Abstract
The present study examined the effects of antiprejudice norms on children’s ethnic attitudes by taking their antiprejudice motivations into account. In a sample of 767 native Dutch preadolescents we found evidence for both an internal and an external motivation to be nonprejudiced which were, respectively, positively versus negatively related to children’s out-group attitudes. Overall, children’s norm perceptions were linked to more positive ethnic attitudes, and this relation was partly explained by their internal antiprejudice motivation. Some normative aspects were found be less effective (moral rule to be nice and honest) than others (equality message) by stimulating an external motivation rather than undermining it and stimulating an internal one. Distinguishing between different normative sources further showed that parents and peers were more influential than teachers. Our findings underline the importance of including motivations in research on norms and out-group attitudes.
One of the key challenges for today’s increasingly diverse societies is to foster acceptance of ethnic and cultural diversity in young people (Civitillo et al., 2017). Therefore, many studies have focused on multicultural education and school-based interventions to improve ethnic attitudes in children (Ülger, Dette-Hagenmeyer, Reichle, & Gaertner, 2018; Verkuyten & Thijs, 2013). Other research emphasizes the crucial role of parents (Degner & Dalege, 2013; Munniksma, Flache, Verkuyten, & Veenstra, 2012) and peers (Jugert, Noack, & Rutland, 2011; Thijs & Verkuyten, 2013). What most of these studies have in common is that they stress the importance of normative influence. One way of promoting positive interethnic relations, for instance, is to transmit norms of tolerance and equality by teaching children the nonacceptability of prejudice and discrimination. However, it is not fully clear how this impacts children’s ethnic attitudes. For instance, multicultural education has a strong normative antiprejudice component (Banks, 2004), but research has shown that it has mixed effects and, on average, is only moderately successful in improving interethnic relations (Aboud et al., 2012; Bigler, 1999; Verkuyten & Thijs, 2013). Likewise, one study found that less than 4% of the variance in the multicultural beliefs of preadolescent children could be explained by the aggregated multicultural beliefs of their classmates (Thijs & Verkuyten, 2013). Thus, it is essential to investigate how, when, and why antiprejudice norms impact children’s interethnic attitudes. The present study addressed these questions by taking children’s motivations to be nonprejudiced into account. While these motivations have been studied among adults for quite some time, very little is known about them in younger populations.
For this research, we examined a large sample of ethnic majority preadolescents (age 7–13 years) and measured their attitudes toward two ethnic minority out-groups. We related children’s ethnic attitudes to their perceptions of the antiprejudice norms of their teachers, parents, and peers, and tested whether these relations were mediated or suppressed by their antiprejudice motivations. More specifically, using structural equation modeling (SEM) we were able to test the unique contributions of different sources (teachers, parents, peers) and aspects (moral, informative) of antiprejudice norms to children’s motivations and attitudes.
Motivations to Be Nonprejudiced
Virtually all research on antiprejudice motivations has examined adults, and much of it is based on Devine’s (1989) theory that people have an automatic tendency to be prejudiced but differ in the degree to which they control or regulate this tendency. Controlling prejudice takes effort, and people can have different reasons for doing this. Although other (Dunton & Fazio, 1997) or more refined (Legault, Green-Demers, Grant, & Chung, 2007) distinctions have been made, an important distinction, and one that is relatively easy to understand for children, is that between the internal and the external motivation to be nonprejudiced (Plant & Devine, 1998). The internal motivation describes an intrinsic desire to be unbiased, which stems from internalized beliefs about the value of equality and the nonacceptability of prejudice. By contrast, the external motivation entails a desire to suppress prejudice to conform to social norms and avoid disapproval by others. Several studies among adults have supported this internal/external distinction, and concluded that both motivations are largely independent of each other and have distinct implications for people’s group attitudes and behavior (Plant, 2004; Plant & Devine, 1998, 2001, 2009).
The internal motivation to be nonprejudiced is assumed to equip people with efficient and consistent strategies to get rid of their biases (Plant & Devine, 2009). Research has supported this notion by showing that it is strongly and negatively related to both the public and the private expression of prejudice (Plant & Devine, 1998; Plant, Devine, & Brazy, 2003), and also associated with less implicit bias (Devine, Plant, Amodio, Harmon-Jones, & Vance, 2002; Hausmann & Ryan, 2004). In addition to this, internal antiprejudice motivation also predicts less interethnic anxiety (Plant, 2004) and a greater openness for interethnic contact (Kunstman, Plant, Zielaskowski, & LaCosse, 2013).
External motivation to be nonprejudiced, on the contrary, has been associated with more, rather than less, prejudiced attitudes. Since more externally motivated individuals aim to come across as nonprejudiced, they respond strategically and only aspire to decrease their bias in those situations where their behavior or thoughts could be perceived as prejudiced by others (Plant & Devine, 2009). Thus, research has shown that external motivation to be nonprejudiced is related to more positive outgroup attitudes in public versus private situations (Plant & Devine, 1998; Plant et al., 2003). In addition to this, external motivation is associated with more, rather than less, explicit prejudice in private situations (Plant & Devine, 1998, 2001) and with implicit bias as well (Hausmann & Ryan, 2004). Externally motivated individuals experience intergroup anxiety and the desire to avoid interethnic interactions (Plant, 2004), and because they feel pressured to live up to the nonprejudiced standards of others, they experience feelings of threat and anger (Plant & Devine, 1998). Such feelings of threat and anger create attitudinal backlash, and as such even increase prejudice (Plant & Devine, 2001).
