Introduction

A social cognitive domain model, known as the social domain approach (Killen and Rutland 2011; Smetana et al. 2014; Turiel 2014), has provided both the theoretical approach and methodology for much of the recent work on children’s evaluations of interracial peer exclusion (Abrams and Killen 2014; Killen et al. 2013). Investigators using this approach, combined with developmental social identity theories (e.g., Abrams and Rutland 2011; Nesdale 2004) have found that children and adolescents use different forms of social reasoning to assess social issues entailing social exclusion, prejudice, and discrimination (Killen and Cooley 2014; Ruck and Tenenbaum 2014). Three distinct domains or categories of knowledge have been identified, including the moral (justice, rights, fairness, and equality), social conventional (social norms, group functioning, and customs) and the personal or psychological (personal decision-making, individual prerogatives) (see Nucci 2001; Turiel 1998, 2002). For example, with respect to evaluations of exclusion, children may use moral (such as focusing on fairness) or social-conventional reasoning (such as focusing on the importance of maintaining group functioning) when required to provide explanations and justifications concerning the acceptability of exclusion (see Killen, et al. 2002; Killen, Henning, Kelly, Crystal, and Ruck 2007; Killen and Stangor 2001). What has been important to demonstrate is that children and adolescents use different forms of reasoning to evaluate social exclusion as a function of the context. For example, adolescents may use moral reasoning, such as the wrongfulness of discrimination to reject racial exclusion while, at the same time, use social conventional reasoning to justify or condone gender exclusion. Thus, multiple forms of reasoning are often brought to bear on complex social issues such as exclusion and discrimination.

Employing a social-cognitive domain approach, researchers (Crystal et al. 2008; Killen et al. 2007) have examined judgments of exclusion in interracial peer situations (e.g., friendship, sleepover, and dating) among majority status and minority status children and adolescents attending racially and ethnically heterogeneous suburban middle-income schools. For situations involving the exclusion of a child from a club or group because of their race or ethnicity, with increasing age participants were more likely to use social-conventional rationales based on maintaining group functioning to justify exclusion. In addition, the findings indicated that, in terms of estimations of the frequency of exclusion, minority status children were more likely than majority status or dominant group children to report that race based exclusion occurs more often than non-race or group functioning based exclusion. These findings highlight the significance of racial/ethnic background, social experience, and context on children’s and adolescents’ interpretations and expectations of interracial exclusion.

Much of the germane available research examining Allport’s (1954) intergroup contact theory, in both U.S. and non-U.S. settings, with children and adolescents has focused on the social and developmental benefits of intergroup contact that results from school diversity such as prejudice reduction and improved intergroup attitudes among dominant and minority status group students (Aboud et al. 2003; Feddes et al. 2009; Killen et al. 2007; Killen and McKown 2005; Turner et al. 2007; Tropp and Prenovost 2008; Verkuyten 2008). However, intergroup contact has been found to have less of a positive influence on minority status group members, thus suggesting that intergroup contact may function differently for ethnic majority status and ethnic minority status children and adolescents (Binder et al. 2009; Feddes et al. 2009; Tropp and Pettigrew 2005; Tropp and Prenovost 2008; Verkuyten 2008). For ethnic minority youth, positive intergroup contact may possibly lead to different outcomes than those primarily documented for ethnic majority or dominate-status youth (e.g., prejudice reduction). The current study attempts to address this issue by examining the influence of intergroup contact on evaluations of exclusion in ethic minority youth.

To our knowledge, only one published study has examined intergroup contact and evaluations of interracial exclusion in a solely ethnic minority U.S. sample. Ruck et al. (2011) investigated evaluations of race-based peer exclusion in a sample of low-income urban ethnic minority 9, 12, and 15 year-olds (African American and Latino/a) attending predominately racial and ethnic minority public schools. Not surprisingly, the overwhelming majority of ethnic minority children and adolescents evaluated racial exclusion as “very wrong”. Also, older participants reported higher estimates of the frequency or occurrence of racial exclusion than their younger counterparts. In addition, the level of intergroup contact significantly predicted low-income participants’ evaluations of the wrongfulness of racial exclusion. Specifically, ethnic minority students reporting high levels of intergroup contact were more likely to evaluate interracial exclusion as wrong than their counterparts with low levels of intergroup contact. Taken together, these findings suggest that, for ethnic minority group youth, positive intergroup contact with majority group peers may also lead to differences in terms of how they evaluate interracial discrimination and prejudice rather than just leading to positive intergroup attitudes. What remains unanswered is whether differing levels of intergroup contact among African American youth lead to differences in how they evaluate and reason about interracial exclusion, including their awareness of what it is about race that makes dominant group members uncomfortable in interracial encounters.

The Present Study

In the current study, we examined African American children’s and adolescents’ intergroup contact and evaluations about interracial peer exclusion and discrimination in a sample of suburban and urban youth. We focused specifically on African American children and adolescents given that they experience high levels of racial discrimination (Brown 2008; Fisher et al. 2000; Romero and Roberts 1998; Sellers et al. 2006). In addition, by interviewing African American students attending heterogeneous middle-class suburban schools and their counterparts enrolled in predominately homogeneous low-income urban schools, the current investigation responds to calls for research addressing within-group variability in ethnic minority populations (Garcia Coll et al. 1996; McLoyd 1998, 2006). Hence, examining African American children’s and adolescents’ reactions to and reasoning about interracial exclusion is an important and, yet, relatively underexplored area of developmental inquiry.

