Disentangling the Role of Migration Background and SES in Student‐Expected Interpersonal Teacher Behavior

ABSTRACT

We investigated student‐expected differences in teachers' interpersonal behavior toward fictitious classmates with different migration and socioeconomic backgrounds, with teacher beliefs as potential moderators. Sixty‐eight teachers (M _age = 41.55, 70.6% female), and 1186 students in Grades 4–6 (M _age = 10.17, 47.8% girls) participated. A conjoint experiment systematically investigated the relative effects of classmates' migration background and socioeconomic characteristics (parental income, education, and the child's favorite hobby) on students' expectations of teachers' nice and angry behavior toward those classmates. Bayesian posterior distributions of Marginal Means were calculated for each child characteristic. Students expected teachers to behave more positively toward children with poor (vs. rich) parents, high‐educated (vs. low‐educated) parents, and with reading (vs. playing games) as favorite hobby. A child's migration background and teachers' beliefs about multiculturalism and meritocracy were not linked to student‐expected interpersonal behaviors. According to students, teachers mainly differentiate in their interpersonal behavior based on the socioeconomic characteristics of children.

1. Introduction

Nowadays, there is a lot of attention in both research and practice on how to diminish ethnic inequalities in schools. One important angle from which to look at this is provided by teachers' interpersonal relations and interactions with their students. Several studies have shown that teachers tend to behave more negatively toward ethnic minority students, leading to ethnic inequalities in educational outcomes (Denessen et al. 2022). Particularly for children in late childhood, there are many developmental and academic benefits of emotionally close and conflict‐free teacher‐student relationships (e.g., Archambault et al. 2024). Also, if ethnic minority students perceive an ethnic bias amongst their teachers, they are less likely to put effort into schoolwork (McKown and Weinstein 2002). Thus, teachers' group‐based differentiation in the interpersonal treatment of their students might contribute to educational inequality.

It is well‐known that students' ethnic background and socioeconomic status (SES) are intertwined and that both influence student outcomes (Codiroli Mcmaster and Cook 2019). Yet research into ethnic educational inequalities often neglects the role of inequalities based on students' socioeconomic background. Although the confounding relationship between ethnic background and SES is often recognized by controlling for SES (e.g., McKown and Weinstein 2002), there is little attention for the practical implication that diminishing SES inequalities in school could also further reduce ethnic inequalities, at least partially. Yet, attempts to counter negative differentiation in interpersonal teacher behavior based on students' ethnic background may be less successful if differentiation based on SES is not, on its own, considered. Hence, to contribute to equal educational opportunities, it is important to disentangle differentiation in interpersonal teacher behavior based on ethnic as well as socioeconomic child characteristics by considering them separately and in isolation.

The current study aims to do so by investigating Dutch students' (Grades 4–6) expectations of differences in teachers' interpersonal behavior toward fictitious classmates with combinations of different migration ¹ and socioeconomic backgrounds. We innovatively apply a conjoint experiment (CE) to systematically and simultaneously investigate the separate, isolated effects of classmates' migration background (native vs. immigrant) and socioeconomic characteristics (parental income, education, and child's favorite hobby) on students' expectations of positive versus negative teacher behavior toward those classmates. We take student reports into account, because it is especially important that students perceive that their teacher does not differentiate between students, as research has shown that this has detrimental effects for those students perceiving that they are treated less positively (McKown and Weinstein 2002). Moreover, if students perceive an unequal treatment toward peers from specific groups, this could have detrimental effects for students' intergroup outcomes (Karataş et al. 2023) as they tend to use the relationships between their teacher and other students in the classroom to form impressions about those students (Hughes and Kwok 2006).

Yet, it is not self‐evident that every teacher (largely) differentiates in their interpersonal behavior based on students' migration and socioeconomic backgrounds. This may depend on several factors, and in particular on teacher beliefs (Pianta et al. 2003). The current study also aims to investigate the link between teacher beliefs that are relevant to dealing with ethnic and socioeconomic diversity (i.e., teacher‐reported multiculturalism [MC] and belief in school meritocracy [BSM]), and student‐expected interpersonal teacher behavior toward students with different migration and socioeconomic backgrounds. Although the link between teacher beliefs and actual teacher practices is not straightforward (see, e.g., Buehl and Beck 2015), there is evidence that such beliefs do manifest in the classroom (see, for a review about the effects of teacher's beliefs about the malleability of intelligence; Laine and Tirri 2023, or about teacher's language diversity beliefs; Gallagher and Scrivner 2024).

Overall, the current study aimed to answer the following research questions; (1) What child characteristics (migration background, parental income, parental education, favorite hobby) are more relevant for student‐expected differences in interpersonal teacher behavior toward different fictitious classmates, and in which way?, and (2) To what extent are primary school teachers' endorsement of MC and BSM linked to student‐expected differences in their interpersonal behavior toward children with different migration and socioeconomic backgrounds?

1.1. Student‐Expected Interpersonal Teacher Behavior

Teachers' interpersonal behaviors are a central component of the relationships with their students, and those relationships are often described in terms of closeness and conflict (Pianta et al. 2003). Whereas closeness involves warm and affectionate teacher (and student) behaviors, conflict refers to hostile and negative interactions (Pianta 1999; Pianta et al. 2003). Research has shown that closeness and conflict contribute uniquely to children's academic and social development, with positive effects of the former and negative effects of the latter ² (for a review, see Baker 2006). Therefore, the current study examines both positive and negative interpersonal teacher behavior.

1.1.1. Cultural Synchrony Hypothesis (CSH)

Generally, teachers tend to develop less favorable interpersonal relationships with students with a migration background and low‐SES students. ³ The CSH (Monroe and Obidah 2004) suggests that less favorable interpersonal teacher behavior might be a result of cultural incongruence between the structure of education (which is based on the ethnic majority cultural norms of behavior), and the cultural norms that ethnic minority students learned at home (Blake et al. 2016; Monroe and Obidah 2004). For example, the Dutch culture tends to be more individualistic, which yields different parenting styles compared to more collectivistic ethnic minority cultures (Prevoo and Tamis‐LeMonda 2017). This could result in a mutual lack of understanding and stereotyping between teachers and students with a migration background (although having an ethnic minority – or migration background is not identical, they largely overlap) – which has been well‐documented by previous research into, for example, ethnic biases in teacher expectations (Glock and Kleen 2023; Kennedy et al. 2023). On the other hand, when there is a cultural match, teachers and students can capitalize on shared backgrounds, which fosters positive relationships (Blake et al. 2016).

