	.42***	.69***	.86***	.60***	–
11. Family Migration History	−.16	−.02	−.08	−.09	.03	−.12	.05	.01	−.12	−.12	–
12. Language Spoken at Home	−.03	.02	.09	−.05	.01	.01	.04	−.02	−.04	−.05	.63***	–
13. Intergroup Contact	−.10	.01	−.08	.02	.06	.03	.04	.26**	.10	.27**	−.07	.06

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

Additionally, pairwise comparisons revealed that for all three targets, participants were more inclusive in their own decisions and concerning what they thought the group should do compared to what they expected their group would do, ps < .001 for both Syrian targets and ps < .05 for the German target (see below). Specifically, for the German target, participants were significantly less likely to expect the group to be inclusive than their own inclusion expectation (p = .008) or than they thought the group should be (p = .029). For the Syrian target with good German skills, participants were significantly less likely to expect the group to be inclusive than their own inclusion expectation (p < .001) or than they thought the group should be (p < .001). Finally, for the Syrian refugee with poor German skills, participants were also significantly less likely to expect the group to be inclusive than their own inclusion expectation (p < .001) or than thought the group should be (p < .001).

Further, there was a between‐subjects effect of gender, F(2,268) = 47.63, p < .001,  = .15. Pairwise comparisons revealed that female participants (M = 5.53, SD = .67) were more inclusive than were male participants (M = 4.86, SD = .69), across all measures and across all targets.

Reasoning Analyses

Reasoning analyses were conducted on the proportional use of the targeted reasoning codes (all of which were used more than 10% of the time). The top codes were as follows: “group functioning,” “language,” “positive attitudes toward foreigners,” “psychological information about the target,” and “no reason for exclusion.” Data were analyzed using repeated measures ANCOVAs with participants’ intergroup contact, language spoken at home, family migration background, and age as covariates in all analyses. ANOVA frameworks are appropriate for repeated measures reasoning analyses because ANOVAs are robust to the problem of empty cells, whereas other data analytic procedures require cumbersome data manipulation to adjust for empty cells (see Posada & Wainryb, 2008, for a fuller explanation and justification of this data analytic approach).

Reasoning for own decision

In order to test for differences in reasoning about the participants’ own inclusion decisions for the three different targets, a 2 (gender: male, female) × 3 (target person: German, Syrian with good German skills, Syrian with poor German skills) ANCOVA was run for proportional use of each code. For group functioning and psychological information about the target, there were no differences in the use of reasoning for each target (German, Syrian with good German skills, Syrian with poor German skills), see Table 3.

Table 3

Means and Standard Deviations of Reasoning for Different Types of Evaluations

Reasoning Type	Own Inclusion Decision			Descriptive Group Inclusion Evaluation			Prescriptive Group Inclusion Evaluation
	German	Syrian (good German skills)	Syrian (poor German skills)	German	Syrian (good German skills)	Syrian (poor German skills)	German	Syrian (good German skills)	Syrian (poor German skills)
Group Functioning	.23 (.39)	.22 (.39)	.27 (.39)	.28 (.44)	.19 (.37)	.24 (.30)	.21 (.40)	.20 (.36)	.20 (.36)
Language	.03 (.15)	.05 (.17)	.12 (.23)	.01 (.10)	.09 (.25)	.12 (.27)	.04 (.19)	.07 (.22)	.05 (.14)
Positive Attitudes toward Foreigners	.02 (.12)	.05 (18)	.16 (.32)	.04 (.11)	.10 (.29)	.13 (.31)	.09 (.28)	.14 (.35)	.20 (.36)
Psychological Information	.20 (.37)	.16 (.35)	.14 (.33)	.16 (.30)	.08 (.27)	.12 (.30)	.14 (.33)	.04 (.33)	.11 (.28)
No Reason for Exclusion	.22 (.40)	.18 (.37)	.07 (.23)	.15 (.34)	.10 (.30)	.08 (.27)	.22 (.40)	.13 (.32)	.11 (.31)

For positive attitudes toward foreigners, there was a significant effect for target, F(2, 271) = 8.89, p < .001,  = .06. Specifically, this revealed that participants used significantly more references to positive attitudes toward foreigners when reasoning about including the Syrian refugee with poor German skills than the German target (p < .001) or than the Syrian refugee with good German skills (p = .006).