In the present study, 4th, 7th and 10th grade (9, 12, and 15 years of age) African American children and adolescents evaluated situations involving interracial peer exclusion. These three age groups were chosen for a number of reasons. First, this age range reflects children enrolled in three distinct types of U.S. school environments: elementary school (4th grade), middle school (7th grade), and high school (10th grade). Available research suggests that experiences with discrimination and prejudice are a social reality for minority children and adolescents across these age levels and grades (Fisher et al. 2000). Second, from a young age, children view racial exclusion as wrong based on moral reasons such as unfairness and discrimination (Killen 2007). Third, prior research on this topic in ethnically diverse middle-class samples has indicated a significant age-related increase during this period regarding the legitimacy of more complex forms of exclusion such as maintaining group functioning as a reason for exclusion as well as age-related increases in understanding when such acts of exclusion results in discrimination (Killen et al. 2007; Smetana 2006). Finally, interviewing participants from these three age groups will allow direct comparisons with recent studies using the same age groups of middle-income suburban minority children and adolescents (e.g., Crystal et al. 2008).

As with previous work on this topic, participants were individually interviewed to assess their judgments and evaluations of scenarios involving such exclusion. In order to make comparisons with recent developmental work on the topic, the present investigation employed the same three scenarios as used in previous studies (Ruck et al. 2011) and designed to match the types of intergroup contexts that children actually encounter in their daily interactions: intergroup friendship exclusion at school lunch (refusing to have lunch with a minority peer), at home (refraining from inviting a minority peer for a sleepover), and at school dance (refraining from inviting a minority peer to a school dance). The three scenarios named, Lunch, Sleepover, and Dance, respectively, were designed to represent familiar peer situations.

We were interested in how intergroup contact influenced African American participants in predominately homogeneous ethnic-minority urban schools and heterogeneous ethnic-majority suburban schools responded to situations involving interracial peer exclusion. For each scenario several possible reasons, including race-based, a specific non-race based reason, and a general reason pertaining to maintenance of group functioning, for the exclusion were described, leaving the protagonists’ intentions ambiguous.

As noted above, the available U.S. work examining interracial exclusion in children and youth has predominately focused on majority-minority comparisons with European American students compared to African American, Asian, and Latino/a students combined (see Crystal et al. 2008; Killen et al. 2002, 2007). However, based on past investigations of majority and minority children’s evaluations of exclusion (e.g., Crystal et al. 2008; Ruck et al. 2011), we were able to make the following general predictions.

Intergroup Contact-Related Hypotheses

In terms of attributions of racial motives, we predicted that students with low levels of intergroup contact would be more likely to view race as a reason for exclusion. This prediction was based on previous work with both heterogeneous and homogenous samples showing that with increasing contact children are more likely to attribute exclusion to non-racial motives (Crystal, et al. 2008; Killen et al. 2002; McGlothlin and Killen 2006). In addition, based on recent research with ethnic minority youth (e.g., Ruck et al. 2011), we expected that students with higher levels of intergroup contact would be more likely to view racial exclusion as wrong and provide lower estimates of the frequency of exclusion than their counterparts with lower levels of intergroup contact. Prior research examining the influence on intergroup contact on reasoning about race-based exclusion has shown that participants use moral reasoning when explaining the wrongfulness of such exclusion (Ruck et al. 2011). Yet, no prior research has examined the influence of intergroup contact on participants’ reasoning concerning non-race based and group functioning based exclusion. Because our peer encounters were interracial, however, we expected that intergroup contact would be related to reasoning for these situations due to the increased awareness of the impact that rejection messages have on ethnic minority peers. Yet, we did not differentiate race-based and group-functioning based exclusion for intergroup contact given the lack of previous research using these contexts.

Grade-Related Hypotheses

First, in terms evaluations of the wrongfulness of exclusion, we expected that for all three scenarios participants across all grades, would evaluate explicit race-based exclusion as more wrong than exclusion due to lack of shared interests and group functioning based exclusion. This expectation was based on the fact that, by early adolescence, ethnic minority children are aware of the pervasiveness of racial bias across a range of contexts (Fisher et al. 1998). Second, with increasing grade level, we predicted that participants would be less likely to evaluate non-race based exclusion and group functioning based exclusion as wrong. This prediction was based on the finding that, with age children become more aware that decisions for exclusion can involve a number of legitimate reasons (Killen and Stangor 2001). The third hypothesis concerned reasoning about exclusion. Based on previous research, we predicted that the majority of participants would employ moral reasoning when explaining the wrongfulness of racial exclusion in school settings and that social-conventional reasoning would be employed when explaining exclusion in the home setting (Crystal et al. 2008; Killen et al. 2007; Ruck et al. 2011). Fourth, based on previous studies indicating that older ethnic minority children typically experience more discrimination than their younger counterparts (Fisher et al. 1998; Szalacha et al. 2003), it was hypothesized that, with increasing grade African American students would provide higher estimates of the frequency of both race-based and non-race based exclusion. Finally, in terms of stereotype assessments we expected that with increasing grade participants would be more likely to consider the historical and social factors leading to the use of stereotypes and racism. This hypothesis was drawn from prior research with ethnic-minority children and adolescents suggesting that, with increasing age, their understanding of stereotypes becomes more elaborated or differentiated and is often linked to the origins of prejudice (McKown 2004; McKown and Weinstein 2003; Ruck et al. 2011).

Method

Participants

The sample included 158 African American children in 4th, 7th and 10th grades attending urban and suburban public schools from the New York City and greater metropolitan Washington, DC areas, respectively. The New York City metropolitan area participants were recruited from four low-income urban predominantly minority public schools. The urban sample consisted of 22 4th grade students (12 girls and 10 boys; M = 9.32, SD = .48), 13 7th grade students (10 girls and 3 boys; M = 12.46, SD = .52), and 28 10th grade students (21 girls and 7 boys, M = 15.39, SD = .79). Overall, the urban schools were 55 % African American, 31 % Latino, 8 % European American, and 6 % Asian. The majority of urban children taking part in the study were from low-income families. Seventy percent of the children attending these schools were eligible for free and reduced meals programs.