Teachers indeed have been found to experience difficulties in their communication with students with an ethnic minority background and have negative interpretations of their behavior, which is the case for Dutch primary school teachers as well (Thijs et al. 2012; Weiner 2016). It is important to note here that a large majority of primary school teachers in the Netherlands are female, and have a nonmigrant, native background (i.e., 8 and 9 out of 10 teachers respectively, which has been stable over 15 years and is thus expected to be accurate to date; Statistics Netherlands 2018). Primary school teachers tend to classify disruptive behaviors of students with an ethnic minority or migration background more negatively (even when they show identical behaviors, Ferguson 2001) and perform more disciplinary actions against them than against nonmigrant students (Inan‐Kaya and Rubie‐Davies 2022), even after (experimentally) controlling for their achievement and behavior (Glock 2016; Rocque and Paternoster 2011). Weiner (2016) found that in a Dutch primary school, Turkish‐ and Surinamese‐origin children were most likely to be subjected to negative classroom practices (e.g., call outs, discouragement, and silencing). Research into student reports echoes these findings: Ethnic minority students experience more teacher dominance and less proximity compared to their nonmigrant classmates (for a review, see Den Brok and Levy 2005), which also holds for the Dutch context (e.g., Thijs and Fleischmann 2015; Van Vemde et al. 2021).

A related line of research focuses on the ethnic (in)congruence between a teacher and their students, arguing that a mismatch between the ethnic background of the teacher themselves and a student could lead to less positive and more negative interpersonal behaviors (Saft and Pianta 2001; Thijs et al. 2012). On the other hand, when teachers have an ethnic migration background themselves, they understand the experiences of their students with a migration background better, which could be beneficial for teacher‐student relationships (Kennedy et al. 2024). As mentioned, many teachers in the Netherlands have a nonmigrant, native background, hence ethnic incongruence may contribute to less positive interpersonal teacher behavior toward children with a migration background.

1.1.2. Social Class‐Academic Contexts Mismatch (SCACM) Model

The CSH might also apply to a socioeconomic cultural (dis)synchronization leading to differences in teachers' interpersonal behavior toward students with different socioeconomic backgrounds. A recent study (Goudeau et al. 2024) developed the SCACM model, which theorizes about the mismatches between the home experiences of low‐SES students, and the cultural ideas and practices that exist in schools. These mismatches occur because education is not SES‐neutral, but structured in a way that it benefits high‐SES students (Bourdieu and Passeron 1990).

According to the SCACM model, one of the mismatches is the cultural capital mismatch. Following Bourdieu (1977, 1986) cultural capital exists in three forms: The objectified (possessions of cultural goods), institutionalized (educational attainment), and embodied form (attitudes, preferences, habitus). Depending on someone's social class (i.e., an individual's or group's relative position in an economic‐social‐cultural hierarchy, which is often determined by someone's SES; Weber 1922), people tend to have different cultural norms of behavior. For example, parents from higher‐SES backgrounds are more likely to instill an interest in books and reading (e.g., Lahire 2019, which is why “reading” is conceptualized as a high‐SES hobby in the current study), and tend to discourage spending too much time on electronic devices for leisure activities such as watching series or playing games (Gebremariam et al. 2020) – which corresponds to the cultural norms in primary schools. Thus, there might also be a cultural incongruence between the structure of education (based on cultural norms of behavior of parents with a higher‐SES background), and the cultural norms students with a lower‐SES background learn at home. Consequentially, schools tend to match and value the behavioral tendencies of students with a higher SES, yet mismatch and devalue those with a lower SES (Goudeau et al. 2024). This could lead to misunderstandings and less favorable interpersonal teacher behavior toward students with a lower SES. Upper primary school students with a low‐SES background indeed experience less close and more conflictual teacher‐student relationships (Zee and Koomen 2017), and studies on primary school teachers' reports echo these findings (e.g., Bloem et al. 2024; Rudasill et al. 2010, 2023). Overall, it is anticipated that children expect teachers to behave more positively and less negatively toward native Dutch children and children with a high SES (high income, high education, reading as a favorite hobby), compared to children with a migration background and children with a low SES.

1.2. Teachers' Multiculturalism (MC)

MC entails beliefs about the treatment of cultural diversity in a multicultural society (Van Oudenhoven et al. 1998). A high endorsement of MC means that people believe that all cultural and ethnic groups should be appreciated and treated with respect, and as equals. Following the CSH, cultural misunderstandings could be one reason why teachers might show more negative interpersonal behavior toward students with a migration background because they contribute to negative perceptions of those students' behavior (e.g., Thijs et al. 2012; Weiner 2016). To process and react to information from the classroom context, teachers use their mental schemas about the world (Pickens 2005). These mental schemas, however, are informed by (cultural) assumptions and beliefs, which might distort perceptions of student behavior and be a reason why teachers (to varying degrees) differentiate in their interpersonal behaviors.

MC might be such a belief that influences teachers' interpersonal behavior toward students with a migration background. A lower endorsement of MC has been found to be related to less positive out‐group attitudes or more in‐group preference in majority group members (Verkuyten 2005). For teachers, more positive attitudes toward other cultural groups might directly translate into positive behaviors but also reinforce their willingness to understand differences based on cultural backgrounds, which could consequentially diminish cultural misunderstandings. A Dutch vignette study (Abacioglu et al. 2021) found that the higher primary school teachers' positive multicultural attitudes are, the more likely they are to use tolerant intervention strategies toward native Dutch and Moroccan‐origin students with disruptive behavior. This suggests that teachers who appreciate ethnic and cultural diversity more show more understanding of individual differences between their students and their behaviors and are thus overall more tolerant of disruptive behavior (Abacioglu et al. 2021). Further, Dutch primary school teachers who, according to their students, seldom spoke about different cultures deserving equal treatment, reported lower levels of closeness with their Moroccan‐origin students compared to native Dutch students, whereas this was not the case for teachers who often talked about this (Thijs et al. 2012).

Looking at it from another angle, but still in line with the CSH, Abacioglu et al. (2020) found that teachers' multicultural attitudes were positively related to the self‐reported frequency with which they engaged in culturally responsive teaching. Thus, teachers with a higher endorsement of MC better connect the personal experiences and frames of reference of students with a migration background to the learning environment (see Gay 2010), thereby decreasing the cultural mismatch between the personal context of students with a migration background and the school context.

1.3. Teachers' Belief in School Meritocracy (BSM)

Meritocracy is a societal belief arguing that status in society is based on merit (Mijs 2016), with merit being typically defined as the combination of ability (intelligence) and effort (Young 1958). This way, the belief in such a societal mechanism justifies existing inequalities (Son Hing et al. 2011). If teachers believe in school meritocracy, they believe that academic success reflects the hard work and ability of students (and, conversely, that less academic success reflects less ability and effort), and they may thus perceive achievement inequality as a fair outcome of the meritocratic process (Darnon et al. 2018). Academic inequalities associated with, for example, students' ethnic background or SES are typically ignored (McNamee and Miller 2004).