For language, there was a significant effect for target, F(2, 271) = 6.05, p = .003,  = .04. Specifically, this revealed that participants used more reference to language when reasoning about including the Syrian refugee with poor German skills than the German target (p = .002) or than the Syrian refugee with good German skills (p = .053).

For justifications stating that there was no reason not to include the target, there was a significant effect for target: F(2, 271) = 4.29, p = .015,  = .03. Specifically, this revealed that participants used significantly more references to the absence of a reason to exclude when justifying their decision about the German target than about the Syrian with poor German skills (p = .018).

Reasoning for descriptive group inclusion evaluation

In order to test for differences in reasoning about the participants’ expected inclusion by the group for the three different targets, a 2 (gender: male, female) × 3 (target person: German, Syrian with good German skills, Syrian with poor German skills) ANCOVA was run for proportional use of each code. For group functioning, positive attitudes toward foreigners, psychological information about the target, and no reason for exclusion, there were no differences in the use of reasoning for each target (German, Syrian with good German skills, Syrian with poor German skills), see Table 3.

For language, there was a significant effect for target, F(2, 244) = 5.83, p = .003,  = .04. Specifically, this revealed that participants used significantly more references to language when reasoning about including the Syrian refugee with poor German skills than the German target (p = .003).

Reasoning for prescriptive group inclusion evaluation

In order to test for differences in reasoning about the participants’ expectations about who the group should include for the three different targets, a 2 (gender: male, female) × 3 (target person: German, Syrian with good German skills, Syrian with poor German skills) ANCOVA was run for proportional use of each code. For positive attitudes toward foreigners and no reason for exclusion, there were no differences in the use of reasoning for each target (German, Syrian with good German skills, Syrian with poor German skills), see Table 3.

For group functioning, there was a significant interaction effect for gender and target, F(2, 230) = 3.73, p = .026,  = .03. Specifically, this revealed that there were no gender differences in use of group functioning reasoning for either of the Syrian targets, but male participants (M = .36, SD = .54) were more likely than were female participants (M = .02, SD = .60) to reference group functioning when reasoning about the German target, p < .001.

For language, there was a significant effect for gender, F(2, 230) = 10.61, p = .001,  = .04. Specifically, this revealed that female participants (M = .10, SD = .26) used significantly more references to language when reasoning about if the group should include the targets than did the male participants (M = .02, SD = .10).

For justifications focused on psychological information about the target, there was a significant effect for target: F(2, 230) = 4.69, p = .01,  = .03. Specifically, this revealed that participants used significantly more references to psychological information about the target when justifying their decision about the German target than about the Syrian with good German skills (p = .012). The difference between the German target and the Syrian with poor German skills approached significance, but was not significantly different (p = .066).

Discussion

The current study investigated adolescents’ inclusivity when considering peers who are native Germans, refugees with good German skills, and refugees with poor German skills. This research is important, given the increasing rates of immigration globally and the rising need of countries to accept refugees fleeing violence and political upheaval in their home countries (Hodes et al., 2018). Our findings provide new insight into adolescents’ cognition concerning inclusion of refugees. Namely, the findings document important differences in individual and group evaluations and results indicate that language may be the key to positive social integration for refugees. Whereas prior research documented the crucial role of language skills for integration in terms of academic achievement (Duong et al., 2016; Stanat & Christensen, 2006) or labor market entry (Auer, 2018; Esser, 2006), the current study demonstrates the importance of language skills for integration during social interactions and for inclusion of refugees in peer groups.

As expected, we found that while adolescents generally expected that they and their peers would be highly inclusive (H1), there were differences in the degree of inclusion depending on the target. Adolescents were significantly less inclusive toward the Syrian refugee with poor German skills than of either the German target or the Syrian refugee with good German skills (H3). Moreover, this pattern held across all measures: Participants expected that they, themselves, would be less inclusive, that their peers would be less inclusive, and that their peers should be less inclusive of the Syrian refugee with poor German skills than of the other two targets. Although we expected less inclusion for both Syrian refugees with good and poor language skills (H2), we found that adolescents’ inclusivity was based on language skills (H3), rather than status as German native or a refugee. This is an important new insight as it suggests that interventions to integrate refugees should target: (1) the language skills of the refugees as quickly as possible and (2) target the native speakers’ perceptions that language will serve as an impediment to successful social interaction. Interestingly, adolescents were less willing to include the target with poor German skills, although the activities in the scenarios (athletic activities) did not really require language skills. This might confirm the idea presented in the introduction that speaking the same language increases perceived similarity and reduces the in‐group–out‐group distinction. It is also important to note that we did not provide information about the German skills of the German target, but we can assume that participants believed that a native German would have good German skills. It may be interesting to replicate this study with Germans who vary in terms of their German skills (for instance, because of hearing impairment or a learning disability).

