In psychological research, “feeling good” is often measured as subjective well-being (e.g., Diener et al., 1999). Subjective well-being is defined as a construct consisting of three often substantially correlated aspects, namely affect, domain-specific satisfaction (e.g., work, relationships, …), and general life satisfaction (Diener et al., 1999). Diener et al. (1985) describe subjective life satisfaction as what one considers to be “a good life.” This form of subjective happiness can be described as “a global assessment of a person’s quality of life according to his own chosen criteria (Shin and Johnson, 1978; p. 478), or as “the harmonious satisfaction of one’s desires and goals” (Chekola, 1975). Subjective well-being can be associated with positive outcomes on a variety of aspects of life, such as higher odds of being married or employed, higher income (Marks and Fleming, 1999), and a reduced risk of cardiovascular disease (Boehm and Kubzansky, 2012).

Although there are many antecedents of well-being, social support is considered to be one of the more important bases. Indeed, social support has been shown to correlate strongly with psychological well-being (e.g., Cohen and Wills, 1985; Chou, 1999; Gallagher et al., 2008; Kong et al., 2013; Lee et al., 2018). Social support has been defined as communication from others that is being interpreted in terms of being cared for and loved, being esteemed and valued, and being part of a network of communication and mutual obligation. In other words, social support can be seen as a conglomerate of emotional support, recognition, and belongingness (see Cobb, 1976).

A distinction in support types has been provided by Helgeson (1993), namely emotional, instrumental and informational, and the various forms of aid and assistance comprised by the term social support can be supplied by different sources, e.g., family members, friends, neighbors, and others (Barrera et al., 1981). Furthermore, a distinction can be made between perceived support, i.e., the extent to which people believe support is available, and received support, i.e., the amount of support that has actually occurred (Helgeson, 1993). Cohen and Wills (1985) suggest that social support influences well-being through the provision of positive emotions and predictability in life, as well as through encouraging a positive self-concept.

The positive relationship between social support and psychological well-being has been reported in children as well (e.g., Rigby, 2000). Child well-being has been described as the outcome of the interaction between risks and resources available to children, thus constantly changing with both their evolving capacities as well as with the variations in their personal situation (Bradshaw et al., 2007). Moreover, as children have a social network that is substantially different from that of adults, both in terms of size (Wrzus et al., 2013) as well as in terms of the nature of relationships, i.e., more vertical than horizontal relationships with adults (Russell et al., 1998), social support may impact child well-being differently than adult well-being (e.g., Hartup, 1989).

In a meta-analysis of the relationship between social support and general well-being in children, Chu et al. (2010) showed that perceived support is associated more strongly with well-being than other measures of social support such as, for instance, enacted support and size of social network. Furthermore, these authors also reported differential effects of social support for different sources, and they particularly showed that social support of parents and of teachers and school personnel is especially important for well-being. Social support of friends was reported to have a somewhat limited effect. It stands to reason that the huge importance of social support for general well-being has been revealed for well-being in specific contexts as well, such as the school context. Perceived teacher support has been found to be important for school well-being of adolescents. Specifically, in a study on adolescents’ emotional, behavioral and cognitive school engagement, it was revealed that although family support is directly linked to school engagement, the greatest impact comes from perceived teacher support (Ramos-Diaz et al., 2016). More specifically, scores on measures of school related affect, like school satisfaction (measured with items such as “I enjoy going to this school” and “I like the classes that I am taking”) and school engagement (measured with items such as “I find school fun and exciting”) coincide with high perceived teacher support, and to a lesser extent, both high perceived parental and peer support (Rosenfeld et al., 2000). Gutiérrez et al. (2017) showed a direct effect of perceived teacher support on satisfaction with school, as well as an indirect effect through school engagement. A longitudinal study demonstrated that children with high teacher-child closeness and low teacher-child conflict reported higher school liking 2 years later (Wang et al., 2016). In the current study, we look at the contribution of perceived social support from teachers, as well as from parents and peers, on both general well-being and school well-being.