However, in practice, academic success does not always reflect hard work and ability; differences in educational attainment are observed even for students with similar abilities. For example, teachers expect students with an ethnic minority background to show lower achievement compared to their majority peers, even when their actual achievement is (experimentally manipulated to be) similar (Glock and Kleen 2023) – which leads to ethnic minority students having lower achievement than their majority peers (D'haeseleer et al. 2021). In the same vein, students with a lower SES tend to receive lower grades and tracking recommendations than students with a higher SES, with a similar level of performance (e.g., Batruch et al. 2023) – also in the Netherlands (Timmermans et al. 2018). One could conclude from this that schools (at least partially) fail to fulfill their meritocratic ideal.

Paradoxically, schools may even contribute to the reproduction of social inequality by claiming to be meritocratic (Van De Werfhorst and Mijs 2010). That is, a BSM could lead to negative attitudes toward those with less educational success, as they are perceived as having an inherent lack of abilities (Reay 2020). This prejudice toward people with a lower education is to date often still seen as an “acceptable” prejudice (Kuppens et al. 2018). Darnon et al. (2023) found that teachers who believe in school meritocracy used practices in their classroom that better match the home experiences of students with a high SES and thus undermine the success of students with a low SES (e.g., encouraging competition and social comparison, valuing the best students). Conversely, teachers with a lower BSM used more cooperative practices in the classroom, which better fit students with a low SES. Although teachers with a high BSM do not intentionally use these practices to discriminate against students with a low SES, the increased cultural mismatch could still lead to less positive interpersonal teacher behavior toward them.

But the consequences of meritocracy go beyond this; a greater belief in meritocratic worldviews is linked to more racist attitudes (Glover 1994), negative attitudes toward poverty (Hoyt et al. 2021), and denial of white privilege amongst White individuals (Knowles and Lowery 2012). This implies that when teachers have a higher BSM, they might have fewer positive attitudes toward students with a low‐SES as well as migration background – because presumably when they show a lower educational attainment, this is due to their own lack of abilities and effort. This reinforces cultural misunderstandings, according to the CSH and SCACM model. For example, teachers tend to have more negative perceptions of the participation in schoolwork of students with an ethnic minority background (i.e., paying attention, following directions; Pyne and Musto 2023).

Meritocratic beliefs also influence how people perceive inequalities. People who believe that success is (and should be) achieved through internal causes (i.e., people with a strong belief in meritocracy) have been found to perceive less economic inequality and accept it more (Filippi 2024). This suggests that teachers who believe that academic success is merely achieved through internal causes (vs. external factors as well) are more likely to perceive less educational inequality and accept it more when, for example, students with a lower SES tend to have lower academic outcomes. Indeed, teachers with a greater BSM have been found to be less supportive of practices and initiatives aimed at reducing ethnic‐ and SES‐based inequities in school (Darnon et al. 2018; Doyle et al. 2023). For teachers' interpersonal behavior toward students with different migration and socioeconomic backgrounds, this means that we expect teachers with a higher BSM to invest less effort in providing equal treatment to different students.

1.4. Conjoint Experiment (CE)

The current study used a CE to measure student‐expected differences in interpersonal teacher behavior toward classmates with different migration and socioeconomic backgrounds. A CE is a variant of factorial designs, which works similarly to providing one vignette but the amount of text is minimalized to reduce cognitive load, which allows presenting more sets of comparisons (Bansak et al. 2021). We adopted the most commonly used “forced‐choice” design, which means that students had to choose between two profiles (peers) whose attributes (migration and socioeconomic characteristics) were randomly manipulated. For each pair, students were asked to choose toward which peer their teacher would behave more positively or negatively. We included the following child characteristics of the fictitious classmates in the CE: Migration background, parental income, parental education, and the child's favorite hobby. Parental income and education were included as separate SES characteristics, as they have supplementary, unique effects on children's school outcomes (Klein et al. 2020). Also, students themselves use them separately to form their group evaluations, and it is not self‐evident that they know these two indicators are related to each other (Boer et al. 2024). Further, we included the child's favorite hobby as this represents cultural capital which plays a substantial role in education, albeit more hidden and implicit (Bourdieu and Passeron 1990).

The CE allowed us to identify and compare the independent and relative effects of each child characteristic (keeping all other characteristics constant), and to compare the magnitude of the effects of the different child characteristics. Gender was added as a fifth child characteristic in the CE because of the consistent and extensive evidence for its impact on interpersonal teacher behavior (i.e., teachers have fewer positive relationships with boys compared to girls, e.g., Rudasill et al. 2023). By including gender in the CE, we were able to control for this when examining the effects of students' ethnic and socioeconomic characteristics. Moreover, we were able to use it as a benchmark by descriptively comparing the magnitude of the effect of gender to the effects of ethnic and socioeconomic characteristics – to see whether migration and socioeconomic background were more or less influential than the established influence of gender.

2. Present Study

Several studies have shown that teachers might interpersonally treat students with a migration background and low‐SES students disadvantageously, leading to inequalities in educational outcomes. The present study aimed to investigate the relative and isolated influence of children's migration and socioeconomic background on student‐expected interpersonal teacher behavior, and whether these differences were linked to teacher beliefs.

Based on the CSH (Monroe and Obidah 2004), SCACM model (Goudeau et al. 2024) and empirical research, it was anticipated that (Hypothesis 1) students expect teachers to behave more positively and less negatively toward native Dutch children and children with (one or more) high SES indicators (high income, high education, reading as favorite hobby), compared to children with a migration background and children with low SES indicators and while controlling for all other characteristics. It was explored which specific child characteristics were relatively more relevant for student‐expected interpersonal teacher behavior. Further, it was anticipated that (Hypothesis 2) the relevance of students' migration background is lower for the expected interpersonal behavior of teachers with a higher (vs. lower) endorsement of MC. Or stated differently: Teachers with a higher endorsement of MC differ less in their interpersonal behavior toward children with a native Dutch versus a migration background (while keeping all other characteristics constant). Finally, it was anticipated that the relevance of students' migration background (Hypothesis 3a) and SES (Hypothesis 3b) is lower for the expected interpersonal behavior of teachers with a lower (vs. higher) BSM. The current study was conducted in the Netherlands, a country where ethnic inequalities are strongly tied to migration backgrounds. The majority of children in the Netherlands with a migration background are Dutch nationals themselves, but have (grand)parents who are born in another country (Statistics Netherlands 2016). These children still face lower educational opportunities (Denessen et al. 2022). Moreover, people with a lower education tend to have a lower social standing as well (Kuppens et al. 2018), and there is currently a lot of attention for these status differences based on educational level in our society (e.g., there are debates about how to refer to low‐ and high‐educated people).