In addition, as expected, we documented differences based on the type of measure, with participants, generally indicating that they would be more inclusive than they expected their peers to be (H4). They also thought their peers should be more inclusive than they thought they themselves would be. This pattern of differences was documented for all three groups, including the native German, the Syrian with good German skills, and the Syrian with poor German skills. This is in line with prior research that documents that adolescents often assume that their group will be less inclusive than they would be, individually (Mulvey & Killen, 2017; Mulvey et al., 2014; Mulvey, Boswell et al., 2018). Further, it is important to document the many contexts in which adolescents assume that their peers will be less inclusive than they should. This also highlights the importance of bullying and inclusion interventions targeting these misconceptions. While adolescents frequently assume that their peers will not be inclusive, data indicate that children and adolescents are very inclusive, even of those who do not speak the same language as them (Mulvey, Boswell et al., 2018). Interventions that focus on inclusion of refugees are particularly important, as exclusion is linked to a host of negative outcomes including mental health concerns (Buhs & Ladd, 2001; Buhs et al., 2006), and as refugees frequently struggle with mental health needs due to prior trauma, victimization, and exposure to violence (Jefee‐Bahloul, Bajbouj, Alabdullah, Hassan, & Barkil‐Oteo, 2016; Nicolai, Fuchs, & von Mutius, 2015). Our results imply that language might be a crucial key to integration and that programs should focus on immediate acquisition of language skills of the respective host country.

Moreover, as expected we found gender differences in the participants’ inclusivity (H5). In all scenarios and across all measures, female participants were more inclusive than male participants. This effect was an overall between‐subjects effect and aligns well with other research that has documented that females are often more inclusive than are males (Killen, Rutland, Abrams, Mulvey, & Hitti, 2013; Killen & Stangor, 2001; Mulvey & Killen, 2017).

The reasoning results add additional insight into adolescents’ underlying motivation and cognition when including or excluding peers who are refugees or Germans. As expected, results documented that participants used different forms of reasoning depending on the target of inclusion (H6). Specifically, participants more frequently referenced positive attitudes toward foreigners when justifying their own inclusion decisions for the Syrian target with poor German skills than for the other two targets. This is important as it suggests that German adolescents recognize the importance of positive intergroup attitudes. Prior research that sampled adolescents and adults documented that even though many Germans are uncomfortable with refugees, particularly unaccompanied minors, they still were very supportive of providing educational opportunities to refugee children and noted the importance of learning German (Plener, Groschwitz, Brähler, Sukale, & Fegert, 2017). More generally, research on positive intergroup attitudes demonstrates that intergroup contact can lead to reductions in prejudice and bias and increase in positive attitudes (Allport, 1954; Tropp & Prenovost, 2008) such as those expressed by our participants. In the current study, intergroup contact was used as a control variable as there was little variation in reported contact (over 85% of participants reported contact with refugees). However, future research should aim to sample both participants who have had opportunities for intergroup contact with refugees and those who have not, in order to better understand how widespread these positive attitudes are. It would also be useful to measure not only the opportunity for intergroup contact, but also the quality of the contact (Crystal, Killen, & Ruck, 2008; Ruck, Park, Killen, & Crystal, 2011). This is important as not all situations of intergroup contact may be positive.

While participants frequently referenced positive attitudes toward foreigners when reasoning about their own decision to include the Syrian refugee with poor German skills, there were no differences in their reasoning about the group's expected decision or what the group should decide in terms of reasoning about positive attitudes toward foreigners.

In terms of language, they also referenced language skills more often when talking about their own inclusion and the group's expected inclusion of the Syrian refugee with poor German skills than the other two targets. Thus, in line with the Likert‐scale findings, participants did believe that language skills may be a barrier when interacting with refugees who do not speak German well. Research on language acquisition among refugees is scant, but findings indicate that language proficiency can take many years and may be complicated by issues such as posttraumatic stress disorder (Kaplan, Stolk, Valibhoy, Tucker, & Baker, 2016). Thus, it may be helpful for interventions to highlight the gradual nature of language acquisition for native youth and to encourage adolescents to recognize the many ways they can interact that do not require strong language skills (such as through athletic activities). Interestingly, when justifying what the group should do, female participants were also, generally, more likely to reason using references to language than were male participants. This finding warrants further inquiry, as it may be that females use language differently than do males in leisure activity contexts, although prior research documents that men and women do not differ in the rates of language used in sporting contexts (Sullivan, 2004).