Based on the literature mentioned above, we formulate the following hypothesis:

Perceived teacher support has been shown to be an important factor in children’s academic development as well as in their school well-being (e.g., Roeser et al., 2000; Natvig et al., 2003; Danielsen et al., 2009; Roorda et al., 2011; Tian et al., 2015; Ruzek et al., 2016). In the present context, our aim is to answer two questions about social support. First, are there differences between majority and minority pupils in their perception of social support, in well-being, and does being part of the majority or minority group moderate the relationship between perceived social support and well-being? Second, are school diversity and class diversity relevant context variables in perceived social support and in well-being, and do these context variables moderate the relationship between perceived social support and well-being?

The original sample consisted of 13,871 children from 163 primary schools located in the Belgian province of Antwerp. One student was removed because he or she did not fill in his class and school information correctly. The 163 schools comprised a total of 754 classes (4th, 5th and 6th year, ages 9–12). Mean age was 10.03 years (SD = 0.944). Virtually identical numbers of boys (50.4%) and girls (49.6%) participated, but due to an administration error gender was not successfully registered for part of the sample (39.1% of the sample).

Given that Hypothesis 2 focused on a formal comparison of Belgian majority youth and minority youth, we decided to limit our scope for the minority youth to pupils which could easily and unambiguously be classified as belonging to either the Middle Eastern minority group or the Eastern European minority group. Both groups could easily be identified from the data using the question “which language do you speak at home?” We refer to Valcke et al. (2022) who used an identical methodology and classification. Middle Eastern minority pupils were classified as such when participants indicated to speak one of the following nine languages at home: Arabic, Turkish, Urdu, Pakistani, Afghan, Kurdish and Persian. These languages are spoken in predominantly Middle Eastern countries (Infoplease, n.d.). A total of n = 1,136 pupils was assigned to the Middle Eastern minority group. Furthermore, participants who indicated to speak one of the following languages at home, were classified as Eastern European minority group members: Armenian, Bulgarian, Georgian, Hungarian, Yugoslavian, Macedonian, Montenegrin, Ukrainian, Polish, Rumanian, Russian, Serbian, Slovakian, Czech and Croatian. A total of n = 434 pupils was assigned to the Eastern European group. The total number of pupils of Belgian descent was n = 10,645. A further n = 1,656 were not retained in our final analyses, because they could not unambiguously be classified as belonging to a specific cultural group based on the language question. For a comprehensive overview of the used classification system, we refer to Valcke et al. (2022). As such, our final sample comprised N = 12,215 (87.1% Belgian majority, 9.3% Middle Eastern minority, 2.6% Eastern European).

Data collection was done by the Welfare and Health Department of the Province of Antwerp. Based on the list on the website of the Ministry of Education, all primary schools of the Province of Antwerp (N = 583), were contacted and invited to register for participation. The schools that accepted the invitation (N = 163) were sent a link to the online questionnaire together with the test instructions. A sample letter for the written consent of the parents was included. The questionnaire consisting of 72 questions was administered between mid-October and early December 2016. To test for the comprehensibility of the items included in the questionnaire, a pilot session was done in a primary school in Beveren in the province of East Flanders.

Table 1 provides an exhaustive overview of reliability and validity statistics (Cronbach’s Alpha, McDonald’s Omega, Composite Reliability and Average Extracted Variance) for each measure of the focal variables.

Pearson correlations among the study’s variables are reported in Table 2.

Initial analyses revealed significant gender differences for all three types of perceived social support (teacher: Mb_oys = 2.97, SD_boys = 0.83, M_girls = 3.07, SD_girls = 0.71; F(1, 7,439) = 31.61, p < 0.001; parental: Mb_oys = 3.60, SD_boys = 0.56, M_girls = 3.62, SD_girls = 0.57; F(1, 7,422) = 4.13, p = 0.042; peers: Mb_oys = 2.86, SD_boys = 0.82, M_girls = 3.20, SD_girls = 0.71; F(1, 7,422) = 356.15, p < 0.001), and school well-being (Mb_oys = 2.58, SD_boys = 0.72, M_girls = 2.79, SD_girls = 0.64; F(1, 7,487) = 178.52, p < 0.001) and general well-being (Mb_oys = 5.64, SD_boys = 1.01, M_girls = 5.56, SD_girls = 1.07; F(1, 7,487) = 11.44, p < 0.001). Furthermore, we also found significant relationships between age and all the focal variables (teacher support: r = −0.04, p < 0.001; parental support: r = 0.04, p < 0.001; peer support: r = 0.08, p < 0.001; school well-being: r = −0.10, p < 0.001), except general well-being (r = 0.00, p = 0.990). As such, we decided to retain these demographic covariates in our main analysis.