Antiprejudice Motivations in Children
Hughes, Alo, Krieger, and O’Leary (2016) were the first to show that the internal and external motivations to be nonprejudiced are present from age 8 already, and relevant for children’s interethnic relations. Their argument was that children should be able to have both internal and external motivation after middle childhood, because by then they have acquired, respectively, a basic understanding of the injustice of prejudice and discrimination (McKnown & Strambler, 2009), as well as sufficient perspective-taking abilities and the awareness of antiprejudice norms (Fitzroy & Rutland, 2010). Hughes et al. (2016) developed a new measure based on the scales by Plant and Devine (1998) and found that the two-factor structure of internal and external motivation could be replicated among 8- to 12-year-olds. Moreover, as in adults, children’s motivations were related to their ethnic attitudes and interethnic anxiety. More specifically, children’s internal motivation was positively associated with their ethnic out-group attitudes, and negatively with their interethnic anxiety and ethnic bias. Children’s external motivation was also related to less ethnic bias, but unrelated to their out-group attitudes and associated with more, rather than less, interethnic anxiety. While interpreting these findings, it is important to note that children’s attitude reports were public as they communicated them directly to an interviewer.
In the present study, we examined children’s private reports of their ethnic attitudes and antiprejudice motivations. For this purpose, we developed a new instrument to measure the motivations, as the research by Hughes et al. (2016) was not published at the time. Similar to theirs, our instrument also involved the internal/external distinction, but it exclusively focused on behavior and not on cognitive and affective aspects of prejudice such as stereotypes and negative thoughts. We wanted to make our measure as straightforward and comprehensible as possible, because although children have a basic understanding of prejudice and stereotypes (Brown, 2017), these concepts and their meaning can be rather abstract for them. In fact, the measure by Hughes et al. (2016) used these terms and their younger participants had difficulties with understanding one or more of the items. Thus, our instrument asked children about different reasons for “being nice to children from other countries or cultures,” which is a formulation that is understandable to children (Verkuyten & Thijs, 2013) and applies equally well to their internal and external antiprejudice motivations. We also used a positive formulation (“being nice”) rather than a negative one, because it can be difficult for children to indicate their disagreement with negatively worded items (in this case, not agreeing with reasons for “not being mean”; Marsh, 1986). As we assessed children’s ethnic attitudes in private rather than in public, we expected positive relations between their internal motivation and their out-group attitudes, but negative relations between their external motivation and their out-group attitudes.
The Motivational Impact of Antiprejudice Norms
Several theoretical perspectives such as cognitive developmental theory (Aboud, 1988) and social identity development theory (SIDT; Nesdale, 2004) hold that out-group attitudes become increasingly dependent on social norms after middle childhood. Research has supported this notion for various normative sources. Many studies have examined the role of peer norms—by experimentally manipulating them or measuring norm perceptions in “real life”—and found that children adjust their outgroup attitudes to them (e.g., Fitzroy & Rutland, 2010; McGuire, Rutland, & Nesdale, 2015; Monteiro, de França, & Rodrigues, 2009; Nesdale & Dalton, 2011; Thijs & Verkuyten, 2013). In addition to this, there is evidence that school or teacher norms affect children’s public expressions of ethnic bias (McGuire et al., 2015; Nesdale & Dalton, 2011) and private ethnic attitudes (Kinket & Verkuyten, 1999; Schwarzenthal, Schachner, van de Vijver, & Juang, 2018; Tropp et al., 2016; Verkuyten & Thijs, 2013). And although few studies have directly focused on the antiprejudice norms of parents, research has supported their normative importance by demonstrating medium-sized parent–child similarity in intergroup attitudes (for a meta-analysis, see Degner & Dalege, 2013). Reasoning that social norms exert their influence through the individual’s subjective awareness and understanding of them, we examined children’s perceptions of the antiprejudice norms of their classmates, teachers, and parents, and expected them to be positively related to children’s out-group attitudes. However, to better understand the potential mechanisms underlying this relation, we took the role of children’s antiprejudice motivations into account.
Depending on who communicates them and how, norms against prejudice can provide children with different reasons to be nonprejudiced. On the one hand, they make a moral appeal by stating that prejudice and discrimination are unjust and unfair because people from different groups are equally valuable and worthy of respect (see e.g., Rosenthal & Levy, 2010). These norms are typically expressed and taught to children under the assumption that they will adopt and internalize them. For example, it is the hope and aim of multicultural education that students learn to respect ethnic and cultural others by embracing egalitarian and antiracist beliefs as personal values and self-endorsed standards (Bigler, 1999). The internal motivation to be nonprejudiced reflects such norm internalization (e.g., “I am nice to children from other countries and cultures because I think everyone is equal”). On the other hand, antiprejudice norms could inadvertently stimulate an external motivation to be nonprejudiced. Children who are repeatedly exposed to the message that prejudice and discrimination are socially unacceptable could become excessively concerned with the reactions of others and the possibility of rejection when expressing prejudice. Hence, the external motivation to be nonprejudiced may also increase as a response to such norms (e.g., “I am nice to children from other countries and cultures because otherwise people might think I am mean”).
Theoretically, both possibilities correspond to the classical social psychological distinction between informational and normative influence (see Turner, 1991). Informational influence describes a process whereby individuals internalize the messages of others because they are convinced by them. Thus, they “accept information obtained from another as evidence about reality” (Deutsch & Gerard, 1955, p. 629). Normative influence, however, implies that individuals will conform to the expectations of others because of perceived normative pressure. And this means that one could “say things which one disbeliefs but which agree with the beliefs of others” (Deutsch & Gerard, 1955, p. 629). The work on antiprejudice motivations suggests that the exposure to antiprejudice norms can fuel both these informational and normative effects.