The suburban sample was recruited from thirteen mixed-ethnicity suburban public schools in the greater metropolitan Washington, DC area. The suburban sample consisted of 21 4th grade students (12 girls and 9 boys; M = 9.98, SD = .41), 34 7th grade students (28 girls and 6 boys; M = 12.78, SD = .93), and 40 10th grade students (23 girls and 17 boys, M = 15.98, SD = .66). The suburban schools were 60 % European American, 14 % African American, 5 % Latino, 12 % Asian and 8 % biracial. All suburban children were from low-middle and middle-income families, with approximately 30 % being eligible for free and reduced meals programs. For both samples only students who self-reported as African American were included in the analyses.

Procedure and Measures

Written parental informed consent (mean response rate approximately 70 %) and child assent were obtained for all participants taking part in the study. Trained female research assistants, matching the participants’ race/ethnicity for the majority of the children, and blind to the specific research hypotheses, individually interviewed participants in a quiet room or area at their school.

The Social Reasoning about Exclusion Interview (Killen et al. 2007) consisted of three short scenarios, each representing a different interracial social context in which exclusion occurred. The pre-established order of story presentation was based on pilot testing and previous developmental work on children’s evaluations of exclusion (see Crystal et al. 2008; Killen et al. 2002, 2007, 2010). Therefore, the three scenarios, presented in the following order, were: Lunch (excluding a friend from a lunch table at school), Sleepover (excluding a friend from a sleepover party at home), and Dance (excluding a friend from a high school dance). For each scenario, participants were asked to evaluate the decision to exclude when it was based on race (What if X excludes because of race?), non-race (What if X excludes because of lack of shared interests/parental unfamiliarity/school rivalry?—for lunch, sleepover, and dance respectively), and group functioning (What if X excludes because the individual will not “fit in” with the group?). In order to capture the historical dimensions of racial exclusion in the United Sates, all three scenarios involved a European American child excluding an African American child. Similar scenarios (where a European American child excludes an African American child) have been used reliably in previous studies examining social reasoning about racial exclusion with participants from a variety of racial and ethnic groups (see Killen et al. 2002).

After the presentation of each scenario, participants were required to respond to a number of assessments:

  1. 1.

    Motive for exclusion (e.g., “why do you think Michael believes that he and Doug won’t have a much in common?”). Responses for the attribution of motive judgments were coded into three categories: race only (e.g., “It is because he’s Black”), non-race only (e.g., “Because they don’t really know each other”), and both race and non-race (e.g., “Doug’s Black and they don’t know him”). Responses for wrongfulness ratings (race-based, non-race and group functioning) ranged from 1 (“very, very good”) to 8 (“very, very bad”);

  2. 2.

    Wrongfulness of race-based exclusion (e.g., “what if Michael thinks they won’t have much in common because Doug is Black? How good or bad is that?”);

  3. 3.

    Wrongfulness of non-race based exclusion (e.g., “what if Michael thinks they won’t have much in common because Doug doesn’t like sports?”);

  4. 4.

    Wrongfulness of group functioning based exclusion (e.g., “what if Michael doesn’t invite Doug to lunch because he thinks Doug won’t fit in?”);

  5. 5.

    Justification for each type of exclusion (after each of the wrongfulness rating, students were asked to provide a rationale or reason as to why they gave that specific evaluation, e.g., “why did you give that response?”). Participants’ justifications for their wrongfulness ratings were analyzed using a coding scheme modified from previous research (Crystal et al. 2008; Killen and Stangor 2001). The coding categories were: Moral (subcategories: references to racial prejudice, references to fairness or empathy for the targeted individual), Social Conventional (subcategories: references to conforming to peer pressure, references to parental “unfamiliarity/wariness”, appeals to personal choice), Stereotypes (subcategories: references to appearance or skin color, references to affirming stereotypes), and Uncodable (“I don’t know”, “other”). Justifications were coded with a score of 1 indicating that the category was used or a score of 0 indicating that the category was not used;

  6. 6.

    Frequency estimations for race-based and non-race based exclusion (for each scenario interviewers asked participants “how often do you think kids your age might not invite someone to lunch because they are a different race?” and “how often do you think kids your age might not invite someone to lunch because they do not share the same interests?”). Participants’ responses to the frequency estimations ranged from 1 (“never”) to 5 (“always”); and

  7. 7.

    Stereotype assessment (responses to an open-ended question: “What is it about race that makes people uncomfortable?”). Responses were coded based on a system used in previous research (Ruck et al. 2011). The coding categories were: Affirming stereotypes (references to affirming or endorsing the use of stereotypes); Stereotype recognition (references to recognizing that others use stereotypes to make decisions); and Social contexts of stereotypes (references to awareness of the social factors and historical circumstances contributing to the manifestation of stereotypes)

Reliability coding was calculated on 30 % of the interviews by two independent raters trained on the coding systems. Cohen’s kappas ranged from .90 to 1.00. For all coding categories uncertainties or disagreements were resolved through discussion.

Developmental Intergroup Contact

At the end of the interview, all participants completed the Developmental Intergroup Contact Survey (Crystal et al. 2008) originally adapted from a 10-item Diversity Attitudes Questionnaire (DAQ) (Kurlaender and Yun 2001). Following Crystal et al. (2008) the 10 DAQ items were subjected to a principal axis factor analysis with varimax rotation (Kaiser normalization), which yielded a primary factor, explaining 34 % of the variance and consisting of six items. Those six items comprised our Intergroup Contact Scale and included: (1) How many students in your school are from racial or ethnic groups different from you own? (2) How often do you work on school projects and/or study with students from other racial/ethnic groups? (3) At school, how many friends do you have who are from a different racial or ethnic group than you? (4) Outside of school, how many friends do you have who are from a different racial or ethnic group than you? (5) In the neighborhood where you live, do you have neighbors from other racial or ethnic groups? and (6) How many of your friends from your neighborhood are from a different racial or ethnic group than you? Responses to these items ranged from one (“none”) to four (“many”), and were summed and then averaged to form the Intergroup Contact Scale, with a Cronbach’s α of .72. In order to include the Intergroup Contact Scale in the analyses, the scale was dichotomously split along the mean into groups of “low” and “high” intergroup contact.