Reasoning analyses also revealed a very intriguing pattern regarding justifications that indicate that there was no reason to exclude the target. This “no reason for exclusion” justification was used more often to support including the German speaker and the Syrian speaker with good German skills, but was used infrequently to justify responses regarding the Syrian with poor German skills. This may suggest that participants were more likely to attribute reasons to their exclusion of the Syrian with poor German skills (such as language differences).

Finally, when participants were considering if the group should include the targets, they were more likely to reference psychological information for the German target than for the Syrian with good German skills. This may suggest that participants had a harder time thinking about the psychological traits of refugees. Recent research has documented that perspective‐taking abilities may be impaired when asked to take the perspective of refugees, with data documenting that for instance Turkish children (aged 9–13 years) were less accurate when ascribing mental states to Syrian targets compared to Northern European targets and Turkish targets. The results also demonstrated that prejudice and perceived realistic threat significantly predicted lower mental state understanding when children attribute mental states to Syrian refugee story characters (Gönültaş, Selçuk, Ruffman, Slaughter, & Hunter, 2019). This indicates that future research should examine more carefully children's and adolescents’ ability to think about the mental or psychological state of others when making inclusion decisions involving out‐group members.

Limitations

While the current study provides novel insight into adolescents’ cognition concerning inclusion of refugees, there are some limitations of the project. The current study used hypothetical scenarios to measure expectations about social exclusion and inclusion. While the advantage of this approach is that confounding variables could be carefully controlled and participants had the opportunity to provide reasoning to support their decisions, authentic situations are often much more complex than hypothetical scenarios. However, recent research demonstrated that participant's responses in hypothetical exclusion scenarios correspond with their authentic decisions in behavioral experiments (Mulvey, Boswell et al., 2018), which provides support for the use of hypothetical scenarios in the context of social exclusion.

Further, in the current study, the three protagonists were not randomized across the different scenarios. While participants did not differ in their reasoning about the different activity conditions, it would be helpful for future research to extend these findings by randomizing across types of activities and testing additional activity choices.

Moreover, the current findings should be extended and replicated with target persons from other countries of origin, and migrants in general (not only refugees). Further, future studies should aim to carefully sample participants who vary based on migration background and intergroup contact in order to examine differences based on these variables. It would also be important for future research to directly measure intergroup attitudes, such as prejudice and bias, in order to more carefully understand factors related to inclusivity. Finally, while the current study examined a cross section of adolescents from 10 to 17 years of age, future research should compare younger and older adolescents to identify whether there are age‐related changes in inclusion, as would be predicted by the Social Reasoning Development model (Rutland et al., 2010). It may also be helpful to follow participants longitudinally, as they potentially experienced increased opportunity for contact over time, to document changes as adolescents gain experience with Syrian refugees.

Additionally, this research investigated only language as one important variable among many potential others which could be relevant for the inclusion of refugee peers. However, refugees often differ on many dimensions, for instance, religion, ethnicity, and socioeconomic status, in addition to the language they speak. Recent research has examined the intersection of refugee status, gender, and disability, with results suggesting the importance of examining multiple types of discrimination that may be faced by refugee youth who are seeking inclusion (Bešić, Paleczek, & Gasteiger‐Klicpera, 2018). It would be helpful for future research to draw on intersectionality theories (Crenshaw, 1991) in order to more carefully understand the impact of different, intersecting dimensions of one's identity that may play a role in inclusion decisions.

The findings of the current study highlight the importance of language skills for peer inclusion in the context of intergroup processes. It was the language that mattered for inclusion decisions. Nationality or refugee status alone did not make the difference, but the level of German skills elicited differences in participants’ answers. This demonstrates how important language skills are for social inclusion and in a broader view for integration. This has strong political and policy implications: Ultimately, language may be a critical bridge to social inclusion and interventions should attend to the very important role that language may play in fostering positive intergroup experiences. Thus, stakeholders, including policy‐makers and educators, should aim to provide language instruction to refugees immediately upon arrival and the education system should put a heavy focus on the language needs of refugees to help ensure smooth integration with peers and inclusion into social life in the host country.