3. Materials and Methods

3.1. Participants and Procedure

This study is part of a longitudinal research project (three waves), approved by the Ethics Committee of the Faculty of Social and Behavioral Sciences of Utrecht University. We preregistered this study at: https://osf.io/ym5qu. In this project, 68 classes from primary schools in the Netherlands participated – all students were in Grades 4–6, except a few students in Grade 3 (n = 29). The teacher sample consisted of 68 teachers (mean age = 41.55, SD = 11.43, 70.6% female, mean teaching experience = 16.31 years, range: 0–44.). Of these teachers, 92.4% had a native Dutch background and 51.5% had at least one high‐educated parent (i.e., who obtained a university [of applied science] or doctorate degree). We considered students and teachers to have a native Dutch background when they themselves and both of their parents were born in the Netherlands, which is in line with the guidelines of Statistics Netherlands (2016). Two teachers dropped out of the research project after wave 1, but their replacement teachers participated in waves 2 and 3. In both cases, data from the replacement teacher were used in this study. For one teacher, this means we used their student data of waves 2 and 3 (and dropped the data of wave 1), for the other teacher, we only used their student data of wave 3 as she had only taught the class for 2 weeks at wave 2.

The student sample consisted of 1186 students (mean age at wave 1 = 10.18, SD = 1.01, 47.8% girls, 62.7% Dutch background). Of the 32.1% students with a migration background, 34.7% were first‐generation immigrants, mostly born in Syria and Poland. The parents of the second‐generation immigrant students came primarily from Turkey, Morocco, Syria, and Poland. Further, 48.6% of the children had at least one high‐educated parent, 26.4% had no high‐educated parents, and 25.0% had no information provided on parental education.

Classes were recruited via a systematic procedure; 1700 primary school principals throughout the Netherlands were contacted with the question of participating in a research project about teacher‐student and peers' interactions, reinforcing equal educational opportunities. We aimed for schools with a range of heterogeneity in terms of students' ethnic and socioeconomic backgrounds according to national statistics, but we oversampled heterogeneous schools (see Supporting Information S1: Appendix S1A for the full procedure). Active informed consent was obtained from teachers, parents/caregivers, and students. On average, 71.84% of the children in a classroom had parental consent and participated in the research project (range: 36.36%–100%).

Classes were visited three times throughout the school year 2023–2024 (during fall, winter, and spring). Before class visits, parents or caregivers filled out their educational attainment to capture family SES. During class visits, a researcher was present to answer questions. Both students and teachers completed an online questionnaire using Qualtrics. For students, this took approximately 30–45 min. The current study used CE data (see below) from all three waves. For teachers, filling in the survey took approximately 20–30 min. The current study only used teacher data from the third wave (April–June 2024), as in this wave both teacher beliefs were asked for, and we assumed these kinds of beliefs to be stable over time for adults (see, e.g., Keskintürk 2021) ‐ especially considering the rather brief timeframe. As this study is part of a larger project, both students and teachers also answered additional questions. As an incentive, teachers were promised a class‐report on their results, and received a gift voucher (for a variety of (web)shops) both for themselves and their class.

3.2. Measures

3.2.1. Student‐Expected Teachers' Interpersonal Behavior

We used a CE (see Figure 1) to measure whether students expected differences in interpersonal teacher behavior toward hypothetical classmates with different ethnic and socioeconomic characteristics. In this CE, students had to choose one of two children. The following child characteristics were manipulated in the CE: Gender (boy vs. girl), migration background (native Dutch vs. migration), parental education (low‐ vs. high‐educated), parental income (poor vs. rich), and favorite hobby (reading vs. playing games on the iPad). We randomized the order in which the child characteristics appeared in the CE between participants, but not within participants, to increase readability of the CE tasks.

Figure 1.

Set‐up of the conjoint experiment. ⁶

During each wave, students received six CE‐tasks. Three times they answered the question “Which child do you think your teacher gets angry at more often?” and three times “Which child do you think your teacher is nicer to?”. As students filled out the questionnaire three times (three waves), each child completed a total of 18 CE‐tasks (9 for both questions). It was not explicitly mentioned in the questionnaire that students could skip a CE‐task (to avoid unserious answering behavior, as this task required more reading than the other questions). Still, students were able to skip tasks by pressing a right‐pointing arrow after confirming this in a pop‐up screen.

All child characteristics in the CE had two options (i.e., levels, e.g., boy‐girl), to create the most simplified version to have more stable estimates of effects. Hence, we included migration background as native Dutch versus migration background instead of specifying different origin countries. We varied the parents' country of birth instead of the child's itself, as most primary school students in the Netherlands with a migration background are second‐generation immigrants. All levels were manipulated independently and at random.

3.2.2. Teachers' MC

To measure teachers' MC, we used a Dutch adaptation (Arends‐Tóth and Vijver 2003) of the Multicultural Ideology Scale of Berry and Kalin (1995). This scale consisted of six items, answered on a 5‐point Likert scale ranging from 1 (completely disagree) to 5 (completely agree). An example item is “Dutch people without a migration background should accept that the Netherlands consists of various cultural and religious groups.” Three items (e.g., immigrants should give up their own culture as much as possible and adopt the Dutch culture) were reverse‐coded, as a high score on those items indicated low MC. Reliability analyses showed an acceptable internal consistency (α = 0.65). Principal Component Analysis using varimax rotation (KMO = 0.69, Bartlett's test of sphericity χ ² (15) = 63.85) yielded support for one factor when inspecting the scree plot (eigenvalue = 2.34, variance explained = 38.94%).

3.2.3. Teachers' BSM

To measure teachers' BSM, we used the Belief in School Meritocracy Scale by Wiederkehr et al. (2015), which was directly translated into Dutch. This scale consisted of eight items, answered on a 5‐point Likert scale ranging from 1 (completely disagree) to 5 (completely agree). An example item is “To succeed at school, one only has to work hard.” Reliability analyses showed a somewhat unsatisfactory internal consistency (α = 0.52), which could be improved by deleting two items (i.e., “Willingness is not always enough to succeed at school” and “students who obtain bad grades at school, are those who did not work hard enough”) to α = 0.57.

We conducted a Principal Component Analysis using varimax rotation (KMO = 0.45, Bartlett's test of sphericity χ ² (15) = 52.01) with the remaining six items. This yielded support for one factor when inspecting the scree plot (eigenvalue = 1.91, variance explained = 31.76%).

3.2.4. Ethnic and Socioeconomic Classroom Composition

As teachers' interpersonal behaviors toward children of different ethnic and socioeconomic backgrounds may also depend on the ethnic and socioeconomic composition of their classroom, we will control for this in the analyses. Students filled out their migration background by indicating their own and their parents' country of birth from a list of provided countries, or by writing down another country. As mentioned, a student was considered native Dutch if they themselves and their parents were born in the Netherlands. We estimated the proportion of Dutch children in a classroom relative to all children, which ranged from 5% to 100%, M = 62.95%. Further, parents were asked to provide their own and – if applicable – their partner's highest obtained level of education. An educational level from primary education, to post‐secondary vocational education is considered low‐educated, and a university of applied sciences to a doctorate is considered high‐educated. The percentage of children with at least one high‐educated parent per class was estimated and ranged from 6.67% to 100%, M = 48.29%.