To satisfy our main research aims, we fitted a single (generalized) linear mixed model (GLMM) with the Lavaan package (Rosseel, 2012) in R (R Core Team, 2022), with all three forms of perceived support as independent variables, and general well-being and school well-being as outcomes. Minority status (Belgian majority youth/Middle Eastern minority youth/Eastern European minority youth) and school diversity (operationalized by the Simpson index) were included as moderating variables, gender and age were included as covariates. In an additional analysis we also included interaction terms between gender and age and the focal predictors (perceived support, minority status and school diversity). Given that these interactions were all found to be insignificant, we decided to report the output of the more parsimonious model below (i.e., with only main effects for gender and age). Missing data were handled within the analysis model, by means of Full Information Maximum Likelihood Estimation. The model output (parameter estimates, standard errors (SEs), significance tests and 95% confidence intervals [CIs]) is displayed in Table 3. In what follows, we present a brief overview of the most important findings for both forms of well-being.

The results of our study both replicate and nuance previous research findings. In line with Hypothesis 1a, our results show that all three sources of perceived support are positively related to school wellbeing and general well-being. In addition, and in line with Hypothesis 1b, it was found that, in general, the relationship between perceived social support and school well-being is the strongest for teacher support, compared to parental support and peer support. Our results match with those of the previous studies, revealing that, in general, teacher support is the most important source for school well-being (Rosenfeld et al., 2000; Ramos-Diaz et al., 2016). Lastly, in contrast to what was expected in Hypothesis 1c, parental support did not show the strongest relationship with general well-being (compared to teacher support and peer support). In contrast to the findings of Chu et al. (2010), peer support seemed to be an equally important source of support for general well-being as parental support. The relationship between teacher support and general well-being showed to be weaker and thus less crucial.

The present research did not obtain supportive evidence for Hypothesis 2 (i.e., children of minority groups will report lower perceived teacher support) Indeed, we could not find any significant differences in perceived teacher support between the three cultural groups in this study – at least, not when type of support was entered into the equation. This result thus does not corroborate studies conducted in the North American context indicating that children of minority groups have less positive relations with their teachers (Hamre and Pianta, 2001; Hughes and Kwok, 2007). A few tentative explanations for this discrepancy can be envisaged. For example, it can be argued that the European and American educational systems differ. Furthermore, the minority groups and their migration background in our Belgian sample are substantially different from those studied in previous investigations conducted in North-America. Indeed, the studies conducted in the United States have mainly focused on African American children as minority group members, whereas we investigated Middle Eastern and Eastern European minority pupils. Unlike the American context, the migration waves of these minorities to Europe have begun only recently, 50 years ago. Moreover, this type of migration has continued up to today. The parents of the children under study in the present investigation can therefore show differing levels of integration in the host culture, and may have different levels of acquaintance with the educational system. Such unfamiliarity with the Belgian school system may have culminated into lower levels of expected support – compared to the US situation – and may consequently have elevated the perception of support when it was de facto provided by the teachers. However, a most marked difference between the present study and previous investigations is the inclusion of pupils pertaining to a younger age group, which distinguishes it from those studies conducted in the United States which investigated secondary school pupils. Hence, one possibility we cannot exclude is that the negative perceptions of teacher relationship and support might grow over the years, and thus becomes negative only in secondary school. Follow-up studies in the United States on primary school samples is thus needed to know if the previously found ethnic-cultural differences in the perception of teacher support are already cultivated at a younger age.

