Being considered a co-national: Social categorization and perceived acculturation of immigrant peers

With increasing immigration, it is increasingly important to understand whether and when children consider immigrant peers as co-nationals. Using an experimental design, we examined among native-born preadolescents (8–13 years of age) in the Netherlands whether and when they perceive immigrant peers as co-nationals. First, and in agreement with the social categorization account, we expected that the use of dual identity (vs single ethnic identity) labels for immigrant peers leads to stronger co-nationality perceptions and a related stronger desire for close social contact. Second, and in line with the acculturation account, we expected that an early age of arrival in the country (vs a later age of arrival) leads to stronger perceived co-nationality and related contact desire. The ﬁndings support the acculturation account, especially among native-born children with higher national identiﬁcation. There was no evidence for the social categorization account. (cid:1) 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/ licenses/by/4.0/).


Introduction
With increasing immigration-related diversity, it is important to study children's responses to peers who have entered their country (Gönültas ß & Mulvey, 2019;Palmer et al., 2022).One potentially important factor in these responses is the degree to which children perceive peers with an immigrant background as co-nationals.According to the common ingroup identity model (CIIM; Dovidio et al., 2007), group boundaries and the potentially negative effects thereof can be transcended by emphasizing shared membership of an overarching category.Several studies, including research among children (Gaertner et al., 2008;Guerra et al., 2010Guerra et al., , 2021;;Vezzali et al., 2015), have shown that the perception and experience of a common identity helps to overcome the distinction between ''us" and ''them" by providing a shared sense of ''we."In the context of immigration, the national category can provide such a common identity and bring natives and those with an immigrant background closer together.There is some literature on young people's attitudes toward first-or later-generation immigrants (for a review, see Gönültas ß & Mulvey, 2019), but we know very little about whether and when children tend to perceive immigrant peers as co-nationals.
The current study was conducted in the Netherlands and used an experimental design to examine when and why native-born preadolescents (8-13 years of age) perceive immigrant peers as conationals.Preadolescence is an important period for the study of intergroup relations: Older children are increasingly able to recognize that others can belong to multiple categories at the same time, and this implies more flexible ways of dealing with group boundaries (Aboud & Amato, 2001).We tested two complementary theoretical accounts-a social categorization account and an acculturation accountby examining the roles of group labeling and age of arrival as possible conditions for children's conationality perceptions of immigrant peers.First, and in agreement with the social categorization account, we expected that the use of dual identity labels for immigrant peers (e.g., ''a Turkish Dutchman")1 would lead to stronger co-nationality perceptions than the use of single ethnic identity labels (e.g., ''a Turkish person living in the Netherlands").Second, and in line with the acculturation account, we expected that an early age of arrival in the country (at 1 year) would lead to stronger perceived conationality than a later age of arrival (at 9 years).Next, we included children's desire for close contact with immigrant peers as an additional outcome in our experiments, and we tested whether identity labeling and early age of arrival predict a greater desire for contact via children's co-nationality perceptions.Finally, we examined whether the effects of labeling and age of arrival depend on children's national identification.