3.3. Analyses

The current study used a fully randomized CE, which is a full factorial design. The number of attributes (child characteristics, e.g., gender) and levels (the two options of the characteristics, e.g., boy–girl) in our study design yields 496 unique profiles of children. Randomization of the selected levels ensures that the probability that each level of an attribute appears in each profile is orthogonal to that of all other attributes, which allowed us to properly estimate the effect of any given level. All analyses were conducted in RStudio (version 4.3.1).

3.3.1. Analyses Plan

3.3.1.1. Marginal Means

To test Hypothesis 1 (about the main effects of children's ethnicity and the SES characteristics), we conducted analyses with only the students' CE data; all the separate CE‐tasks of the full sample were aggregated together. Logistic regression models were used, as the outcome variable was binary (either child 1 or child 2 could be selected). We calculated Marginal Means ⁴ for each level of the child characteristics that occur in the CE‐tasks. Marginal Means are descriptive average outcome values of the two levels in a child characteristic, averaged across all possible combinations of the other child characteristics' values (while keeping all other child characteristics constant). With this, we could calculate the percentage point difference between the probability of a child being chosen when they had, for example, a native Dutch versus a migration background. To illustrate, if migration background had no influence on student‐expected interpersonal teacher behavior, we expect the averages for each level (i.e., Dutch and migration background) to be (close to) 50% which would mean that both levels have an equal probability of being chosen in the CE. Thus, we aimed to model the chance that a child was chosen, based on the level of each of their characteristics.

There was a multilevel structure in the data, as students' individual profile choices were nested inside respondents, and respondents were nested within classes. If a student completed all CE‐tasks, there were 18 tasks with displayed child characteristics (Level 1) nested within this student (Level 2), and this student was, together with classmates, nested within a classroom (Level 3). Following Heiss (2023), we estimated the Marginal Means Bayesianly, in which we compared the posterior distributions of Marginal Means for all characteristic levels. To account for the multilevel structure of the data, we allowed respondent‐and class‐specific offsets – and thus created random intercepts between persons and classes. We specified weakly informative priors given the structure of the data and our lack of previous evidence to build on: a normal distribution prior for the child characteristics levels with a mean of 0 and a standard deviation of 1 for all of them, and an exponential prior, with a lambda of 1 for the between‐respondent and between‐classes variability.

To test Hypothesis 1, the credible interval (CI) of the difference in Marginal Means between the two levels of a child characteristic (i.e., native Dutch vs. migration background) was inspected to see which specific position yielded more positive and less negative student‐expected interpersonal teacher behavior. A CI means that given the priors and observed data, there is a 95% probability that the true difference lies within this range (Hespanhol et al. 2019). Further, we descriptively inspected which characteristics were more relevant.

3.3.1.2. Marginal Means and Teacher Variables

To test Hypotheses 2 and 3, the teacher's belief variables BSM and MC were included in the analyses as continuous variables. First, we inspected these variables and checked for assumptions. There were no missing data or outliers in these variables. Again, a Bayesian approach was used with similar priors by running a separate model with interaction terms added. More specifically, we added the interaction terms MC with migration background and BSM with migration background, income, education, and favorite hobby. To test Hypotheses 2 and 3, posterior distributions of the interaction terms were inspected to see how teacher beliefs affected the influence of the child characteristic in the CE.

3.3.2. Missing Data

For both the questions about negative (CE‐angry tasks) and positive (CE‐nice tasks) teacher behavior, there could be a maximum of 10.551 chosen answers in total (1186 students × 9 CE‐tasks minus 123 tasks for the two teachers who dropped out and thus had no CE‐data on wave 1 [9 students] or 1 and 2 [15 students]). For both the CE‐angry and CE‐nice tasks, there were 597 cases (5.7%) with missing values due to student absence during a wave, resulting in missing values on the corresponding six CE tasks. Of the remaining 9954 cases, for the CE‐angry tasks, in 1836 cases (17.4%), children did not choose one of the two fictitious children, but skipped the CE task. Thus, for the CE‐angry tasks, there was a total of 8118 completed tasks. For the CE‐nice tasks, in 1725 cases (16.3%), children skipped the question, resulting in a total number of 8244 completed tasks.

Regarding the sample size per class for the CE‐angry‐tasks, this ranged from 37 to 205 cases. The percentage of CE‐tasks that were skipped by students ranged from 2.02% to 53.97%, which indicates there was great variation between classes. For the CE‐nice‐tasks, the sample size per class ranged from 35 to 212 cases. The percentage of tasks that were skipped ranged from 2.02% to 53.17%, again indicating great variation between classes. All missing responses were automatically excluded, and all the available information was used. This is because one cannot assume that the missing data was missing completely at random (in fact, it could reasonably be not at random since students deliberately chose to skip this task – possibly for substantive reasons), hence data imputation was not an option. It was also not possible to treat ‘skipping the task’ as a separate outcome within the CE, as we then should have explicitly mentioned that this was an option. Thus, excluding missing responses is the most reasonable way of dealing with the data, which does not cause any power issues.

4. Results

4.1. Marginal Means Child Characteristics

4.1.1. Negative Teacher Behavior

Figure 2 shows the Marginal Means of the probabilities for all levels of the child characteristics that appeared in the CE‐tasks – for the questions about the teacher acting angrily. First, it shows that, when keeping all other child characteristics constant, children expected their teachers to act more angrily toward classmates if that classmate is a boy (as opposed to a girl), which is as anticipated. That is, boys were 8.30% more likely to be chosen than girls (i.e., 45.85% for girls and 54.14% for boys). This median posterior difference of 8.30% has a 95% CI of 6.86–9.74. Thus, there is a 95% probability that the true difference of teachers' differentiation as perceived by students lies within this range, meaning the difference found in favor of girls is robust.

Figure 2.

Marginal means of probabilities for all student characteristics levels – Angry.

Second, contrary to Hypothesis 1, the classmate's migration background was not associated with student‐expected angry interpersonal teacher behavior (95% CI = −0.01 to 0.02). So, classmates with a native Dutch and migration background were equally likely to be chosen when asked toward which classmate the teacher would act more angrily. Third, in line with Hypothesis 1, children expected their teachers to act more angrily toward classmates if that classmate had low‐educated parents (as opposed to high‐educated). These classmates were 9.43% more likely to be chosen (45.24% for children with high‐educated parents and 54.67% for low‐educated parents; 95% CI = 7.99–10.87).

Fourth, unexpectedly, we found that children expected their teachers to act more angrily toward a classmate with rich parents (as opposed to poor). These classmates were 17.64% more likely to be chosen (41.16% for poor and 58.81% for rich; 95% CI = 16.18–19.07). Finally, as expected, children expected their teachers to act more angrily toward classmates if that classmate had playing games on the iPad as a favorite hobby (as opposed to reading). This child characteristic had relatively the largest association; classmates who played games on the iPad as a favorite hobby were 28.74% more likely to be chosen (35.65% for reading and 64.40% for playing games; 95% CI = 27.27–30.17).