Moreover, the results did provide an answer for Research question 1. The results showed significant differences between the Belgian majority group and both cultural minority groups for perceived parental support and peer support, showing that minority members report lower feelings of both sources of support compared to the majority group. Both minority groups did not differ from each other. Through this lens, in relative terms, the minority pupils in the present sample thus have rated teachers support at a high level, as compared to parents and peer support. In other words, even though minority pupils were inclined to feel less support of parents and peers, they did not experience less support from teachers. From this it can in fact be concluded that teachers elicited a fairly high level of perceptions of social support among their pupils – at least, relative to the other focal socialization figures (parents and peers).

With respect to Research question 2, no significant main effect was obtained of minority status on school well-being. However, with regard to Research question 3, our results revealed a weaker relationship between perceived teacher support and school well-being for Eastern European pupils, compared to the majority group. This finding runs counter to previous studies showing no significant moderating effect of minority status on the teacher support-school well-being relationship (e.g., Bird and Markle, 2012; Griffith, 2002). Furthermore, this results does not mirror our observations among Middle Eastern minority group pupils, who benefited from teacher support (in terms of school well-being) to a similar extent as majority group pupils. One potential explanation for this finding may reside in the fact that the Eastern European minority group constitutes a relatively “new” migration group, compared to the Middle Eastern group (Myria, 2016). Although migration from countries like Poland to Belgium has remained quite stable during the 20^th century, migration waves from various other Eastern European nations have only accelerated over the last two decades. In this respect, they clearly differ from the Middle Eastern group, who are in fact often second-or even third-generation migrants. As such, there may be various other barriers for Eastern European minority pupils – e.g., unfamiliarity with the Belgian educational system, language barriers, etc. – that prevent teacher support from enhancing school well-being. Further research incorporating these variables is needed to answer these questions.

Secondly, the results showed that for both minority groups, perceived parental support was more strongly related to school well-being, compared to the Belgian majority group. A similar relation has been shown with academic achievement as an outcome variable (Ceballo et al., 2014). Specifically, Ceballo et al. (2014) revealed that the positive relation between parental involvement and academic achievement was stronger for immigrant youth. Our results add to these scholars’ findings the notion by showing that minority pupils are not only more responsive to parental support when it comes to school achievement, but also in terms of school well-being.

In addition, for Middle Eastern minority group pupils, a weaker relationship between perceived peer support and school well-being was also found, compared to the majority group. An explanation this weaker relationship may be that students’ peers are part of a different group. Minority students are clearly outnumbered by Belgian majority students in our sample, and we therefore can assume that most pupils have a number of cross-ethnicity friendships in their school social network. Especially for the Middle Eastern pupils, this often entails interactions with pupils that have very different cultural customs and traditions. If these relationships would be less harmonious than same ethnicity relationships, then support perceptions may drop. However, as we did not administer the ethnicity of the peer social network in the present study, these hypotheses await further empirical test.

Hypothesis 3 stated that school diversity would have beneficial effects on school well-being. Hence, besides the effect of individual-level minority status on individual perceived support and reported well-being, we also looked into the cross-level effect of school-level diversity. School diversity was calculated based on students’ home language using the Simpson index. To the best of our knowledge, only two previous studies investigating school well-being have used the Simpson index as a measure of school diversity, and have included both cultural minority and cultural majority groups in the same sample. The first of these studies has shown that higher school diversity yields positive effects, because it is associated with feeling less lonely, less victimized, and safer, regardless of students’ ethnicity. Thus, the more equally each ethnic group is represented in a school, the better students feel (Juvonen et al., 2018). The second study points towards similar benign effects of school diversity – at least for White majority students. Specifically, school well-being (operationalized as suspension rates) was significantly and negatively associated with school diversity among White (but not Black) students (Bottiani et al., 2016). Our results, however, failed to corroborate these prior studies and Hypothesis 3: No significant main effects of school diversity on the well-being variables were found. In addition, in answer of Research question 4, no significant moderation effects were found, indicating that school diversity did not affect the relationship between perceived support (teacher, parental and peer) and both types of well-being. One potential explanation for these unexpected results may be that school diversity was less “visible” in our study, compared to the work described above. That is, school diversity has been characterized in these studies in terms of the presence of members of clearly distinct and salient racial groups – for example in terms of skin tone (e.g., Black, Latino, Asian minority pupils). Conversely, in the present research, school diversity was operationalized in terms of the number of languages spoken at home. It stands to reason that language diversity is a less salient school attribute than racial/ethnic diversity, and the former type of diversity may simply not be sufficient to yield the previously observed benign effects. Another possible explanation for our diverging findings is that, although school diversity was found to be medium-to-high in general (i.e.,. D = 35), this substantial average diversity did not translate into equality in numbers of the various (language) groups. Indeed, in our sample, certain groups were clearly overrepresented (e.g., Dutch-speaking, Turkish-speaking), whereas others were noticeably underrepresented (e.g., Nepali-speaking, Finnish-speaking). As Juvonen et al. (2018) observed, equal group representation is particularly key for school well-being, and these intergroup differences in terms of group size may thus have further prevented school diversity to yield its purported positive effects. Further research implementing alternative diversity indicators could help clarify such questions.