Social categorization and ethnic minority labeling
People can be included or excluded on different grounds, and children learn that individuals can be categorized in different ways based on cues that are socially meaningful in a particular context.Categorizing people, and hence differentiating between them, happens everywhere in society and for a variety of reasons.It occurs in everyday life and is a prerequisite for group functioning, but it also forms the basis for ingroup bias from a young age onward (Bigler & Liben, 2007;Liberman et al., 2017).Labels or linguistic representations of social categories draw group boundaries that can have important implications for children's intergroup attitudes (e.g., Waxman, 2010).Already at preschool age, children tend to use linguistic labels for differentiating between social categories.And infants show a preference for ingroup members who are marked by the same label because these members are considered similar to themselves, are often more familiar, and can provide relevant group information (Begus et al., 2016;Buttelmann & Böhm, 2014;Cameron et al., 2001).One implication is that the use of dual identity labels (e.g., ''Turkish Dutch") can be expected to improve children's attitude toward immigrant peers compared with a single ethnic label (e.g., ''Turk").The former emphasizes a qualified form of common national belonging, whereas the latter draws a stronger group boundary by focusing on the distinctive ethnic background.According to the CIIM, not only a single common identity (''Dutch") but also a dual identity in which subgroup memberships and a common identity (''Turkish Dutch") are simultaneously emphasized leads to more positive outgroup attitudes (Dovidio et al., 2007).After middle childhood, children have developed multiple classification ability, which means that they can recognize that others can be both ingroup and outgroup members at the same time (Aboud & Amato, 2001).Thus, preadolescents should be able to understand that immigrant peers can be co-nationals in addition to having another ethnic background.
A few studies have examined whether group labeling can stimulate more positive outgroup attitudes among children.For example, Guerra et al. (2010Guerra et al. ( , 2021) ) conducted intervention research among chil-dren in Portugal (9-and 10-year-olds) that also included the use of specific labels.In the intervention, the differences between Portuguese-and African-origin children were first emphasized (by putting children in ethnically segregated teams and labeling them in terms of their ethnicity).Subsequently, in one condition the original labeling was maintained, but in the other conditions children's common national identity was stressed, either alone (by labeling their team as ''the Portuguese group") or in combination with their ethnic identities (by labeling their team as ''the Portuguese team with Portuguese-and African-origin children").It was found that the native children had more positive attitudes toward their African peers in the two common identity conditions.However, the intervention involved more than labeling per se given that seating arrangements and team interdependencies were also manipulated for stimulating a common identity.Moreover, the perceived co-nationality of the African-origin children was not examined as a central outcome of their intervention.Thus, it is unclear whether the mere labeling of immigrant-origin peers in dual identity terms is sufficient for children to perceive them as conationals.
In an experimental study that did examine the effect of mere labeling (Verkuyten & Thijs, 2010), native-born Dutch adolescents (14-18 years of age) were asked to evaluate immigrant-origin people using a dual identity label (e.g., ''Turkish Dutch") or a single identity label (e.g., ''Turkish").Participants in the dual identity condition provided more positive evaluations (small to medium effect size), and this effect was similar across four target ethnic minority groups.However, this study also did not investigate perceived co-nationality and focused on an older age group.In the current study, we directly examined whether children were more likely to perceive immigrant peers as co-nationals if they are labeled in dual identity terms (e.g., ''a Turkish Dutchman") than if single identity labels are used (e.g., ''a Turk living in the Netherlands").

Cultural adoption and age of arrival
The use of different labels not only refers to social categorization with the related boundary and ingroup and outgroup distinctions (''us vs them") but also might trigger different minority subgroup understandings.Especially in the absence of other information, it is possible that native-born majority children perceive dual identity peers as having adopted the mainstream culture (e.g., language, practices, beliefs) more than single ethnic identity peers (Verkuyten & Thijs, 2010).Dual identity labels like ''Turkish Dutch" might conjure up images of peers that are culturally more similar to the native-born majority group than single ethnic identity labels.In that case, there would be a confound between labeling and perceived acculturation.In the current study, we addressed this potential confound by also testing a cultural adaptation account for immigrants' perceived co-nationality.
A study in the Netherlands found that native-born majority preadolescents were more likely to perceive immigrant peers as co-nationals-and to subsequently like them more-if these peers had (partly) adopted Dutch majority culture (Verkuyten et al., 2014).Furthermore, research in Northern Ireland showed that children (4-11 years of age) were more likely to consider a Polish immigrant peer as Northern Irish (vs Polish; but not more Northern Irish and Polish vs Polish) if the immigrant had a Northern Irish accent rather than a Polish accent (Feeney et al., 2020).Thus, cultural adaptation appears to be a positive predictor of perceived co-nationality.An earlier age of arrival implies a longer stay in the country, and although not identical to cultural adaptation, such a longer stay facilitates immigrants' learning of the national language and adoption of mainstream cultural practices.Being a preadolescent who is raised in the country from 1 year of age compared with 9 years of age implies a clear difference in enculturation process.Thus, age of immigration can be used as a relevant proxy for perceived cultural adaptation, and we expected that the native Dutch children would be more likely to consider immigrant peers as co-nationals, and accordingly to express a stronger desire for contact with them, if they came to the country at an earlier age.