In two seminal experiments, Legault, Gutsell, and Inzlicht (2011) examined different motivations to regulate prejudice in adults after exposing them to antiprejudice norms. Rather than measuring the internal and the external motivation, they used a scale based on the taxonomy of self-determination theory (Legault et al., 2007; Ryan & Deci, 2000), which distinguishes between intrinsic motivation, internalized versus noninternalized forms of extrinsic motivation, and amotivation. Still, their results indicate that people become more internally motivated to be nonprejudiced when presented with arguments for the importance of prejudice reduction and when reminded of freedom of choice, yet more externally motivated when pressured to comply with social norms against prejudice (Legault et al., 2011). Likewise, the results of a recent survey study—which also used the self-determination taxonomy—indicate that adolescents’ perceptions of multicultural norms in daily life are positively associated with both their internal and their external motivation to be nonprejudiced (Thijs, Gharaei, & de Vroome, 2016). In line with these findings, we anticipated similar relations for the present study. Moreover, given our hypotheses about the effects of children’s antiprejudice motivations on their ethnic out-group attitudes, we expected that their internal motivation would mediate the (anticipated) norm–attitude links (positive indirect effect), but that their external motivation would suppress them (negative indirect effect). However, in line with the expectation of an overall positive relation between perceived antiprejudice norms and children’s ethnic attitudes, the positive pathway via internal motivation was anticipated to be stronger than the negative one via external motivation.
Normative Sources and Aspects
To examine the impact of children’s norm perceptions, we developed a measure based on previous research on multicultural education (Verkuyten & Thijs, 2013). This measure involves children’s teacher, classmates, and parents as different normative sources in order to explore the relative contributions of each source to children’s antiprejudice motivations and ethnic out-group attitudes. To our knowledge, the three of them have not been simultaneously examined in research on children’s intergroup relations, although there is some evidence for independent effects of parental and peer norms (Mähönen, Jasinskaja-Lahti, & Liebkind, 2011; Thijs et al., 2016), and peer and school norms (Tropp et al., 2016).
Apart from different sources, our measure involves three different but interrelated aspects of antiprejudice norms (see Verkuyten & Thijs, 2013). Taken together, they reflect the notion that group prejudice is not only a negative orientation toward out-group others, but also an unjustified one (Augoustinos & Reynolds, 2001). The first two aspects involve the moral rules to respect ethnic others (“you should be nice and honest to people from other cultures”; positive moral rule) and to not discriminate against them (“it is wrong to be mean to people from other countries”; negative moral rule). These rules directly address the negativity and lack of positivity characteristic of out-group prejudice, by telling individuals how (not) to behave. The third aspect targets the unjustified nature of prejudice. It involves the message that people from all cultural groups are equal. It is an important component of multicultural education (e.g., Bigler, 1999) and explains why prejudice is wrong.
Because we assessed these three different aspects with parallel items for the teacher, classmates, and parents, we could aggregate their scores and explore their differential impact on children’s motivations and out-group attitudes. Unlike the moral rules against prejudice, the message of equality does not explicitly prescribe (“you should. . .”) or proscribe (“it is wrong to. . .”) a particular behavioral orientation toward the out-group. Thus, compared to those rules, it should be less likely to heighten children’s concern with others’ reactions to their attitudes and behaviors. Rather than drawing their attention to the social unacceptability of prejudice, the equality message provides children with a valid reason to reject prejudice. Theoretically, its impact should therefore be informational rather than normative (Deutsch & Gerard, 1955). Thus, it could be anticipated that, after disentangling these different aspects, the equality message would be more likely than the moral rules to fuel the positive pathway expected for children’s internal antiprejudice motivation, but less likely to fuel the negative pathway expected for their external antiprejudice motivation.
Overview of the Present Study
The goal of the present study was to further understand the link between children’s perceptions of antiprejudice norms and ethnic attitudes by taking their antiprejudice motivations into account. We conducted our research in the Netherlands, and examined native Dutch ethnic majority preadolescents’ attitudes towards Moroccan and Turkish peers. Moroccan and Turkish people are the largest non-Western minority groups in the Netherlands and originally came to the country as low-skilled labor migrants. They still hold a relatively low-status position and are among the least liked minority groups in Dutch society, also among children (Verkuyten & Kinket, 2000). Moreover, the attitudes towards these two groups are strongly associated with each other among native Dutch children (Thijs & Verkuyten, 2012).
We created a new measure for children’s antiprejudice motivation and evaluated seven different hypotheses. Our first two expectations were that children’s internal motivation would be related to a more positive out-group attitude (Hypothesis 1) and that children’s external motivation would be related to a less positive out-group attitude (Hypothesis 2). Next, we expected an overall positive link between children’s perceptions of antiprejudice norms and their out-group attitudes (Hypothesis 3), which would be mediated by their internal antiprejudice motivation (positive indirect effect; Hypothesis 4) but suppressed by their external antiprejudice motivation (negative indirect effect; Hypothesis 5). Lastly, we tested the hypotheses that the positive indirect effect via internal motivation would be most pronounced for the equality message aspect of the antiprejudice norms (Hypothesis 6) and that the negative indirect effect via external motivation would be least pronounced for this aspect (Hypothesis 7). In addition to this, we explored the relative contributions of children’s teacher, classmates, and parents as normative sources.