4.1.2. Positive Teacher Behavior

Figure 3 shows the Marginal Means of the probabilities for all levels of the child characteristics that appeared in the CE‐tasks – for the questions about the teacher acting nicely. Overall, the results showed a similar pattern as for the question about the teacher acting angrily, but for some child characteristics, the magnitudes of the effects were smaller.

Figure 3.

Marginal means of probabilities for all student characteristics levels – Nice.

First, we found as anticipated that, when keeping all other characteristics constant, girls were 3.40% more likely to be chosen than boys when asked which classmate the teacher would act more nicely toward (48.30% for boys and 51.70% for girls; 95% CI = 1.95–4.88). Second, the classmate's migration background was again not associated with student‐expected nice interpersonal teacher behavior (95% CI = −0.005 to 0.02). Third, in line with Hypothesis 1, classmates with high‐educated parents were 10.16% more likely to be chosen than classmates with low‐educated parents (44.89% for low‐educated and 55.06% for high‐educated; 95% CI = 8.71–11.63). Fourth, contrary to Hypothesis 1, classmates with poor parents were 9.58% more likely to be chosen than classmates with rich parents (45.22% for rich and 54.79% for poor; 95% CI = 8.12 – 11.01). Finally, as expected, classmates with reading as a favorite hobby were 25.76% more likely to be chosen than classmates with playing games on the iPad as a favorite hobby (37.08% for playing games and 62.84% for reading; 95% CI = 24.34%–27.19%). Again, this child characteristic has relatively the largest association.

4.1.3. Variation Between Classes

To investigate whether there was variation between classes (and thus teachers) in children's expectations of their teachers' interpersonal behavior, we ran a random slopes model. With this, we calculated conditional ICCs, which capture the variability explained by both the intercept and random slopes of the different classes. The ICC of the random slopes model indicated that, respectively, 2.21% and 2.62% of the variance in student‐expected teachers' negative and positive behavior was explained at the class‐level. Thus, there were very little systematic differences between teachers.

4.2. Teacher Beliefs

At this point, teachers' MC (M = 3.68, SD = 0.82) and BSM (M = 3.16, SD = 0.82, SD = 0.91) were added to the model. These two teacher beliefs were significantly and negatively correlated r (68) = −0.29, p = 0.017. To test whether teachers' MC and BSM were linked to student‐expected differences in teacher behavior toward different children, we looked at the posterior distribution of the interaction effects between the teacher beliefs and relevant child characteristics. We specifically looked at the following interaction effects: Migration background x MC, Migration background x BSM, Income x BSM, Educational level x BSM, and Favorite hobby x BSM. The interaction effects all showed that teacher's MC and BSM were not linked to any of the student‐expected differences in teacher behavior based on the child characteristics in the CE. For the question about the teacher acting angrily all interaction term beta's were between 0.00 and 0.05 (see Table 1 in Supporting Information S2: Appendix S1B for all specific beta's), with all 95% CIs ranging from at least −1.11 to 1.07, indicating a lack of support for the presence of interaction effects, contrary to Hypotheses 2 and 3. For the question about the teacher acting nicely, all interaction term beta's were between 0.00 and 0.06 (see Table 2 in Supporting Information S2: Appendix S1B for all specific beta's), with all CIs ranging from at least −1.11 to 1.11 indicating again a lack of support.

4.3. Robustness Checks and Exploratory Analysis

First, we checked whether our conclusions hold when testing the models without the two classes with less student CE data due to teacher drop‐out, to ensure that the lower number of cases did not influence the results. When leaving those two classes out, our conclusions remained the same. Second, we checked whether our conclusions hold when controlling for the ethnic and socioeconomic classroom composition. When we added the classroom percentage of native Dutch students and the percentage of students with at least one high‐educated parent as interaction effects with the child characteristics to the model, our conclusions remain the same – both for the question about the teacher acting angrily and nicely.

As mentioned, a relatively large number of CE‐tasks were skipped by students, and those had to be treated as missing in our analyses. However, there was great variation between classes in the percentage of skipped CE‐tasks. One possible explanation for this – based on the researchers' impressions while administering the questionnaire – is that students were not willing to choose a child, because they felt their teacher did simply not differentiate based on (some of) the child characteristics involved. Hence, we exploratively examined whether teachers' BSM and MC were linked to the percentage of skipped CE‐tasks. Linear regressions showed no significant links between this, both for the skipped CE‐tasks about the teacher acting angrily (β_mc = 0.095, p = 0.474, β_bsm = 0.18, p = 0.174), and about the teacher acting nicely (β_mc = 0.09, p = 0.571, β_bsm = 0.17, p = 0.195).

5. Discussion

There is ample research that aims to better understand ethnic educational inequalities with a focus on teachers' differential interpersonal behavior (e.g., Denessen et al. 2022). Yet, this often neglects the role that diminishing differences based on socioeconomic backgrounds can play in mitigating ethnic inequalities – as ethnic background and SES are highly intertwined. The current preregistered study aimed to address this study gap by simultaneously investigating the separate, isolated effects of various SES indicators and migration background of fictitious children, for student‐expected differences in teachers' interpersonal behavior toward these children. Further, we studied whether these differences were linked to teacher beliefs. Our findings for positive and negative teacher behavior mirrored each other: As anticipated, students expected their teachers to behave more positively and less negatively toward children with high‐educated parents, or with reading as a favorite hobby. In contrast to our expectations, they also expected teachers to behave more positively and less negatively toward children with poor parents, and students' migration background had no influence. Also, teacher beliefs about MC and meritocracy were not linked to student‐expected interpersonal teacher behavior, and there was only little variation between teachers in these beliefs.

5.1. SES Indicators

It was hypothesized (Hypothesis 1) that children would expect their teacher to behave more positively and less negatively toward children with high‐SES indicators (i.e., high parental income and education, and reading as a favorite hobby). Our findings are partly in line with this; student‐expected interpersonal teacher behavior was in favor of children with high‐educated parents or with reading as a favorite hobby (keeping all other characteristics constant). Yet, for parental income, our results were in favor of children with poor parents. Compared to gender, which showed the expected effects (more positivity and less negativity toward girls), all three SES indicators had a larger influence on student‐expected interpersonal teacher behavior. Given the fact that migration background had no influence on this, the present study shows that, according to students, teachers mainly differentiated in their interpersonal behavior based on the socioeconomic characteristics of children. Within those, the favorite hobby clearly had the largest influence. The influences of parental income and education were more similar in magnitude, but for negative teacher behavior, the influence of parental income was larger.