The role of national identification
How native-born majority children respond to immigrant peers may also depend on the psychological importance of their national belonging.In some European non-settler countries, the linguistic representations of nationhood and the native population tend to correspond.National labels such as ''German" and ''Dutch" are typically taken to mean ''ethnic German" and ''ethnic Dutch" also by ethnic minority group children (van de Weerd, 2020), and there are no separate labels to indicate the ethnic majority group (e.g., not ''German German" or ''Dutch Dutch"). 2ccording to the social identity perspective, high group identifiers are more likely to draw clear boundaries between their own group and other groups because they are more strongly motivated to achieve a distinctive and positively valued group identity.They tend to react rather negatively to blurred intergroup distinctions with the related distinctiveness threat (Branscombe et al., 1999;Crisp & Hewstone, 2007).Thus, in the Dutch national context in which the national category is implicitly defined in ethnic majority terms, children with high national identification might be less likely to perceive immigrant-origin peers as co-nationals and therefore have less desire for social contact with them.More important for the current study, higher identifying majority children are not only less likely to perceive immigrant peers as co-nationals but also probably less affected by dual identity (vs single ethnic) labeling.The reason is that higher national identifiers tend to endorse a stricter one-group majority perspective in which group boundaries should not be blurred by dual identities or subgroup identities.
However, national identification can be expected to increase the effect of early age of arrival as a proxy for enculturation and majority cultural adaptation.In general, majority group members have a preference for immigrants adopting the majority culture (Brown & Zagefka, 2011) because this implies that the majority culture is valued and maintained, which is especially important for higher national identifiers (Branscombe et al., 1999).Research among preadolescents has shown that high national identifiers like culturally assimilated immigrant peers more than culturally segregated peers (Verkuyten et al., 2014).Thus, to the extent that age of arrival indicates enculturation and cultural adaptation, the anticipated positive effect of early (vs later) age of arrival can be expected to be stronger for higher versus lower national identifiers.

Participants and procedure
Initially, 572 children (Grades 4-6) from eight schools in different parts of the Netherlands participated in the study.After acquiring informed parental consent, these children filled out a questionnaire in their classroom.This questionnaire covered various topics, including ethnic diversity, helping behavior, and school attitudes, and participation was anonymous and voluntary.For the current analyses, we selected those children who reported to have lived in the Netherlands all their lives and who identified both of their parents as ethnic Dutch and did not indicate to be non-Dutch (for measures, see below).This sample consisted of 378 participants (M age = 10.58 years, SD = 1.02; 183 girls) and was large enough to detect medium-sized main effects and interactions in a design with four groups (two experimental conditions) and one covariate (national identification) with power =.90 and p < .50.Of these children, 124 were in Grade 4 (M age = 9.56 years, SD = 0.64), 131 in Grade 5 (M age = 10.64 years, SD = 0.60), and 123 in Grade 6 (M age = 11.59 years, SD = 0.60).

Measures
The measures were completed in the order in which they are described.Although we used latent variables in our main analyses, we do provide information on the internal consistency here.

Ethnic background
Children's ethnic background was assessed by asking children to indicate the cultural groups to which their mother and father belonged.Response categories included Dutch and Turks, Moroccans, and Surinamese (the three largest non-Western minority groups in the Netherlands) and a remaining category where other groups could be listed.In addition to this, the items on national identification (see below) also included the response option ''I am not a Dutch person."Children were not considered as native Dutch if they selected this option at least once or if they had missing scores for the national identification items.This was the case for only 2.3% of the children with two Dutch parents.