Results

Data Analytical Plan

The data for intergroup contact and attributions of race as a motive for exclusion were dichotomous; thus, analyses for both were conducted using binary logistic regression. Follow-up tests were conducted using Bonferroni-adjusted paired comparisons. To examine possible differences between scenarios in terms of motives for exclusion, wrongfulness ratings, frequency estimations, justifications, and stereotype assessments repeated measure ANOVAs were used. Follow-up tests included univariate ANOVAs for between-subject effects using Bonferroni corrections in the post hoc analyses. In those cases where sphericity was not met, corrections were made using the Huynh–Feldt method. Since preliminary analyses indicated that gender was not highly significant it was omitted from the following analyses.

Intergroup contact

Binary logistic regression was performed on the intergroup contact scale with type of school, and grade serving as independent factors. Significant main effects were found for setting, Wald’s χ2 = 11.46, p < .001, η 2p  = .30. Follow-up univariate analysis indicated that suburban students (M = .64, SD = .48) had significantly higher levels of intergroup contact than urban students (M = .37, SD = .49). Since intergroup contact was highly correlated with whether participants attended suburban or urban schools the intergroup contact variable was employed as a proxy for type of school in order to increase the power and validity in subsequent analyses.

Attribution of Racial Motives

Individual binary logistic regressions were used to determine if attribution of motives differed for each of the three scenarios. The analysis revealed that significant differences were found for the sleepover scenario, intergroup contact emerged as a significant predictor of participants’ attribution of racial motives, Wald’s χ2 = 4.562, p < .05, Exp(B) = 2.07. Follow-up univariate analysis indicated that students with low levels of intergroup contact (M = .50, SD = .50) were significantly more likely to indicate that the story character’s race was the reason for exclusion than participants with high levels of intergroup contact (M = .34, SD = .48), p < .05.

To determine if there was a significant difference between the three types of scenarios and attribution of racial motives, a 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) × 2 (context: lunch, dance) univariate ANOVA was conducted. Analysis indicated significant differences between the lunch (M = .26, SD = .44) and sleepover (M = .42, SD = .49) scenarios, F(1, 154) = 8.91, p < .005, η 2p  = .06, between the lunch (M = .26, SD = .44) and dance (M = .29, SD = .46) scenarios, F(1, 155) = 14.67, p < .005, η 2p  = .09, and between the sleepover (M = .42, SD = .49) and dance (M = .29, SD = .46) scenarios, F(1, 155) = 9.15, p < .005, η 2p  = .06. Findings indicated that respondents were more likely to attribute the story-character’s exclusion from a sleepover party as being due to “race” than their being excluded from a school lunch setting or a school dance setting.

Wrongfulness of Race-Based Exclusion in School Settings

To examine how children and adolescents evaluated each of the three types of exclusion in the school context, a 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) × 2 (context: lunch, dance) ANOVA with repeated measures on the last factor was conducted on children’s wrongfulness ratings of the two scenarios where participants were asked to evaluate the wrongfulness of race-based exclusion. Means are presented in Table 1.

Table 1 Wrongfulness of exclusion ratings for three contexts and types of exclusion by grade and intergroup contact
Full size table

The vast majority of participants evaluated race-based exclusion as wrong (M = 7.29, SD = .94). There were no differences in terms of grade or type of scenario. There was an overall effect for intergroup contact, F(1, 149) = 5.95, p < .05, η 2p  = .04. Participants with high levels of intergroup contact were more likely to view race-based exclusion as wrong (M = 7.46, SD = 1.01) compared to those with low intergroup contact (M = 7.11, SD = 1.11) across the two scenarios. Follow-up analysis revealed that intergroup contact was a significant predictor of race-based exclusion in the lunch scenario, F(1, 149) = 9.25, p < .01, η 2p  = .06. For the lunch scenario, participants with high intergroup contact (M = 7.34, SD = .72) were significantly more likely to evaluate race-based exclusion as wrong as than their low level counterparts (M = 7.08, SD = 1.14). There were no significant differences for the dance scenario.

In addition, individual 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) × 3 (type of exclusion: race-based, non-race, group functioning) ANOVA with repeated measures on the last factor were conducted on participants’ wrongfulness ratings on all three types of exclusion for the school lunch and dance scenarios. For the lunch, F(2, 298) = 96.81, p < .001, η 2p  = .39, and dance, F(2, 298) = 62.84, p < .001, η 2p  = .29 scenarios. Findings indicated that children viewed race-based exclusion as more wrong than non-race and group functioning based exclusion in both scenarios.

Post-hoc analyses using the Bonferroni correction revealed significant between subject effects for intergroup contact and grade in the lunch scenario. Participants with high levels of intergroup contact (M = 7.55, SD = .720) were significantly more likely to rate racial exclusion in the lunch scenario as wrong than their counterparts with low intergroup contact (M = 7.08, SD = 1.14), F(1, 149) = 9.25, p < .005, η 2p  = .06. Additionally, 4th (M = 6.05, SD = 1.21) and 7th (M = 6.07, SD = 1.08) grade participants were significantly more likely to view non-race based exclusion as wrong than 10th (M = 5.40, SD = 1.27) graders in the lunch scenario, ps > .05. There were no significant differences found for the school dance scenario.

The types of reasoning or justifications participants employed for their wrongfulness ratings were examined using descriptive statistics and revealed that the majority of participants used moral reasons when discussing race-based exclusion in the lunch (M = .93, SD = .26) and dance (M = .84, SD = .36) scenarios. Further analyses of the subtypes of moral reasoning used revealed significant differences for the lunch scenario. The majority of children referred to the wrongfulness of racial prejudice (M = .87, SD = .04) when discussing race-based exclusion. In addition, univariate analyses indicated significant differences for intergroup contact were found. Specifically, children with high levels of intergroup contact (M = .96, SD = .19) were more likely than their low level counterparts (M = .88, SD = .32) to cite reasons of racial prejudice as explanations for the wrongfulness of race-based exclusion in the lunch setting, F(1, 150) = 4.10, p < .05, η 2p  = .03.