We have now only discussed the effects of students’ minority status on perceived social support and well-being. However, we also included gender and age as control variables in our study, as well as the interaction effects of gender and age with minority status. Girls perceive more support and report higher levels of school well-being, but lower levels of general well-being. Our results also show that girls especially perceive greater teacher and peer, and to a lesser extent – but still significantly – more parental support. This replicates results of previous studies showing that girls report higher levels of teacher (Ellonen et al., 2008) and peer (Ellonen et al., 2008; Rueger et al., 2010) support. With regard to parental support, our results did not corroborate a previous study of Helsen et al. (2000) who showed that there are no differences in perceived parental support between boys and girls. The difference in perceived teacher support may be partially explained by the interaction style from teachers towards students, which may generally include more disciplinary actions towards boys than towards girls, which in turn might cause boys to perceive teachers as less supportive (Finn and Rock, 1997). Because of such an interaction style, boys may also perceive less parental support. However, this would not explain the finding of lower perceived peer support in boys. Another, possibly additional, explanation of our results can be found in the fact that boys and girls share different beliefs about communality (i.e., the needs and desires of the interaction partner determine social behavior) versus exchange (i.e., social behavior is determined by the comparability of benefits for each interaction partner) in relationships, and that these different beliefs might influence perceptions of support (Murstein and Azar, 1986; Jones and Costin, 1995). It has been shown that girls cherish communality beliefs more than boys, which relates positively to perceptions of support. In contrast, boys score higher on exchange beliefs than girls, which relates negatively to support perceptions (Colarossi and Eccles, 2003).

We also obtained some interesting age effects. Specifically, both types of well-being declined with increasing age, and we noted a shift in the importance of support sources as children grow older. Previous studies suggest that the importance of the different support sources varies with age (e.g., Tian et al., 2015; Wang and Eccles, 2012) indicating a shift in social priorities from adults to peers happening in adolescence (Wang and Eccles, 2012). In the present study, the effects of age were rather modest, but still we found a similar pattern in a sample of children with an average age as young as 10.3 years. Specifically, whereas perceived peer and parental support increased with a small but significant effect, a decrease in perceived teacher support with age was also observed. One possible explanation for this decrease might be the growing need for autonomy as children grow older, combined with the authority roles of teachers (e.g., Claes et al., 2001; Cattley, 2004).

Finally, The present results revealed that the effects of minority status on school well-being were not significantly moderated by gender and age, implying that the effects of minority status applied to an equal extent for boys and girls, and for younger and older pupils A last interesting result with regard to the demographic variables, is that teacher support had its most beneficial effects on school well-being for boys. As we argued above, the interaction style towards boys, with more disciplinary actions towards boys, lead to relatively low levels of perceived teacher support. Because of this low level of perceived teacher supports, incremental gains in this type of support may have greater effects.

In terms of practical implications, these latter results seem particularly relevant for both educational practitioners and policy-makers in Belgium. On the one hand, teachers may have to be attentional to support male pupils (in general, not only those with an ethnic minority background) in a sufficient manner, since this group may be most beneficially influenced by these acts of support. Additionally, policy-makers should pay extra attention to the decrease in well-being with increasing age, and how this decline may be buffered in a school context.