Evaluations of immigrant peers
Four vignettes were used to assess the perceived co-nationality of immigrant peers as well as the desire for close contact in relation to their labeling and age of immigration (two experimental factors).Each vignette described a hypothetical 12-year-old immigrant peer, and the vignettes were presented in the following fixed order: a boy from Turkey (''Kadir"), a girl from Morocco (''Naima"), a girl from Greece (''Tekla"), and a boy from Spain (''Juan").The latter means that the national origin and gender of the peers were not manipulated, unlike labeling and age of immigration.A random betweenparticipants design was used; within each condition, each responding child was presented with the same version of the four vignettes.Labeling was manipulated by referring to the peer by using a dual identity label or a single ethnic label, and all peers were either 1 year old or 9 years old when they immigrated to the Netherlands.Thus, the first vignette, for example, read, ''Kadir is twelve years old, and he is from Turkey.He came to live in the Netherlands when he was [one/nine] years old.He is [a Turkish Dutchman/a Turk living in the Netherlands].".
Directly after each vignette, children completed a single question on how strongly the target peer was perceived as a co-national (''How Dutch do you think [. ..] is?"), which has been successfully used in previous research (Verkuyten et al., 2014), followed by two questions on desire for contact (''Would you want to play with [. ..]?" and ''Would you want to become friends with [. ..]?").The perceived conationality question was answered on a 5-point scale ranging from not at all to very much.The two contact questions were answered on 5-point scales ranging from definitely not to definitely, and for each vignette their correlation exceeded .79,indicating high internal consistency.Table 1 shows the mean scores for co-nationality and desire for contact (average of the two items) for each target peer.All scores were above the midpoint of the scale, p < .01,which indicates that children were relatively likely to consider the target peers as co-nationals and wanted to have contact with them.There were small to medium differences among the four target peers, g 2 = .025and g 2 = .028,respectively, which might be due to order effects, and therefore we did not further interpret these.Moreover, the responses to the same questions were consistent across the four target peers, with Cronbach's alphas of .96for perceived co-nationality (''how Dutch?"), and .93 for contact (eight items; ''play?" and ''become friends?" combined).

National identification
Children's national identification was assessed with the following three questions: ''Do you like being a Dutch person?"; ''Are you proud to be a Dutch person?"; and ''Do you think it is important to be a Dutch person?"These three questions were based on previous studies among Dutch preadolescents (e.g., Verkuyten et al., 2014), and the response scale ranged from 1 (no!) to 5 (yes!).Cronbach's alpha was .57.This moderate reliability indicates the need to correct for measurement error by using latent variables.

Results
We used latent variables and structural equation modeling in Mplus Version 8.5 (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 the CFI and TLI have values of .95 or higher and the RMSEA and SRMR are lower than .05.CFI and TLI values higher than .90and RMSEA and SRMR values lower than .10 are considered acceptable (Kline, 2011).

Latent constructs
First, not taking the experimental conditions into account, we tested a factor model for participants' national identification, perceived co-nationality, and desire for contact across the four hypothetical peers, and given the confound between gender and ethnicity in the vignettes, we explored whether the factor structure was invariant across participant gender.To minimize the ratio of parameters to cases, we included the means of the desired contact questions per vignette.In addition, error correlations were allowed between children's vignette-specific responses to the co-nationality and desired contact questions.The fit of this so-called correlated uniqueness model (Kline, 2011) was nearly acceptable, v 2 (37) = 221.376,CFI = .939,TLI = .910,RMSEA = .115,SRMR = .050,and multiple group analyses (see Table S1 in online supplementary material) indicated that it was difficult to obtain a well-fitting factor model that proved to be invariant across gender.Presumably, this was due to the confound between the gender and the ethnicity of the four target peers.Therefore, we retested the correlated uniqueness model after matching the peer evaluation scores to the perspective of the participants' gender.That is, we analyzed the evaluations of the first same-gender peer that children evaluated (the Turkish boy for boys and the Moroccan girl for girls), the second same-gender peer (the Spanish boy for boys and the Greek girl for girls), the first other-gender peer (the Moroccan girl for boys and the Turkish boy for girls), and the second other-gender peer (the Greek girl for boys and the Spanish boy for girls).After this gender matching, the fit of the model was acceptable to good, v 2 (37) = 156.545,CFI = .962,TLI = .944,RMSEA = .092,SRMR = .047,and there was evidence for full metric and partial scalar invariance across gender (see Table S1).In this model, national identification was weakly and not significantly associated with perceived co-nationality (r = .11,p = .06)and with desire for contact (r = .11,p = .08).However, the more children perceived the immigrant peers as Dutch, the stronger was their desire for contact with them (r = .44,p < .01).