Wrongfulness of Non-Race Based Exclusion in School Settings

A 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) × 4 (types of non-race based exclusion at schoolFootnote 1) ANOVA with repeated measures on the last factor was conducted on participants ratings of non-racial exclusion in interracial peer contexts at school (described as “non-race” and “group functioning” in Table 1). Significant main effects were found for this type of non-racial exclusion, F(3, 447) = 4.49, p < .005, η 2p  = .03. Children were significantly more likely to view group functioning based exclusion in the dance scenario (M = 6.24, SD = 1.42) as more wrong than non-race based exclusion in the lunch setting (M = 5.77, SD = 1.24).

Employing descriptive statistics, the types of reasons given by participants to explain their wrongfulness ratings for non-race based exclusion at school indicated that the majority of children invoked moral reasons (M = .55, SD = .45) with a smaller majority of participants employing social-conventional justifications (M = .38, SD = .41). Univariate analyses revealed a significant interaction between intergroup contact and grade in the dance scenario for participants’ use of social conventional reasoning. Fourth grade participants with low levels of intergroup contact were significantly more likely to employ social conventional justifications (M = .50, SD = .49) than their 4th grade peers (M = .19, SD = .38) with high levels of intergroup contact. However, 7th (M = .52, SD = .51) and 10th (M = .45, SD = .49) grade participants with high levels of intergroup contact were significantly more likely to use social conventional reasoning to explain non-racial exclusion due to school rivalry in the dance scenario than 7th (M = .32, SD = .48) and 10th (M = .32, SD = .46) graders with low levels of intergroup contact in the dance scenario, F(2, 149) = 3.87, p < .05, η 2p  = .05.

Wrongfulness of Race-Based Exclusion in the Home Setting

As predicted, the overwhelming majority of children (M = 7.05, SD = 1.21) viewed the decision to exclude a child from a sleepover due to parental concerns about race as wrong. To examine children’s ratings of the wrongfulness of race based exclusion for the sleepover context a 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) ANOVA was conducted on participants’ wrongfulness ratings race-based exclusion in the sleepover scenario. Significant effects were found for grade for race-based exclusion in the sleepover scenario. Fourth grade (M = 6.27, SD = 1.31) participants were significantly more likely to rate race-based exclusion in the sleepover context as more wrong than 10th graders (M = 5.69, SD = 1.37), F(2, 249) = 5.11, p < .01, η 2p  = .06.

Descriptive statistics for the types of reasoning or justifications participants employed for their wrongfulness ratings were examined and analyses revealed that the majority of participants used moral reasons when discussing race-based exclusion in the sleepover scenario (M = .85, SD = .35). Further analysis revealed that the majority of participants cited racial prejudice (M = .71, SD = .45) as the explanation for race-based exclusion in the sleepover context. No significant differences were found for references made to racial prejudice.

Wrongfulness of Non-Race Based Exclusion in the Home Setting

To examine children’s ratings of the wrongfulness of non-race based exclusion in the form of parental unfamiliarity and group functioning based exclusion for the sleepover context a 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) × 2 (type of exclusion: non-race, group functioning) ANOVA with repeated measures on the last factor was conducted on participants’ wrongfulness ratings for non-race and group functioning based exclusion. A significant main and grade effect for type of exclusion was found for participants’ wrongfulness ratings of non-race based and group functioning based exclusion. Participants were significantly more likely to rate group functioning based exclusion (M = 6.08, SD = 1.35) as wrong than non-race based exclusion (M = 4.62, SD = 1.53) in the sleepover scenario, F(1, 148) = 111.74, p < .001, η 2p  = .43. In terms of the effect of grade, 4th graders (M = 5.86, SD = 1.45) were significantly more likely to rate non-race based exclusion from a sleepover party as wrong than their 10th grade (M = 5.07, SD = 1.43) counterparts when comparing the two types of exclusion, F(2, 148) = 6.90, p < .001, η 2p  = .09.

Post-hoc analyses using the Bonferroni correction revealed significant between subject effects for grade in the sleepover scenario. Follow-up analysis found significant grade effects for participants’ wrongfulness ratings of non-race based and group functioning based exclusion. For non-race based exclusion, 4th grade (M = 5.22, SD = 1.61) participants were significantly more likely to rate non-race based exclusion as wrong than their 7th (M = 4.41, SD = 1.41) and 10th grades (M = 4.39, SD = 1.48) counterparts, ps < .05. In addition, for group functioning based exclusion, 4th graders (M = 6.48, SD = 1.31) were significantly more likely to indicate that group functioning based exclusion was wrong than 10th graders (M = 5.76, SD = 1.39), F(1, 149) = 4.07, p < .05, η 2p  = .05.

We also examined the types of reasons participants employed for their wrongfulness ratings for a child’s decision not to invite a friend for a sleepover due to parental unfamiliarity and group functioning concerns. Descriptive statistics indicated that the majority of children employed social-conventional justifications (M = .64, SD = .47) with fewer using moral explanations (M = .28, SD = .44) for their wrongfulness ratings for exclusion due to parents’ lack of familiarity with the friend. In terms of subcategories of reasoning, the use of moral explanations was tied to children’s references to “empathy” for the story character (M = .24, SD = .42) while social conventional reasoning pertained to concerns with parental “unfamiliarity/wariness” (M = .62, SD = .48) about having a minority child in their home. For group functioning concerns that a child would not “fit in” the majority of participants used moral reasons (M = 67, SD = .47) with fewer employing social conventional reasoning (M = .22, SD = .41). Moral explanations were tied to references to “empathy” for the excluded individual (M = .50, SD = .49) while social conventional reasoning referred to “personal choice” (M = .16, SD = .36.) in being able to decide who to invite to a sleepover.