In our analyses, we also took into account children’s gender and grade level, which strongly correlates with their age. Research has shown that girls tend to report more out-group positivity than boys (e.g., Thijs & Verkuyten, 2012), thus it is important to control for this effect of gender to prevent spurious findings. Hughes et al. (2016) found that older children had stronger internal and external motivations to be nonprejudiced, but to our knowledge, they did not examine whether their motivation measures were invariant across age. We tested this in our study. 1
Method
Data and Participants
Participants were 767 children (51.8% girls) with a mean age of 9.93 years (SD = 1.18; range 7–13 years). These children were from 48 classrooms (Grades 3–6) in different schools in various parts of the Netherlands. Most of these classrooms included children from one grade, but 10 and eight of them combined, respectively, two and three adjacent grades. The classrooms differed in ethnic composition: of the participating students per classroom 72.2% were Dutch (SD = 19.8) and 5.4% of Turkish or Moroccan descent (SD = 12.8). It is important to note, however, that our analyses focused on within-classroom differences only. After receiving informed parental consent, students anonymously and voluntarily filled out a pen-and-paper questionnaire in class. A researcher or research assistant was present to help them with the questions if necessary. For the present study, we first selected those children who identified themselves and both of their parents as native Dutch (N = 838), and then used list-wise deletion to remove children with missing values on one or more of the study variables. Missing value analysis suggested that data were missing completely at random, χ2(26) = 20.70, p = .76.
Measures
Antiprejudice motivations
Our measure to assess children’s antiprejudice motivations consisted of 10 items that are shown in Table 1. These items were newly developed and based on work on antiprejudice motivations among adults (Legault, Green-Demers, & Pelletier, 2006; Plant & Devine, 1998). Children were asked to rate their endorsement of different reasons for “acting nice towards children from other countries or cultures” on a 5-point Likert-type scale (1 = no! 5 = yes!). Five of these reasons captured their internal motivation and involved their personal endorsement of equality and positivity toward ethnic or cultural others, as well as their intrinsic appreciation of intergroup contact. The other five reasons were external in nature and involved children’s concerns with social unacceptability and the reactions of others. A study that used a preliminary six-item version of this measure found evidence for a two-factor structure corresponding to an internal and an external motivation. Children’s internal motivation was positively related to their out-group attitudes, but their external motivation was unrelated to these attitudes. However, both measures had moderate reliabilities which might have undermined the power to find a unique negative effect of external motivation (Geerlings, Thijs, & Verkuyten, 2017).
Items and factor loadings for children’s antiprejudice motivations.
To test whether the current 10 items corresponded to two different factors (internal vs. external), we specified a correlated two-factor model without cross-loadings in Mplus Version 8 (Muthén & Muthén, 1998–2017). We relied on four fit indexes: the comparative fit index (CFI), the Tucker Lewis index (TLI), the root mean square error of approximation (RMSEA), and the standardized root mean residual (SRMR). Model fit is considered good if CFI and TLI have values of 0.95 or higher, and RMSEA and SRMR are lower than 0.05. CFI and TLI values larger than 0.9 and RMSEA and SRMR values smaller than 0.1 are considered acceptable (Kline, 2011). To take the nested structure of our data into account, we used the clustering option in Mplus. Results showed that the fit of the two-factor model was barely acceptable, χ2(34) = 193.22, CFI = 0.90, TLI = 0.87, RMSEA = 0.08, SRMR = 0.08. Further inspection, however, showed that one item (“. . .because I want other people to like me”) had a comparatively low standardized loading (0.33), and after removing this item, model fit became satisfactory, χ2(26) = 86.63, CFI = 0.96, TLI = 0.94, RMSEA = 0.06, SRMR = 0.04. Table 1 shows the standardized factor loadings for this final model. The correlation between both latent factors was relatively weak, r = .11. Based on the factor analyses, we calculated two measures: a five-item Internal Motivation Scale for which Cronbach’s alpha was .77, and a four-item External Motivation Scale with a Cronbach’s alpha of .81.
Given the novelty of our instrument, we also conducted a set of multiple group analyses to examine the metric and scalar invariance of each factor separately for children in the lower grades (3 and 4; n = 327; Mage = 8.84 years, SD = 0.67) versus the higher grades (5 and 6; n = 440; Mage = 10.75 years, SD = 0.73). Comparisons of more versus less constrained models revealed that there was metric invariance for both measures; for internal motivation: χSB2dif (4) = 0.89, p = .93; for external motivation: χSB2dif (3) = 5.11, p = .16; but no scalar invariance: χSB2dif (4) = 42.43, p < .01 and χSB2dif (3) = 16.10, p < .01, respectively. Thus, it was appropriate to compare the correlates of these measures for children in the higher versus the lower grades, but not to directly examine age differences in children’s antiprejudice motivations. Hence, we did not control for grade level in our main analyses, although we explored its moderating effects.
Perceived antiprejudice norms
To assess their perceptions of antiprejudice norms, children were presented with a neutral doodle character who expressed the following three statements in a speech balloon: “You should be nice and honest to people from other cultures,” “It is wrong to be mean to people from other countries,” and “People from all cultural groups are equal.” These statements were based on research on multicultural education in the Netherlands (Verkuyten & Thijs, 2013). After each statement, children were asked to indicate whether, respectively, their teacher, classmates, and parents ever made it. The response scale ranged from 1 (absolutely never!) to 5 (very very often!).
To analyze the factor structure behind these nine items, we tested two correlated uniqueness models in Mplus (Kline, 2011), again using the clustering option. In the first model we specified three factors for normative source (teacher, classmates, parents) and correlated the error correlations for each statement, and in the second model we specified three factors for each statement (i.e., aspect) and correlated the error correlations for each normative source. Both models fitted the data well, χ2(15) = 17.35, CFI = 1.00, TLI = 1.00, RMSEA = 0.01, SRMR = 0.02; and χ2(15) = 18.37, CFI = 1.00, TLI = 1.00, RMSEA = 0.02, SRMR = 0.02, respectively. In the first model, the correlations between the source factors were very strong, r = .78 for teacher and classmates, r = .74 for teacher and parents, and r = .70 for classmates and parents. In the second model, the correlation between the two normative rule aspects (“you should be nice” and “wrong to be mean”) was .78, and their correlations with the equality message aspects were both .70. This indicates that there was considerable overlap between the perceived norms of different sources, and between the different components of these sources.