Main effects
We continued by examining the effects of our experimental manipulations on perceived conationality and desired contact with the immigrant peers.Thus, we specified a path model that included the latent factors for both experimental variables (and their error correlations) and national identification.In this model, perceived co-nationality and desired contact were regressed on contrasts for labeling (dual identity = .50vs single ethnic identity = À.50) and early age of arrival (1 year old = .50vs 9 years old = À.50) and on participants' national identification. 3Desired contact was regressed on perceived co-nationality as well, to test whether the effects on the latter were mediated by the former, and all factors were also regressed on gender and age as control variables.The fit of the model was acceptable to good, v 2 (71) = 212.583,CFI = .957,TLI = .940,RMSEA = .073,SRMR = .044.The standardized regression coefficients are shown in Table 2.Note that the (total) effects for desire for contact are split into direct and indirect effects.Because our model included indirect effects, we used bootstrapping (1000 samples) to estimate the standard errors of all coefficients.
The experimental labeling of the immigrant peers did not affect the degree to which children perceived them as co-nationals or wanted to have contact with them.However, early age (vs older age) of immigration had significant positive effects on both variables, and the total effect on desire for contact was fully explained by the indirect effect via perceived co-nationality.Thus, children were more likely to see the immigrant peers as co-nationals and to subsequently desire contact with them if those peers had immigrated at an early age compared with a later age.Next, national identification had a nonsignificant effect (p > .05) on perceived co-nationality, but the control variables gender and age had positive significant effects.Girls were more likely to see the immigrant peers as co-nationals and to express a desire for close contact, and the latter effect could be partly explained by the former effect.Age was unrelated to perceived co-nationality, but older children reported a greater desire for close contact.

Interactions with national identification
Next, we tested our expectations that the experimental impact of labeling and early age of arrival depends on children's national identification.We used ''type = random" to estimate the interactions with these latent variables (and ''algorithm = integration" to obtain standardized effects).Of the two interactions, the one between national identification and age of arrival was positive and marginally significant for perceived co-nationality (b = .12,SE = .06,p = .055)and for desire for contact, even when controlling for perceived co-nationality (b = .11,SE = .06,p = .063).However, this interaction was in line with our expectations and was significant with one-sided testing.This indicates that the positive effects of earlier arrival were more pronounced for higher identifiers compared with lower identifiers.Fig. 1 shows the estimated scores for perceived co-nationality (and their 95% confidence intervals) for the two arrival conditions at different levels of national identification.The confidence intervals overlapped for respondents with national identification scores of nearly .50standard deviations or more below the mean.Thus, age of arrival had no effect for lower identifiers.However, for higher identifiers, a younger age of arrival led to stronger perception of co-nationality and a later age of arrival led to weaker perception of co-nationality.The total model is shown in Fig. 2.
Although it is not possible to examine indirect effects with the ''type = random" option, we could plot the nonstandardized indirect effect of age of arrival on desired contact (via perceived conationality) against children's national identification.As shown in Fig. 3, the 95% confidence interval for the indirect effect exceeded the value of zero but only for high identifiers (with scores above the mean on the latent variable).Thus, we could conclude that identification moderated the indirect effect of early age of arrival.

Additional analyses
In a last set of analyses, we explored whether the (non)effects of dual labeling and earlier arrival, and their interactions with national identification, depend on the age of the participants.First, we added the interactions of participant age (standardized) with the two experimental factors to the model shown in Fig. 2. The total, direct, and indirect effects of these interactions were not significant (see Table S4 in supplementary material).Next, we added the three-way interactions among participant age, national identification, and dual labeling and among participant age, national identification and early age of arrival as well as the two-way interactions of national identification with both factors and participant age.The direct effects of these three-way interactions were not significant, b = .03,SE = .06,p = .60and b = .06,SE = .06,p = .32,respectively, for perceived co-nationality and b = À.06,SE = .06,p = .25 and b = .10,SE = .06,p = .08,respectively, for desire for social contact.Thus, our findings appeared to be similar for younger and older respondents.