Our results for parental education (but not parental income) were in line with the SCACM model, which posits a mismatch between the home experiences of low‐SES students and the cultural ideas and practices that exist in schools (Goudeau et al. 2024). Apparently, this mismatch is due to the educational level rather than the income of their parents and the different cultural norms associated with this (i.e., cultural capital; Bourdieu 1977, 1997). Moreover, children's favorite hobby (reading vs. playing games on the iPad) was, from all characteristics in the CE, most strongly related to teachers' expected interpersonal behavior – also in line with the SCACM. Compared to parental income and education, which are direct indicators of family SES, a favorite hobby is more an expression of family SES (the embodied form of cultural capital). This dimension of capital is more directly related to behavior, and thus more directly related to behavioral tendencies within the classroom that are (de)valued by the teacher. Students probably felt that teachers value the behavior of children with reading as a favorite hobby more. As parents' SES has a large influence on, for example, students' interest in reading (e.g., Lahire 2019), this could be detrimental for the school outcomes of students with a lower SES.

Although in a different direction than expected, parental income still had a large effect on student‐expected interpersonal teacher behavior. Students' answers were highly in favor of poor children. On the one hand, this might be due to students' stereotypical ideas about poor versus rich people – which they then thought to cause more positive and less negative teacher behaviors. Previous research has shown that when children have negative attitudes toward rich peers, this is because they think they are mean, arrogant, and brag about their possessions (Boer et al. 2024; Elenbaas et al. 2022). At the same time, they tend to have positive attitudes toward poor peers out of empathy (e.g., they want to befriend them because they feel sorry for the child being poor; Boer et al. 2024). If students think that rich peers are mean and arrogant, and they feel sorry for poor peers, they might consequentially expect that teachers act more negatively toward rich children, and more positively toward poor peers.

On the other hand, it might be that teachers actually actively try to interpersonally behave more positively toward poor students, because they believe those students especially need this, given their home situation (students with a lower‐SES background have indeed been found to benefit more from high‐quality teacher‐student relationships; see for a meta‐analysis Roorda et al. 2011). To date, most research into the role of students' SES for teachers' interpersonal behavior conceptualized SES as parental education – and then found that teachers tend to have lower‐quality relationships with students with a lower SES (which is in line with our findings for parental education). However, according to students, this might work differently for parental income. In the current study, we focused on the notion of inequality by examining whether teachers, according to their students, negatively differentiate in their interpersonal behavior toward students with a migration or low‐SES background, as previous research has found that teachers tend to do so, which is undesirable for equal opportunities. However, following the notion of inequity, it could also be that teachers positively differentiate in their interpersonal behavior toward students with a migration or low‐SES background (i.e., act nicer and less angry toward these students) because they are aware that these students need this more – which we found in the current study for parental income. In this case, differentiation based on students' SES background might in fact lead to more equal opportunities. It is interesting for future research to investigate this idea of equity within teachers' interpersonal behavior toward students with different backgrounds further.

It is important to note that our findings indicate the independent and isolated effects of each SES indicator, since we controlled for the effects of the other two. These separate independent effects of SES indicators clearly showed these indicators have contradictory connotations; the results are in favor of children with high‐educated parents or with reading as a favorite hobby, but at the same time for poor children as well. Thus, the results are not uniformly in favor of high‐SES children when isolating effects of various SES indicators – which is in line with previous research in other areas that found contradictory effects of different SES indicators (Antonoplis 2023). In reality, SES indicators typically go together (Duncan and Magnuson 2003), and this means that the opposite effects of parental income and education might cancel each other out. This highlights the need for more research taking multiple SES indicators simultaneously into account – to be able to investigate their separate as well as combined effects.

There can be differences between how students expect teachers to behave and how teachers actually behave. However, if students perceive a differentiation in their teachers' behavior that implies negative behavior toward themselves, this could in itself already have detrimental effects for their school outcomes (which has been found for ethnic minority students; McKown and Weinstein 2002). Thus, if students feel that teachers interpersonally behave more positively to students with a higher SES, students with a lower SES might feel less belonging at school – which has also been found to be the case (e.g., Bakchich et al. 2023).

5.2. No Influence of Migration Background

Unexpectedly, it did not matter whether a child had a native Dutch or a migration background for student‐expected positive as well as negative interpersonal teacher behavior toward that child. This is not in line with previous research showing that, according to teacher and student perceptions about their actual teacher–student relationships, the quality of those relationships is better for ethnic majority students (Den Brok and Levy 2005; Inan‐Kaya and Rubie‐Davies 2022).

One reason why children might expect their teachers not to differentiate based on migration background is that teachers may be trying to communicate the idea of ethnic equality more than socioeconomic equality – and students might be aware of this. In teacher education, there is more attention for the equal treatment of all students regardless of their ethnic background than regardless of their socioeconomic background, and celebrating diversity often means ethnic diversity (i.e., multicultural education; Banks and Banks 2019). In line with this, Dutch primary school teachers have been found to deem it important to pay attention to ethnic differences in class (Thijs et al. 2025). This higher attention and awareness among teachers of ethnic compared to socioeconomic inequalities is also reflected in other research. For example, the review study of Batruch et al. (2023) shows that teachers' track recommendations are systematically biased based on students' socioeconomic background, whereas for students' ethnic background, findings are more mixed.

Another reason why we found no effect of migration background might be due to our focus on native Dutch versus immigrant children (with parents born in, respectively, the Netherlands vs. another country). It might be that the category of children with immigrant parents was too broad and thus abstract for our participants. Students might not have had a clear idea about who these children are and thus had no specific ideas about interpersonal teacher behavior toward them. However, there are also studies showing that the distinction native versus migration background in general is meaningful for children (e.g., Thijs 2017). Future research could take more specific migration backgrounds into account to see whether the SES indicators still have a larger influence on student‐expected interpersonal teacher behavior.

5.2.1. Teacher Beliefs

Further, the teachers in our sample mainly had a high endorsement of MC and a moderate BSM. We would expect teachers with these beliefs to differentiate less in their interpersonal behavior based on students' migration background. This might also explain why we found no link between teachers' beliefs and their differentiation based on migration background; perhaps in our sample, there was simply not enough variation between teachers in their beliefs. Yet, the link between teacher beliefs and actual teacher practices is not straightforward (see e.g., Buehl and Beck 2015), so further research is necessary to examine the impact of teachers' endorsement of MC and BSM.

Also, we found only a little variation between classes (and thus teachers) in student‐expected interpersonal teacher behavior. The lack of systematic differences between teachers might suggest that our CE captured children's stereotypes of teachers in general rather than their expectations of their particular teacher. However, there are arguments against this interpretation, as there was a substantial number of children who skipped CE‐tasks, which implies they felt uncomfortable giving this information about their own teacher. Further, in the questionnaire, there were many other questions about their specific teacher, so children were already prompted to think about this. Instead, it is more likely that the lack of between‐teacher effects was due to the relative homogeneity of the teacher sample.