For our analyses, we calculated several measures: a measure for the total perceived antiprejudice norm consisting of all nine items (Cronbach’s α = .91), three scales for the perceived norms of children’s teacher (Cronbach’s α = .81), classmates (Cronbach’s α = .82), and parents (Cronbach’s α = .82), and three scales for the normative aspects (the positive moral rule “you should be nice,” Cronbach’s α = .77; the negative moral rule “it is wrong to be mean,” Cronbach’s α = .83; and the equality message, Cronbach’s α = .85).
Ethnic attitudes
To assess children’s ethnic attitudes, they were asked to evaluate Moroccan and Turkish children on three stereotypic traits. These measures have been successfully used in previous research in the Netherlands (Thijs, 2017) and have been shown to strongly correlate with smiley-face measures of ethnic attitudes, which are commonly used among children (Thijs & Verkuyten, 2016). The participants had to estimate whether most of the children in each group were “honest,” “fun to play with,” and “eager to help,” and the response scale ranged from 1 (no, certainly not!) to 5 (yes, certainly!). Confirmatory factor analysis in Mplus with the clustering method supported a two-factor structure in which the items for each group evaluation loaded on a separate factor. After allowing two cross-factor error correlations (between the items “honest” and “fun to play with”), the model fit was acceptable, χ2(6) = 25.60, CFI = 0.98, TLI = 0.95, RMSEA = 0.07, SRMR = 0.03. As the correlation between the factors for the attitudes toward the Moroccan and Turkish children was very high, r = .85, we combined the six items into one scale for children’s out-group attitude. Cronbach’s alpha was .90 for this scale.
Results
Our data had a nested structure, as we sampled whole classrooms. To account for this nesting, we tested our hypotheses using the clustering option in Mplus 8 (Muthén & Muthén, 1998–2017). To avoid estimation problems due to having more parameters than clusters, all analyses were performed on observed scales rather than latent variables.
Preliminary Findings
Simple correlations between all study variables are shown in Table 2. As anticipated, children who were more internally motivated to be nonprejudiced had more positive ethnic out-group attitudes, whereas more externally motivated children had less positive out-group attitudes. All measures for the perceived antiprejudice norms, that is, for the total norm, the norms by each source (teacher, classmates, and parents), and the three normative aspects were associated with a stronger internal antiprejudice motivation and more positive out-group attitudes. There were no bivariate correlations between children’s norm perceptions and their external antiprejudice motivation. Table 2 also shows the importance of taking gender into account. Girls reported a stronger internal motivation, a weaker external motivation, and more positive out-group attitudes.
Intercorrelations of study variables.
Note. *p < .05. **p < .01.
Path Models
To further test our hypotheses, we specified a set of path models in Mplus, taking into account the classroom clustering of our data. First, however, we examined the impact of our decision to focus on the within-classroom level by calculating the intraclass correlation (ICC) for the scale measures (see Table 2). These ICCs are shown in the right column of Table 2 and indicate the proportion of the total variance that exists between classrooms. For all of the scales, most variance was found within rather than between classes. However, classroom variance was larger for the norm perception measures (8.3% to 19.1%) than for the other ones (⩽ 6.4%), and largest for the perceived teacher norms, which makes sense as children in the same class had the same teacher.
Total norm
For the first model, we used the total norm measure. We regressed children’s out-group attitudes on their internal and external antiprejudice motivations, and we regressed each of these three measures on the perceived norm and the gender control variable. We also specified a correlation between the residuals of both motivation measures. The standardized results are shown in Figure 1. The total norm had a positive effect on children’s internal antiprejudice motivation, but no effect on their external antiprejudice motivation. As expected, children’s out-group attitudes were positively linked to their internal motivation (H1) and negatively to their external motivation (H2), and the total effect of the total norm on their attitudes appeared to be positive and significant, b = 0.26, SE = 0.04, p < .01 (H3). In addition to this, there was a positive indirect effect of the norm on children’s out-group attitude via their internal motivation, b = 0.13, SE = 0.02, p < .01. This pathway was consistent with H4. As shown in Figure 1, there was still a direct effect between the perceived norm and out-group attitudes, indicating only partial mediation. Moreover, there was no indirect effect of the norm via children’s external motivation, b = −0.00, SE = 0.01, p = .80, which means that our expectation about suppression (H5) was not supported.
Standardized results for the first SEM model including the perceived total norm.
Figure 1 also shows that girls reported a stronger internal but a weaker external motivation. Moreover, there were indirect effects of gender, implying that girls reported a more positive out-group attitude due to a higher internal and a weaker external antiprejudice motivation, b = 0.08, SE = 0.02, p < .01 and b = 0.01, SE = 0.01, p = .057, respectively.
Normative sources
Next, we specified the same path model with the perceived norms of the teacher, classmates, and parents instead of the global norm measure. The result (standardized effects) is shown in Figure 2. Please note that the effects of gender were not included, which were similar to those in Figure 1 (with a higher internal motivation and a lower external motivation for girls, p < .01 and p < .05, respectively. Whereas the perceived teacher norm had no effect, the perceived norms of children’s classmates and parents had positive effects on children’s internal antiprejudice motivation, but not on their external one. Unlike the perceived parental norm, the perceived peer norm also had a direct effect on children’s ethnic attitude, and the total effect was largest for this predictor. Further inspection showed that the perceived norms by classmates and parents had positive indirect effects on children’s out-group attitude via their internal motivation, b = 0.06, SE = 0.02, p < .01 and b = 0.11, SE = 0.02, p < .01, respectively.