Discussion
The goal of the current study was to examine when and why native-born majority children perceive immigrant peers as co-nationals.A social categorization perspective suggests that perceptions and attitudes toward immigrants are related to processes of boundary drawing in which a stronger or weaker categorical distinction between ''us" and ''them" is made (Bigler & Liben, 2007;Liberman et al., 2017).However, these perceptions and attitudes might also result from the degree to which immigrants are considered to have adopted the majority culture, and the mere use of dual identity labels could not only imply social categorization but also indicate culture adoption (see Verkuyten & Thijs, 2010).We used two independent experimental manipulations to distinguish and test these two possibilities, and we examined the moderating role of national identification.In general, the findings are in line with a cultural adaptation interpretation and not with a strict social categorization account.
In support of a cultural adaptation interpretation was the relatively strong experimental effect of age of arrival on perceived co-nationality and desire for social contact.Preadolescent immigrants who arrive in the country at 1 year of age will be much more adapted to the mainstream culture than those who arrive at 9 years of age.Their early and relatively long process of enculturation implies a good knowledge of the national (here Dutch) language and familiarity with mainstream practices, norms, and beliefs.This makes it more likely that they are seen as co-nationals and that nativeborn children want to have social contact with them.The findings show that when immigrant peers were considered more strongly Dutch, children were more willing to have social contact with them (see Verkuyten et al., 2014).Furthermore, the cultural interpretation might also explain why we found that especially higher national identifiers responded more strongly to the difference between early and later age of arrival.Compared with lower identifiers, those who felt more strongly Dutch considered immigrant peers who arrived at a later age less as co-nationals and considered those who arrived at a young age more strongly as co-nationals.This pattern of findings makes sense from the perspective of cultural adaptation because higher identifiers are more focused on the ingroup culture and value that newcomers adapt to and adopt the mainstream culture (Branscombe et al., 1999;Brown & Zagefka, 2011).Thus, our findings support a cultural interpretation, but it should be noted that we used early age of arrival as a proxy for perceived cultural adaptation.Future research could examine the role of this perception more directly and in doing so could also examine which aspects of cultural adaptation-for example, language, practices, preferences, and beliefs-are especially meaningful for recognizing immigrant peers as co-nationals.
We tested the social categorization account by examining whether children consider dual identity labeled immigrant peers (e.g., Turkish Dutch) more strongly as co-nationals than single ethnic labeled peers (e.g., Turkish).The former implies a categorization as a (dual) national ingroup member, whereas the latter emphasizes an ethnic group distinction.The findings showed that there were no differences between the two experimental conditions in perceived co-nationality and the desire for social contact with immigrant peers.Furthermore, this lack of effects was found for both lower and higher national identifiers.In general, higher identifiers are more likely to draw clear boundaries between their own group and other groups and therefore are more responsive to social categorization processes (Branscombe et al., 1999).However, national identification had no negative main effect in this study, and the experimental labeling had no differential effects for level of identification, which further suggests that the single and dual identity labeling did not involve a difference in group boundary drawing.Peers with a dual identity were not more likely to be perceived as co-nationals than those with a single ethnic identity, which might be explained by the tendency in non-settler contexts, such as the Netherlands, to focus on the ethnic group belonging of ethnic minority members rather than their national group membership (Verkuyten, 2018).Future research could examine this interpretation further by adding an experimental condition in which immigrant-origin peers are exclusively defined in national terms of the host society (e.g., as Dutch).This might lead to a stronger perception of co-nationality, but at the cost of not recognizing the specific ethnic identities of newcomers.
Despite a positive effect of age on desire for contact that corresponds with the finding that ingroup bias tends to decrease during late childhood (Raabe & Beelmann, 2011), the (non)effects of the experimental manipulations were similar for both younger and older children.We focused on preadolescents because their multiple classification ability would allow them to respond to dual labels (see Aboud & Amato, 2001).However, there was no effect of labeling for all age groups, and based on findings by Feeney et al. (2020), who found little evidence for age-related changes in nationality views among 4-to 11-year-olds, we suspect that similar findings might be obtained in younger children as well.We have no reason to expect different outcomes for older teenagers either.Earlier research found positive effects of dual labeling in this age group, but this might be due to the confound between labeling and perceived acculturation (Verkuyten & Thijs, 2010).In addition, our analyses yielded some interesting main effects for gender of the participating children.Compared with boys, girls were more likely to regard the immigrant peers as co-nationals and to express a desire for positive contact with them.This corresponds with earlier findings that girls are more open to ethnic others (e.g., Jargon & Thijs, 2020), which might be related to average gender differences in socialization whereby girls are socialized to be more other-oriented and prosocial than boys (Brody, 1999).