We also found no link between teachers' BSM and student‐expected differentiation of teachers' behaviors based on socioeconomic characteristics, which could again be due to the homogeneity of our teacher sample. Yet, despite our ‘positively biased’, homogeneous teacher sample, we still see that teachers are highly differentiated based on children's socioeconomic characteristics – according to students. Maybe, to find empirical support for our expectation, teachers' BSM should be even lower (i.e., teachers should really disagree with the fact that there is school meritocracy, and it is not enough if their answers reflect a moderate BSM). For future research, it is important to gain more insight into this by investigating student‐expected interpersonal teacher behaviors with a larger, more diverse teacher sample.

5.3. Limitations and Directions for Future Research

There are some limitations worth noting. First, the CE had a forced‐choice design, so there was no third option to choose ‘equally angry/nice toward child 1 and child 2’. This has the potential to bias the results, as children might have thought that their teacher does not necessarily differentiate in their behavior based on these child characteristics, but still chose one child. Hence, our conclusions might be too strong, and if children were provided with the third option, the effects could have been smaller. However, providing a third option could also lead to other issues, such as social desirability and statistical power. Second, all missing values in the data of the CE were excluded from analyses. Missingness could be either due to children being absent at one or two waves or because children skipped a CE task. Although the latter was not explicitly mentioned as an option in the task, children could skip tasks. A substantial number of children did this, and thus, in fact, indicated that their teacher did not differentiate based on these characteristics. Although the percentage of skipped tasks per class was not related to the teacher's beliefs, it is still unfortunate that it was not possible to take these missing values into account in the analyses. Third, the reliability of the teacher‐reported belief in school meritocracy scale was relatively low (α = 0.57), so related results should be interpreted with some caution.

Fourth, students in the current study were prompted to think about their teachers' interpersonal behavior toward children, based on only limited information about those children – while in reality, there is more information available. The information was also presented to the children with only two, rather extreme, positions per characteristic (e.g., poor and rich parents). There are countless variations on these child characteristics, but due to methodological restrictions, we were not able to add more nuance to the CE. Since the social reality is more nuanced than our experiment, with often smaller differences between children, our conclusions might not extend directly to this reality. Yet, this design allowed us to examine the isolated, separate effects of various SES indicators and ethnic background – and compare these effects.

Finally, related to this, our conceptualization of favorite hobby (reading vs. playing games on the iPad) might evoke other associations besides SES‐related ones. Although hobbies are more often used to represent someone's SES (e.g., playing violin and field hockey vs. dancing at the community center; Bloem et al. 2024), the hobbies in the current study could be related to (mal)adaptive classroom behavior, such as paying attention in the case of reading, and getting distracted in the case of playing games on the iPad. Teachers' expected differentiation in their behavior could therefore be student behavior – rather than SES‐based. Future research could use different hobbies to rule out this explanation. Further, future studies could qualitatively examine students' reasoning behind their choices. It would be interesting to know more about why students think that teachers differentiate in their behavior based on children's socioeconomic characteristics.

5.4. Implications for Research and Practice

It is important to consider the results of the current study when researching the underlying mechanisms for ethnic educational inequalities. Research to date often investigates the effects of students' ethnic background while controlling for SES, while our results show that especially socioeconomic characteristics might be a reason why teachers differentiate in their interpersonal behavior toward children. At least, children have the idea that their teachers tend to differentiate based on SES characteristics. Even though in the current study we disentangled and separated students' ethnic and socioeconomic backgrounds, in reality, the two are highly intertwined (Codiroli Mcmaster and Cook 2019). Thus, if a student with a migration background is poor (which is more likely to happen than for students without a migration background, given the fact that people with a migration background tend to have a lower income; OECD 2019; Netherlands Institute for Social Research 2023), this might in fact have positive consequences for interpersonal teacher behavior for this student. At the same time, if a student with a migration background has parents with a lower level of education, this might have detrimental effects. For future research, it is interesting to take both students' migration and SES background into account to gain more insight into its combined effects on interpersonal teacher behavior. Thereby, it is interesting to gain more knowledge into teacher perceptions, and whether they align with student perceptions.

It is important for teachers to be aware of and acknowledge the cultural mismatch between the home experiences of students with a lower‐SES background and the cultural ideas and practices of schools. This way, teachers might already better understand the behaviors of these students (Goudeau et al. 2024). Naturally, it can be difficult for teachers to act upon this awareness, but, following the notion of equity, there are teaching practices that have been found to diminish this cultural mismatch (i.e., resulting in students with a lower SES being more engaged and feeling more seen in the classroom). For example, teachers could use more cooperative teaching styles instead of competitive ones such as encouraging competition and social comparison, since the latter is especially detrimental for the self‐efficacy of students with a lower SES (Jury et al. 2018). Teacher training programs should pay more attention to this socioeconomic‐cultural mismatch, both in terms of raising awareness about this and giving practical tools to diminish this mismatch.

5.5. Conclusion

To date, research into ethnic educational inequalities neglects the practical implication that diminishing socioeconomic inequalities in schools could also further reduce ethnic inequalities – as the two are highly intertwined. Hence, this study investigated children's expectations of teachers' differentiations in their interpersonal behavior toward children with different ethnic as well as socioeconomic characteristics. Its findings show that children expect their teachers to not differentiate in their interpersonal behavior based on students' ethnic background, but they expect them to do so based on students' socioeconomic characteristics. On the one hand, they expect teachers to act more positively toward children with high‐educated parents or who read as a favorite hobby. On the other hand, however, they expect teachers to act more positively toward poor children. Whereas the first finding, as expected, implies that teachers tend to have better‐quality relationships with high‐SES students, the latter finding implies that teachers actively try to have better teacher‐student relationships with poor students – maybe because they think those students need this more. It is important, both for research and practice, to pay more attention to differentiation in interpersonal teacher behavior based on all ethnic and socioeconomic background characteristics – to gain more insight into how we can create equal opportunities in education for all students.

Author Contributions

Iris Boer conceptualized the study, participated in its design and coordination, performed the data collection and statistical analyses, and drafted the first version of the manuscript. Duco Veen performed statistical analyses and provided extensive feedback on the methods and results section of the manuscript. Fenella Fleischmann and Jochem Thijs participated in the conceptualization of the study and its design and provided extensive feedback on each part of the manuscript. All authors read and approved the final manuscript.

Ethics Statement

Approval was obtained from the ethics committee of Utrecht University.

Consent

Teachers provided written informed consent, and for students, written active parental consent was obtained.

Conflicts of Interest

The authors declare no conflicts of interest.

1. Transparent Peer Review

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1002/jcop.70040.

Supporting information

Appendix A.docx.

Appendix B.docx.

Boer, I. , Veen D., Fleischmann F., and Thijs J.. 2025. “Disentangling the Role of Migration Background and SES in Student‐Expected Interpersonal Teacher Behavior.” Journal of Community Psychology 53: 1–16. 10.1002/jcop.70040.

Data Availability Statement

Due to the fact that the data set contains sensitive and personal data of young people, the data set is not published openly available, but is available upon request at the first author. The analyses documents (computer code and output) are openly available in the Open Science Framework at https://osf.io/wvs3t/.