The present research also suffers from some potential limitations. For example, it must be noted that we only sampled primary school pupils pertaining to two ethnic-cultural minority groups. Given that, in the Belgian context, diversity is characterized by the presence of a plethora of ethnic-cultural groups, it seems conceivable that the lack of any main effects of minority status may be an artifact of the absence of other ethnic-cultural groups in our research design. Future studies could circumvent this limitation by focusing on the perspective of a wider array of ethnic-cultural minority groups.

Another possible limitation is that we only had information on the languages the children in our sample spoke, but not on where they and their parents were born. A further challenge for the categorization by language is the fact that pupils who indicated to speak only Dutch at home were automatically categorized in the ‘majority group’. It is however, possible that some of these pupils actually belong to a family of third or fourth generation immigrants in Belgium². We here want to refer to the formal definition that is often used in research using Belgian samples to classify ethnic minorities. Specifically, the definition used to categorize people as ‘allochthonous’ pertains to persons residing in Belgium, regardless of whether they possess Belgian nationality, at least one of whose parents or grandparents was born outside Belgium - commonly outside West Europe - and who have a disadvantaged position in society because of their ethnic origin (Brans et al., 2004). Following this definition, one is not considered to be “allochthonous” anymore when he/she belongs to the fourth generation. Actually, this definition implies that people from the fourth generation do not differ much from native Belgians and for this reason they are not categorized as ethnic minorities anymore. We have cross-validated the present language variable in another large sample where we also asked the participants to indicate whether they have a (grand)parent who was born in a non-West European country. From the 3,745 participants who indicated to speak Dutch exclusively at home, only 70 (or 1.87%) indicated to have a (grand)parent born in a Middle Eastern country. Hence, we believe that the number of wrongly classified Middle Eastern pupils is limited and unlikely to have a large influence on the results. However, future studies should supplement the language use variable with direct questions about the migration history of their parents.

Related to the previous point, it should be stated that we did not have any information about how our primary school pupils identified in terms of minority group membership (and thus on how the participants socially identify with the cultural group in which they are categorized). In this vein, it should be stressed that reliance on linguistic characteristics to determine whether a given participant is a minority group member or not can be tedious, and this aspect of our research design could have obscured factual differences between the studied ethnic-cultural groups. We therefore strongly encourage future researchers to investigate the effects of minority status on school well-being by operationalizing minority group membership in terms of how the youth population under study self-identifies, in order to scrutinize the robustness of the present results.

Finally, although our findings with respect to the moderating role of minority status are interesting in their own right, they should nonetheless be interpreted with caution. That is, a closer look at Table 3 also reveals that, following Cohen (1988); pp. 413–414) classification of magnitudes of R² effect sizes (i.e., “small” = 0.02, “medium” = 0.13, “large” = 0.26), the effect sizes associated with each of these interactions were all rather small (i.e., R²s < 0.01). In this regard, it should be noted that our sample size was quite substantial, and, as Hayat (2010) notes, “An increasingly large sample size yields a decreasingly smaller p value, [sometimes] regardless of scientific importance.” In other words, given that large sample sizes can inflate p-values – and hence, lead to overestimation of the practical significance of an effect – we must thus refrain from drawing strong conclusions based on these observed, yet rather small effects. Further research is needed to substantiate the robustness of the moderating role of minority status on the relationship between perceived social support and school well-being. In addition, it is noteworthy that the presently obtained effects of minority status are rather modest compared to the magnitude of the effects of gender and age. In primary schools, cultural ethnic background may have more limited effects than in other age groups. One of the interesting avenues for further study, therefore, is to probe into the role of minority status in a sample with a wider age range.

The present study investigated the joint effects of perceived social support, minority status, school diversity, age and gender on school and general well-being. As predicted, teacher support emerged as the most relevant source of support predicting school well-being, and peer support was shown to play an equally significant role in general well-being as parental support. Minority status was shown to moderate these associations, although we must nuance these findings because of the relatively small associated effect sizes. School diversity did not yield any relevant effects. Taken together, and given that perceived teacher support was also found to be positively associated (albeit less strongly than peer and parental support) with general well-being, the present results thus highlight the pivotal role that teachers can play in the wellbeing of primary school pupils – both in the education context and beyond – of all ethnic groups.