Standardized results for the second SEM model including different normative sources.
Normative aspects
Finally, we tested a model that regressed children’s antiprejudice motivations and out-group attitude on the scales for the three normative aspects (“be nice,” “don’t be mean,” and the equality message), as well as their out-group attitude on their motivations. We controlled for gender, which again had similar effects as those presented in Figure 1. As shown in Figure 3, the positive moral rule to be nice and honest to people from other cultures had no unique effect on children’s internal antiprejudice motivation, but a positive effect on their external motivation. Further inspection showed that this rule had a marginally significant indirect negative effect on children’s ethnic attitude via their increased external motivation, b = −0.02, SE = 0.01, p = .067, which is consistent with H5. Still, as shown in Figure 3, the total indirect effect of this normative aspect was not significant.
Standardized results for the third SEM model including different normative aspects.
The negative moral rule against being mean to people from other cultures had a positive effect on children’s internal motivation, and as such a marginally positive indirect effect on their out-group attitude, b = 0.04, SE = 0.02, p = .055, but no effect on their external motivation. Again, however, the total indirect effect was not significant. Finally, the equality message had a positive effect on the internal motivation to be nonprejudiced but a negative effect on the external motivation. As there was a negative relation between children’s external motivation and their out-group attitude, this led to a more positive out-group attitude in both cases: b = 0.07, SE = 0.02, p < .01, via internal motivation; and b = 0.03, SE = 0.01, p < .05, via external motivation. As shown in Figure 3, there was also a direct positive effect of the equality message.
Taken together, these results support the expectation that the positive indirect effect via children’s internal motivation would be most pronounced for the equality message aspect of the antiprejudice norm (H6). Although we did not anticipate a positive indirect effect via children’s external motivation, our results are also consistent with the hypothesis that the negative indirect effect would be least pronounced for this aspect (H7).
Additional Analyses
To examine the robustness of our findings, we conducted three sets of additional analyses. First, we explored whether the effects of the perceived norms on children’s antiprejudice motivations, and the effects of these motivations, were similar for older versus younger children. We conducted a set of multigroup analyses in which we tested the models in Figures 1–3 and compared them for children in the higher (5 and 6; n = 440, Mage = 10.75 years, SD = 0.73) versus the lower grades (3 and 4; n = 327, Mage = 8.84 years, SD = 0.67). Results showed that there was no significant deterioration of model fit when all paths and the correlation between both motivations were constrained to be equal, χSB2dif (9) = 12.96, p = .16 for Model 1; χSB2dif (15) = 16.24, p = .37 for Model 2; and χSB2dif (15) = 17.73, p = .28 for Model 3. Thus, our results appeared to be similar for younger and older respondents.
Next, we retested the model in Figure 2 without the perceived classmate norms as a predictor. Classmates are exposed to the same teacher and, as shown in Table 1, the perceived norms of both parties were strongly related. This could mean that the perceived norms of the teacher had effects that were nonsignificant due to the inclusion of the perceived norms of the classmates. Results did not support this possibility, however. When classmate norms were not included in the model, the perceived teacher norms were still unrelated to children’s antiprejudice motivations, although they had a small and positive direct effect on children’s out-group attitude, b = 0.10, SE = 0.04, p < .05.
Finally, we compared our path models to alternative versions in models in which we changed the order of the endogenous variables (antiprejudice motivation and out-group attitudes). The reason for this is that a positive ethnic attitude might strengthen the importance of internal reasons to be nice to ethnic others (liking it, finding it important) and diminish the importance of external reasons (expectations of others). Because these alternative models were not nested in the original ones, we relied on the Akaike index to compare them (see Kline, 2011). To make meaningful comparisons, we “trimmed” the original and alternative models by allowing no direct paths between the predictor variables (norms and gender) and the dependent variable (out-group attitude in the original models, and antiprejudice motivations in the alternative models). 2 For each original model, the Akaike index was lower than for its alternative counterpart: 5592.66 versus 5626.05 for Model 1; 5589.56 versus 5627.41 for Model 2; and 5584.35 versus 5621.83 for Model 3. This suggests that the models with the antiprejudice motivations preceding the out-group attitude could be preferred over models with the out-group attitude preceding the antiprejudice motivations.
Discussion
The goal of this study was to shed light on the effects of antiprejudice norms on children’s ethnic attitudes by taking into account the role of their antiprejudice motivations. Theoretically, the influence of such norms could be informational, meaning that children would adopt and internalize them, or normative, in which case children would comply with them for social reasons (Deutsch & Gerard, 1955). In this study, we investigated both types of influence by studying children’s internal and external motivations to be nonprejudiced. Moreover, we examined if different kinds of normative sources (peers, teacher, and parents) and aspects (moral rules and equality information) would have differential effects.
Like the first and only research in this area so far (Hughes et al., 2016), our study tested a new measure to assess antiprejudice motivations in ethnic majority children and found clear evidence for an internal and an external motivation. However, unlike Hughes et al. (2016), we related our measure to children’s privately reported ethnic out-group attitudes. Results are largely in line with our expectations and comparable to what has been found among adults (Hausmann & Ryan, 2004; Legault et al., 2011; Plant & Devine, 2001; Wyer, 2007). Children’s internal motivation involved their personal beliefs in the importance of ethnic equality and the correct treatment of ethnic and cultural others, but also their personal interest in those others. This motivation had a considerable positive effect on children’s evaluations of ethnic minority peers. By contrast, children’s external motivation included their concern with external pressure to be nonprejudiced and possible negative reactions from others, and it showed a small but negative relation to their out-group attitudes. This indicates that promoting the “right reasons” for being nonprejudiced is crucial, not only for adults but for younger populations as well.