Estimations of the Frequency of Racial and Non-Racial Exclusion

We were also interested in children’s estimations of the frequency of exclusion across the three interracial contexts (see Table 2 for means). A 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) × 3 (context: lunch, sleepover, dance) ANOVA with repeated measures on the last factor was conducted on participants’ estimations of the frequency of race-based and non-race based exclusion. Significant main and grade effects were found for estimates of race-based exclusion across the three scenarios. Specifically, participants were significantly more likely to provide higher estimates of the frequency of race-based exclusion in the sleepover scenario (M = 3.32, SD = .90) and school lunch (M = 3.14, SD = .94) scenario than in the school dance (M = 2.58, SD = .99) scenario, ps < .001. In addition, 4th graders (M = 2.86, SD = .98) were significantly more likely to provide higher estimates of the occurrence of race-based exclusion across scenarios than 10th graders (M = 3.15, SD = .88), F(4, 294) = 3.17, p < .05, η 2p  = .04. No significant differences were found for estimations of the frequency of occurrence for non-race based exclusion across the three scenarios.

Table 2 Estimations of the frequency of non-racial and racial exclusion for three contexts by grade and intergroup contact
Full size table

For each of the three scenarios, a 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) × 2 (type of exclusion: race-based, non-race based) ANOVA with repeated measure on the last factor were conducted on children’s estimations of the frequency of the two types of exclusion. Significant findings were reported for the lunch and sleepover scenarios.

For the lunch scenario a significant main effect for type of exclusion was found, participants estimated that non-race based exclusion (M = 3.14, SD = .94), in the form of lack of shared interests, occurred more frequently than racial exclusion (M = 2.47, SD = 1.14), F(1, 149) = 42.45, p < .001, η 2p  = .22. Post-hoc analyses using the Bonferroni correction revealed significant differences for intergroup contact and grade for estimations of race-based exclusion in the lunch scenario. Participants with high levels of intergroup contact (M = 3.28, SD = .85) were significantly more likely to provide higher estimates of the frequency of race-based exclusion in the lunch scenario than their counterparts with low levels of intergroup contact (M = 2.97, SD = 1.02), F(1, 149) = 6.48, p < .05, η 2p  = .04. In addition, with regard to grade differences, 10th graders (M = 3.36, SD = .88) were significantly more likely to estimate that non-race based exclusion in the lunch scenario occurred more frequently than either 4th (M = 2.98, SD = .95) or 7th graders, (M = 2.95, SD = .96), F(1, 149) = 4.35, p < .05, η 2p  = .06.

For the sleepover scenario there was a significant main effect for type of exclusion. Participants suggested that non-race based exclusion (M = 3.32 SD = .90) due to parental unfamiliarity, occurred more frequently than race-based exclusion (M = 2.60, SD = .95), F(1, 149) = 69.96, p < .001, η 2p  = .32. Post-hoc analysis using the Bonferroni correction for the sleepover scenario found significant differences for estimations of the frequency of non-race based exclusion by grade. Tenth grade participants (M = 3.51, SD = .81) were significantly more likely to provide higher estimates that non-race based exclusion occurred in the sleepover scenario than their 4th grade counterparts (M = 2.98, SD = .92), F(1, 149) = 3.96, p < .01, η 2p  = .06. No significant effects were found for the dance scenario.

Analyses for Stereotype Explanations in Home and School Settings

We were also interested in students’ response to the question, what is it about race that makes people uncomfortable? (i.e., the use of stereotypic explanations). For the sleepover context, 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) ANOVAs were run on each of the three response categories (e.g., stereotype affirmation, stereotype recognition, social contexts of stereotypes). Participants were more likely to refer to the social contexts of stereotypes (M = .47, SD = .49) than stereotype affirmation (M = .18, SD = .33) when considering racial exclusion in the sleepover context, F(2, 290) = 12.23, p < .001, η 2p  = .08.

We next examined participants’ responses in the school setting for each of the categories of stereotype explanations using 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) × 2 (context: lunch, dance) ANOVAs with repeated measures on the last factor. For the stereotype affirmation response (e.g., affirming or endorsing stereotypes) a significant main effect was found for scenario. Interestingly, participants were significantly more likely to endorse or affirm stereotypes as a reason for exclusion in the lunch scenario (M = .26, SD = .43) than in the dance scenario (M = .15, SD = .36), F(1, 117) = 7.32, p < .01, η 2p  = .06. Additionally, significant grade effects were found in participants’ awareness of the social and historical factors contributing to stereotypes across the lunch and dance scenarios. Seventh graders (M = .47, SD = .47) were significantly more likely to recognize the social and historical factors leading to stereotypes than 4th graders (M = .24, SD = .41), across the school settings, F(2, 117) = 4.30, p < .05, η 2p  = .07. Post-hoc analysis revealed significant grade differences for the dance scenario. Seventh graders (M = .48, SD = .50) were significantly more likely to be aware of the social contexts of stereotypes leading to racial exclusion than 4th graders (M = .19 SD = .39), F(2, 117) = 3.33, p < .05, η 2p  = .05. No significant differences were found for the school lunch.

Finally, to examine participants’ responses to the stereotype assessment question for each category of response across the three scenarios we performed a 3 (grade: 4th, 7th, 10th) × 2 (intergroup contact: high, low) × 3 (context: lunch, sleepover, dance) ANOVAs with repeated measures on the last factor. No additional significant differences were found.

Discussion

The present study examined African American children’s and adolescents’ social judgments and reasoning about intergroup exclusion in familiar peer situations. Unlike past U.S. work on the topic where majority-minority comparisons have been a major theme, an important feature of the current investigation was the focus on within-group differences in terms of how African American public school students with differing levels of positive intergroup contact evaluate intergroup exclusion. The findings of the study are consistent with previous work examining young people’s evaluations of intergroup exclusion and also extends the research literature in a number of novel ways. The main findings are discussed below.