Our findings might have some implications for the use of group labels in research on children's intergroup relations.Such labels are sometimes scrutinized for being divisive and exclusionary and for constructing otherness (e.g., Moffitt & Juang, 2019).There is little doubt that this can happen and that one must be sensitive to this possibility in research.However, our results indicate that labeling per se does not need to be a problem.In fact, researchers often need to use group labels if they want to examine children's prejudicial attitudes toward particular minority groups.Furthermore, in contrast to immigrant countries such as the United States and Canada, in countries like the Netherlands and Germany there is no clear distinction between the national category label and the ethnic majority label, and ethnic minority children themselves sometimes do not self-identify as nationals (e.g., van de Weerd, 2020).In these contexts, the use of dual identity labels might be confusing to (young) research participants, and single ethnic identity labels are simpler and closer to the ways in which children themselves think about ethnic group differences.
To evaluate the findings, we consider a few qualifications.First, we examined children's responses to hypothetical immigrant peers who were described with a few words only.Thus, just as with other experimental research, there is the issue of ecological validity and it is an open question whether children would respond similarly to the labeling and age of arrival of ''real life" peers about whom more information might be available.However, there are also many situations in which children make judgments about newcomers based on very limited information that they hear from others.Furthermore, the fact that a weak and proximal manipulation of cultural adoption (age of arrival) quite strongly affected perceived co-nationality and the related desire for positive social contact suggests that even relatively mild communications can influence the degree to which children consider newcomers as ''one of us nationals."It is likely that more extensive (narratives) and vivid (visual and auditive) manipulations may yield stronger effects.
Second, we examined whether and how children's judgments were moderated by their national identification, but there might be other important moderating factors such as perspective taking ability and cognitive flexibility.For example, children who are better able to take the perspective of the other might be more focused on the situation of immigrant peers and less sensitive to their degree of acculturation, whereas children with higher multiple classification ability might be more responsive to single versus dual category labels.Furthermore, peer norms, contact with immigrant peers, and feelings of outgroup threat might be important for whether and when labeling and age of arrival matter for the perception of newcomers as co-nationals.
Third, we examined children's responses to immigrant peers who had particular combinations of national origin and gender.We focused on immigrants from Turkey and Morocco and from Spain and Greece, and in Dutch society the former two groups have lower social status and are accepted less by adults than the latter two groups (e.g., Gijberts et al., 2016;Hagendoorn, 1995).Children's average responses to the vignettes are somewhat in line with this status group difference (Table 1), but our design prevented us from testing whether immigrants' national origin (and gender) affected their perceived co-nationality as well as the experimental impact of labeling and early age of arrival on this.The internal consistencies and the factor structure of our measures indicated that it was meaningful to aggregate children's responses toward the different immigrant peers.Yet, future research could systematically examine the effects of labeling and perceived acculturation on majority children's evaluation of immigrant peers who differ in status group attributes.
To conclude, with increasing immigration, it is increasingly important to understand whether and when children consider immigrant peers as co-nationals.Our experimental findings indicate that age of arrival matters for being considered a co-national and the related desire for social contact, which supports a perceived acculturation interpretation for co-nationality.We found no evidence for dual (vs single) group labels having an effect on perceived co-nationality of immigrant peers, which suggests that this form of social categorization does not matter.Although we did not find any age differences, future studies should examine whether these findings replicate among a broader range of age groups.In addition, the findings could be examined in other non-settler European societies such as Germany and France as well as settler societies such as Canada and the United States that have a much longer history with the accommodation of immigrants and in which dual identity labels (e.g., Mexican Americans) are much more common.

Fig. 1 .
Fig. 1.Interaction effect of national identification (ID) and early age of arrival on perceived co-nationality.AGE9, 9 years of age at arrival; AGE1, 1 year of age at arrival.Lower and Upper are 95% confidence interval boundaries.

Fig. 3 .
Fig. 3. Indirect effect of early age of arrival on desired contact (via perceived co-nationality) depending on national identification (ID).Lower and Upper are 95% confidence interval boundaries.

Table 1
Evaluations of immigrant peers.

Table 2
Standardized effects on perceived co-nationality and desire for contact.Note.Standard errors are between parentheses.Bootstrapping was used (1000 samples).PCN, perceived co-nationality.aIncludes (nonsignificant) indirect effects via national identification as well.