Our findings further demonstrate the importance of children’s antiprejudice motivations by demonstrating their role in the link between children’s norm perceptions and their ethnic attitudes. This role was most straightforward for children’s internal motivation. As expected, the perception of norms that reject prejudice and promote ethnic out-group positivity predicted a more positive orientation toward ethnic minority peers, and this association could be largely explained by children’s internal antiprejudice motivation. This indirect positive pathway implies that children internalize the antiprejudice norms they perceive in their daily lives, and demonstrates that these norms can have informational influence (Deutsch & Gerard, 1955). However, unlike we expected, children’s total norm perceptions were unrelated to their external motivation, suggesting that, overall, there was no normative influence (Deutsch & Gerard, 1955). Moreover, this indicates that the perception of antiprejudice norms does not create a so-called “attitudinal backlash” via an increased external motivation (see Legault et al., 2011). Together, these findings show that it can be useful to emphasize norms against prejudice as a means to improve intergroup relations in children, and as we will discuss in what follows, parents and especially peers appear to be most important in this respect.
Although the different normative aspects in our sample were strongly related, we could disentangle them and thus analyze their unique effects. 3 The results indicate that it may be more effective to emphasize some normative aspects over others. We found that the message that people from different cultures are equal had an unambiguously positive impact by being associated with both a stronger internal antiprejudice motivation and a weaker external one, and partly as a result of this, with a more positive out-group attitude. By contrast, the overall indirect effects of the two moral rules were nonsignificant, indicating that such rules are ineffective when decoupled from the message that people from different cultural backgrounds are equal. Still, whereas the moral rule against being mean to ethnic and cultural others was related to more out-group positivity via an increased internal motivation, the moral rule to be nice and honest to those others predicted a less positive out-group attitude via a stronger external motivation. Everything else being equal, children who perceived more of this positive moral rule were more concerned about social sanctions if not living up to it and, in turn, less positive about ethnic out-group children.
We can only speculate why this negative pathway was found for the moral rule to be positive to ethnic and cultural others rather than for the rule not to be negative to them. Earlier research has shown that positive moral messages are more effective than negative moral messages in promoting positive behavior in (younger) children (Lee et al., 2014). Our results clearly differ from those findings, but it is important to note that, unlike the moral rules that we studied, the moral messages in that research explicitly addressed the consequences of desired or undesired behavior. Instead, our findings seem to be consistent with the positive–negative asymmetry effect in intergroup research (Mummendey & Otten, 1998), as well with the distinction between proscriptive and prescriptive morality (Janoff-Bulman, Sheikh, & Hepp, 2009). The positive–negative asymmetry effect involves the tendency to regard the presence of negative out-group treatment as more problematic than the absence of positive out-group treatment (Mummendey & Otten, 1998), and our finding that children appeared to internalize the negative rule (“don’t be mean”) rather than the positive one (“be nice”) is clearly consistent with this conclusion. Related to this, the negative rule can be regarded as an instance of proscriptive morality, whereas the positive rule is exemplary of prescriptive morality. Proscriptive morality focuses on the avoidance of bad behavior (“you should not”), and violating proscriptive moral rules is blameworthy. By contrast, prescriptive morality focuses on the desirability of good deeds (“you should”), and because people cannot be expected to do good all the time, prescriptive moral omissions are much more acceptable than proscriptive ones (Janoff-Bulman et al., 2009). Accordingly, it would be easier to discard the rule to be nice than the rule not to be mean, and that is what our findings suggest. As mentioned, this is mere speculation, and it is important not to overemphasize the different effects of the positive and negative moral rules, as the overall indirect effect of each rule was nonsignificant. 4 Still, future research could test these interpretations with more extensive measures of different normative aspects.
With respect to different normative sources, parents and especially classmates appeared to be important for the stimulation of an internal antiprejudice motivation and positive out-group attitudes, while there was no evidence for an influential teacher role. These findings underline the normative power of the peer group for preadolescent children (Berndt, 1979), and indicate that it is highly relevant for research on multicultural education and teachers’ diversity norms to control for the impact of peers and parents as normative agents—which is typically not done in that kind of research (Verkuyten & Thijs, 2013). Our additional analyses showed that the lack of effects of perceived teacher norms on children’s antiprejudice motivations could not be explained by their overlap with the perceived norms of the classmates (who had the same teacher as the respondent children). Still, there was a direct effect of teacher norms on children’s out-group attitudes when classmate norms were not included in the model. This indicates that these teacher norms could be important for children via their impact on the norms of the classroom peer group, but not by stimulating or discouraging specific antiprejudice motivations.
Although not its main focus, our study showed interesting results for age and gender. We found that antiprejudice motivations were metrically invariant across age groups, which means that we could compare the effects for them. Importantly, all effects were similar for both age groups, which indicated that they processed the antiprejudice norms in similar ways. Still, there was no evidence for the scalar invariance of our motivation measures, which suggests that researchers should be careful in their conclusions about the direct relations between age and antiprejudice motivations (see Hughes et al., 2016). 5 We also found indirect effects of gender. Girls were more positive about their ethnic out-groups via a stronger internal motivation to be nonprejudiced and a weaker external one. Thus, the relation between gender and out-group attitudes that is sometimes found in the literature (e.g., Thijs & Verkuyten, 2012) might be accounted for by girls’ stronger tendency to be nonprejudiced for the “right reasons.” We do not have a clear-cut explanation for this difference, but it may be related to gender differences in socialization. Girls are sometimes socialized to be more prosocial and other-oriented than boys (see Brody, 1999). Although the presently found gender differences in antiprejudice motivation existed independent of the perceived norms (and although gender was unrelated to those norms), these general socialization differences could make girls comparatively more open to cultural others and less focused on how others might react to their out-group behaviors. Future research could test this possibility by including parents’ socialization goals.