In support of our expectations, African American students with low intergroup contact were more likely to suggest that the story characters’ race was the reason for exclusion than their high intergroup contact counterparts. This finding is in accord with research suggesting that students with less opportunity for positive interaction with out-group peers are more likely to attribute racial motives to ambiguous situations of exclusion (Crystal et al. 2008; Killen et al. 2002; McGlothlin and Killen 2006). In addition, consistent with past research (e.g., Crystal et al. 2008; Ruck et al. 2011), intergroup contact also influenced respondents’ ratings of the wrongfulness of exclusion. In the current study, African American students with high levels of intergroup contact were more likely to acknowledge the wrongfulness of race-based exclusion than those students with low levels of intergroup contact. This finding documents important within-group differences in African American youth with regard to the influence of intergroup contact on evaluations of racial exclusion.

Regarding grade differences, we found that younger participants were more likely to view non-race based exclusion as wrong than older participants when considering exclusion in the school setting due to lack of shared interests and in the home setting due to parental discomfort. In accord with previous research (Crystal et al. 2008; Killen 2007; Killen et al. 2007; Killen and Stangor 2001), these findings reflect the fact that with increasing grade, youth are aware that decisions for interracial exclusion may often involve considerations other than race. These age-related decreases in viewing explicit non-race based exclusion as wrong require further research to fully explore the basis for this decrease. On the one hand, there are legitimate bases for social exclusion when all parties recognize the criteria for exclusion and agree on it (e.g., excluding the slow runner from the track team). In other cases, however, ambiguity lies with exclusion decisions and interpreting whether the source is race- or non-race based is difficult (e.g., excluding an African-American runner from the all-White track team by citing his performance may be a cover if it turns out that his/her performance was not different from other European-American team members). Thus, it is possible that ethnic minority status students’ previous experiences with discrimination would lead them to view exclusion in non-race based situations as a proxy for bias or racial prejudice. The sleepover or home context may reflect a unique setting where the actions or motives of a third-party (i.e., parents) must be taken into consideration when assessing bias. We find evidence of this in the sleepover situation where the majority of African American students were more likely to attribute racial motives to the parental figure than to protagonists in the other two scenarios.

While the reasoning findings were in accord with a domain-specific approach by demonstrating the multifaceted nature of children’s social reasoning (Killen and Cooley 2014; Ruck and Tenenbaum 2014), novel findings also emerged in terms of how participants reasoned about intergroup exclusion. Supporting our general prediction, as well as past research and work on social domain theory (Crystal et al. 2008; Ruck et al. 2011), the findings for justifications of exclusion indicated that the majority of African American students generally employed moral reasoning when explaining why intergroup exclusion was wrong in school and home settings. However, previous studies on interracial exclusion with both majority and minority status children have not examined the influence of intergroup contact on reasoning about intergroup exclusion (see Crystal et al. 2008; Ruck et al. 2011). Hence, for the first time, it was found that participants’ reasoning about race-based exclusion was influenced by intergroup contact and grade: In such settings, students with high levels of intergroup contact were more likely than their peers with low intergroup contact to talk about the wrongfulness of race-based exclusion in terms of moral considerations such as racial prejudice. In addition, novel findings, which were qualified by interactions with intergroup contact and grade, also emerged for non-race based exclusion in the school dance scenario. Specifically, in the school dance scenario, 4th grade participants with low levels of intergroup contact were more likely to employ social conventional justifications than their 4th grade peers with high levels of intergroup contact when discussion non-race based exclusion.

In contrast, for the same scenario, older participants (7th and 10th graders) with high intergroup contact were more likely to invoke social conventional reasoning to justify non-racial exclusion than their same grade peers with low intergroup contact. However, an explanation for why differences in intergroup contact would lead to differences in reasoning observed for the youngest participants remains to be determined. A possible explanation with respect to the older participants may be that higher levels of intergroup contact may lead to heightened awareness that racial exclusion in social situations, such as school dances, is due to social conventional concerns such as conforming to peer pressure.

Furthermore, for the sleepover scenario, which involved parental discomfort the majority of African American participants as we indicated above, referred to social conventional justifications to explain the exclusion decision. Negative parental messages about cross-race friendships are often couched in terms of concerns over safety (Edmonds and Killen 2009; Smetana 2006). Thus, those who are targets of such exclusion may become adept at detecting and interpreting such potentially harmful social situations (McKown 2004). Hence it was not surprising that over 60 % of participants in the current study referred to parental unfamiliarity or wariness as a reason White parents did not want to have a Black child in their home. Clearly future research should more closely consider the extent to which non-racial exclusion may be seen by ethnic minority children and adolescents as a proxy for racial exclusion.

Past research has suggested that dominant group participants often use social-conventional reasons such as those involving parental jurisdiction to justify exclusion decisions (Killen and Cooley 2014; Rutland, et al. 2010). However, it should be noted that, in the current investigation, African American students’ social-conventional explanations were not used to condone or justify the legitimacy of the exclusion decision but rather to explain the behavior of majority peers or parents. This was especially the case in the sleepover scenario where the majority status child might be viewed as having little ability to control the actions or fears of his/her parents (Edmonds and Killen 2009). Future research would benefit from examining the degree to which ethnic minority targets of parental-initiated exclusion view their cross-race peers as condoning their parents’ views.

All students reported higher estimates of the frequency of race-based exclusion in the lunch and sleepover situations compared to the school dance situation. Furthermore, another novel finding was the influence of intergroup contact in the school lunch scenario with African American students with high intergroup contact providing higher frequency estimates of non-racial exclusion than those with low levels of intergroup contact. Previous work on this topic has only considered estimations of race-based exclusion with students with high contact providing higher estimates of race-based exclusion than their peers with low contact (e.g., Crystal et al. 2008). The finding pertaining to intergroup contact extends the available literature by suggesting that African American students who have greater opportunities for cross-race friendship may also have more familiarity with peer exclusion being due to non-race based reasons such as lack of shared interests.

For the friendship and sleepover contexts, the data also revealed interesting grade effects with regard to participants’ estimates of non-race based exclusion. The oldest participants were more likely than their younger counterparts to provide higher estimates of the frequency of exclusion for both types of scenarios. This provides further support for previous work indicating that, with increasing age, children view non-race based exclusion as occurring more often and across a range of contexts and situations (Killen et al. 2007).

Interesting results were found regarding African American youths’ awareness of racial stereotypes in home and school contexts. For the question of what it was about race that would make people uncomfortable in a sleepover setting, perhaps not surprisingly, the majority of African American students referred to the historical and social factors contributing to the use of racial stereotypes by majority group members than other explanations. In addition, as predicted, older participants were more likely than the youngest participants to be aware of the historical and social circumstances contributing to the manifestation of racial stereotypes used to justify exclusion. Taken together, these findings (i.e., estimates of frequency of non-race based exclusion and stereotype assessments) suggest that not only are African American participants more likely with increasing grade to experience discrimination across a variety of settings (Fisher et al. 2000; Sellers et al. 2006), but they are also more likely to link that treatment to historical and social circumstances.

An unexpected and novel finding emerged for the school lunch scenario where independent of level of intergroup contact participants used explicit racial stereotypes as reasons for interracial peer exclusion. Such statements were typically in the form of references to general differences between African Americans and Whites that would lead to exclusion. For example, as one 10th grade African American male suggested, “Black people do different things than White people instead of playing sports that make like video games, Michael [the White story-character] probably doesn’t like stuff like that.” Along the same lines, a 7th grade African American female noted, “Maybe Black and White people just think differently,” as a possible reason why the White child would believe that she has little in common with a Black child. Previous work with high status European-American students attending schools with limited opportunities for intergroup contact found that those students used stereotypes to explain racial discomfort more than dominant group students in schools with greater opportunities for such contact (Killen et al. 2010). Our findings revealed that African American students used stereotypes to explain interracial exclusion in peer contexts by European-American students regardless of African American students’ level of intergroup contact. It is important to note, however, that in the current study African American students’ stereotypic views were not derogative or focused on negative traits associated with their European-American peers or members of their own group but rather emphasized more subtle differences that might account for a lack of common interests between the two groups and serve as a legitimate reason for exclusion. An important question for intervention programs, such as educational curricula designed to reduce bias (Hughes et al. 2007; Verkuyten 2008), is whether these participants were reporting their own personal stereotypes about race or reporting more broadly held societal racial stereotypes. It would be useful for future research to examine the origins of African American children and adolescents’ personal stereotypes about race and whether they interact with their awareness of societal racial stereotypes (McKown and Weinstein 2003).

While the current findings make an important contribution to our knowledge concerning African American children’s and adolescents’ judgments and reasoning concerning intergroup exclusion, there are a number of limitations that should be noted. First, the cross-sectional nature of the study limits any inferences regarding causality and developmental change. Future studies would benefit from employing a longitudinal research design. Second, the small sample size with regard to the number of male urban participants necessitated that we omit gender from the analyses to address concerns about statistical power. It would be useful in future work to replicate the current findings with a larger sample, which would also allow an examination of possible gender differences. Third, the study’s exclusive focus on African American youth, while a major strength of the current study given the lack of available research examining within-group differences in this population, restricts the generalizability of the findings to other groups of racial or ethnic minority youth. Fourth, our reliance on participant self-report data was also a limitation that should be addressed in future research. As others have noted (see Way and Greene 2006), employing multi-informant data sources reduces the potential of bias resulting from social desirability and common method variance. Also, observational or ethnographic design would also enrich our understanding of possible differences in how African American youth not only experience but also deal with intergroup relations across various settings and situations. Finally, preliminary analyses indicated that in the current investigation students attending suburban schools had significantly higher levels of intergroup contact than their peers in urban schools. However, rather than comparing suburban and urban students directly, we used opportunities for positive intergroup contact as a proxy for type of school. Unfortunately, the degree to which socioeconomic differences between these two groups of students influenced intergroup contact and evaluations of exclusion was not explored in the current investigation. However, this limitation may be less problematic for the interpretation of the current findings given available research suggesting that socioeconomic advantage does not insulate African American youth from discrimination and prejudice as they are subject to the same types of discriminatory encounters as their low-income counterparts (Comer 1995; Scott and House 2005). Nevertheless, future work would benefit from more closely examining the influence of socioeconomic differences on African American youth’s evaluations of intergroup exclusion.

African American children’s and adolescents’ awareness of the ubiquitousness of racial prejudice and unfair treatment directed towards their group is often acquired through their own experiences with discrimination (Fisher et al. 2000) and/or conveyed through parental racial socialization messages (Hughes et al. 2008; Varner and Mandara 2013). Actual experiences of unfair treatment as well as parental racial socialization messages likely both play an important role in how ethnic minority youth evaluate intergroup exclusion. Future work in this area would benefit by considering the role of these factors in ethnic minority children’s and adolescents’ understanding of interracial exclusion.

Conclusion

The current findings extend the available literature by providing new insights on the benefits of positive intergroup contact beyond improved intergroup attitudes, demonstrating that African American students with high levels of positive intergroup contact had more differentiated judgments and reasoning about intergroup prejudice and discrimination. In addition, African American youths’ evaluations of interracial exclusion also provide support for a domain-specific perspective by demonstrating that their views about intergroup exclusion were influenced by both individual and contextual factors. Furthermore, the views of African American students were not completely uniform or monolithic but in many cases revealed significant within-group variability. The findings of the current investigation are not only important for theoretical reasons but also for interventions to improve intergroup peer relations and promote the well being of ethnic minority children and youth in schools and communities.