In evaluating the current study, some limitations and directions for future research should be considered. First, our data were cross-sectional, which means that we cannot make claims about causality. Like other researchers (e.g., Hughes et al., 2016; Legault et al., 2007), we examined antiprejudice motivations as predictors rather than outcomes of out-group attitudes, and our additional analyses favored this approach. Yet, it is likely that the relationships between these variables are bidirectional, as children might also become more internally motivated to be nonprejudiced and less focused on the social unacceptability of prejudice if they like out-group others. Therefore, future studies should use longitudinal designs to confirm our interpretations and examine the relations of children’s antiprejudice motivations and ethnic attitudes over time. Relatedly, there is the possibility that children’s norm perceptions partly depended on (rather than predicted) their antiprejudice motivations and ethnic attitudes via a process of social projection (Degner & Dalege, 2013). That is to say, they may have not been fully sure about the norms of their peers and parents, and wrongfully assumed that these were compatible with their own orientations toward out-group others. A recent study indeed found that children’s minority-group attitudes predicted their perceptions of a corresponding peer-group norm over time, but it is important to note that the relationship was bidirectional (Thijs & Zee, 2019). Moreover, whereas the norm perception measures in that study focused on the opinions of others, our measures focused on their communications. Children are probably more certain about what others say than about what others think, and therefore the need for social projection to diminish uncertainty may have been relatively low in our study. Still, longitudinal research is needed to support such a claim.
Second, as we relied on children’s self-reports of norms, motivations, and attitudes, we cannot rule out the possibility of common method effects (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Our decision to use subjective measures was appropriate given our research questions. However, future studies could examine children’s antiprejudice motivations in relation to norms reported by normative agents themselves or norms that are experimentally manipulated, which would also rule out the possibility of social projection.
Third, future research could use multilevel modeling to examine differences in antiprejudice motivations and ethnic attitudes at the classroom level, and study how students’ aggregated perceptions of the norms of their teacher affect those differences. As mentioned, much of the variance in perceived teacher norms was at the level of the classroom, and it is possible that students’ aggregated, consensual perceptions are important even if their individual perceptions are not. 6 Related to this, future research should examine other ways in which teachers could influence children’s antiprejudice motivations. Recent research has shown that the quality of students’ relationship with their teachers has a positive impact on their out-group attitudes (Miklikowska, Thijs, & Hjerm, 2019) and also that this is independent of perceived teachers’ antiprejudice norms and children’s perceived relationship with their parents (Geerlings et al., 2017). Thus, teachers may make children more open to ethnic and cultural others by providing them with relational security rather than with antiracist messages, regardless of the perceived norms of peers and parents.
Fourth, our motivation measures focused on the behavioral aspect of prejudice but not on its affective (dislike) and cognitive aspects (stereotypes). It might be difficult to measure children’s motivations for these other aspects, as they may have a hard time thinking about motivations for particular feelings or beliefs. Still, future research could try this and examine the relations between children’s cognitive and affective antiprejudice motivations, and their norm perceptions and ethnic attitudes. We have no strong reason to believe that those relations would be substantially different from the ones obtained in the present research.
Fifth, our study included the perceived norms of children’s teacher, but future research could also consider the perceived norms of the school at large (cf. Nesdale & Dalton, 2011). Teachers working in the same school can strongly differ from each other in their teachings about diversity (Agirdag, Merry, & van Houtte, 2016), which suggests that the norms stressed by the school as a whole do not necessarily align with those expressed by individual teachers. Thus, it is still possible that children’s perceptions of the norms of their school, unlike those of their teacher, uniquely impact their antiprejudice motivations and ethnic attitudes.
Finally, the variance explained in children’s antiprejudice motivations was relatively low in this study. Specifically, the finding that children’s external motivation was unrelated to the total norm might be regarded as surprising, as this motivation involves the very desire to conform to social norms. Still, this absent relation could conceal different and opposing individual responses that depended on the communication of those norms. Presumably, antiprejudice norms only increase external motivation when there is social pressure to conform. Indeed, Legault et al. (2011) found that people’s antiprejudice motivation became less self-determined (i.e., more external) when “urged to combat prejudice and to comply with the social norms of non-prejudice” (p. 1473), but more self-determined (i.e., more internal) when presented with arguments for the importance of prejudice reduction and reminded of freedom of choice. Although the results for the different normative aspects are somewhat consistent with this interpretation, future research should directly examine how antiprejudice norms are delivered to children. Related to this, it is important to study factors that interact with children’s norm perceptions. For example, based on self-determination theory (Ryan & Deci, 2000), it can be expected that children are more likely to internalize the norms of others if they experience a sense of relatedness to them.
Despite its limitations, the current study contributes to the social psychological and developmental literature on intergroup relations by demonstrating the intermediate role of antiprejudice motivations in the link between norms and ethnic attitudes. Our results indicate that perceived norms against prejudice have a positive impact on children’s out-group attitude partly because they promote an internal motivation to be nonprejudiced. However, some normative aspects appear to be less effective than others, and if not accompanied with a good argumentation, the prescriptive rule to be nice to ethnic and cultural others might backfire by stimulating an external motivation. We hope that future research will build upon our study to obtain a more complete understanding of when and how social norms influence intergroup attitudes in children and adults, and thereby help to create a climate of mutual acceptance in today’s increasingly diverse societies.
