Young Children’s and Teachers’ Perceptions of Affective Teacher-Child Relationships: A Cross-Cultural Comparison Between the Netherlands and China

ABSTRACT

Research Findings: The present study focused on the affective quality of dyadic relationships between teachers and young children in China and the Netherlands. We examined whether there were cross-cultural differences in teachers’ and children’s relationship perceptions, as well as the degree to which teachers and children agreed on their relationship perceptions. The Dutch sample included 234 kindergartners and first-graders (47.7% girls; M_{age =} 5.82 years) and 35 teachers. The Chinese sample consisted of 190 kindergartners (50.5% girls; M_{age =} 5.60 years) and 19 teachers. Both teachers and children reported about their mutual relationships. Multiple group analyses showed that, unexpectedly, Chinese children experienced lower warmth and higher conflict with teachers than Dutch children. In contrast, teachers from China and the Netherlands did not differ in the degree of closeness and conflict they experienced in their relationships with young children. In both countries, teachers and children had significant agreement on conflict but lacked agreement on closeness. Practice or Policy: Thus, it seems important to include child perceptions in cross-cultural comparisons of early teacher-child relationships as well.

Previous research has repeatedly shown that the affective quality of teacher-child relationships (TCRs) is important for children’s behavioral and academic adjustment, such as children’ self-control, externalizing behaviors, school engagement, and academic achievement (see H. Lei et al., 2016, 2018; J. B. Li et al., 2021; Roorda et al., 2011, 2017 for meta-analyses). For instance, in a recent meta-analysis, higher relationship quality with teachers appeared to be associated with better self-control in children (J. B. Li et al., 2021). Research about teacher-child relationships, however, was predominantly conducted in Western countries such as the United States, Germany, and the Netherlands. For example, in the meta-analysis of Roorda et al. (2017), 161 out of the 189 studies (85.2%) were conducted in Western countries. In another meta-analysis (H. Lei et al., 2016), only four out of the 57 studies (7.02%) were based on samples from Eastern countries. The few existing cross-cultural studies found that the quality of teacher-child relationships can differ across Western, individualistic countries and Eastern, collectivistic countries (M. Chen et al., 2019; Pianta et al., 2003; Yang et al., 2013). As such, findings from Western, individualistic countries may not simply generalize to teachers and children in Eastern, collectivistic countries. It thus appears relevant to know more about how teacher-child relationship quality may differ across countries. Investigating cross-cultural differences in teacher-child relationships could have both theoretical and practical implications. From a theoretical perspective, such research can enrich current knowledge about the formation of teacher-child relationships in different contexts, and thus could help to get a more comprehensive picture of the development of teacher-child relationships. Furthermore, the few existing cross-cultural studies mostly focused on teachers and children in upper elementary school and middle school (e.g., M. Chen et al., 2019; Jia et al., 2009; Yang et al., 2013). Not much is known, however, about whether the relationships between teachers and young children differ across countries. As younger children are exposed to cultural values and expectations for a shorter period, it remains unclear whether previously found cross-cultural differences in the relationship between teachers and older children also exist in the early stage of education. Regarding practical implications, it is important to state that most relationship-focused interventions have been developed in Western, individualistic countries (e.g., Vancraeyveldt et al., 2015). Still, there have been increasing attempts and discussions about transferring educational programs and interventions from one country to another (Gabrenya & Glazer, 2022). Looking at teachers’ and young children’s relationship experiences in different countries may provide researchers and educational practitioners with insights to make educational practices and interventions more culture-sensitive.

Furthermore, previous studies in Western countries have shown that children and teachers have different understandings of their mutual relationships (Hughes, 2011; Jellesma et al., 2015; Koomen & Jellesma, 2015). The only two cross-cultural studies looking at the relationship between teachers and young children (i.e., 5-to-8-year-old children), however, were both based on teachers’ relationship perceptions (Acar et al., 2019; Beyazkurk & Kesner, 2005). As such, not much is known about cross-cultural differences in young children’s relationship perceptions. Thus, it is important to investigate whether young children’s perceptions of teacher-child relationships differ across countries. The present study, therefore, focused on both teachers’ and young children’s relationship perceptions and made a cross-cultural comparison between the Netherlands (a Western, individualistic country) and China (an Eastern, collectivistic country). Furthermore, M. Chen et al. (2019) showed in their study about teachers and older children (i.e., upper elementary students) that the agreement between teachers’ and children’s relationship perceptions differed across countries as well. Hence, the second aim of this study was to investigate whether there are also cross-cultural differences in the agreement between teachers’ and young children’s relationship perceptions.

Theoretical Framework

Research about early teacher-child relationships is usually based on attachment theory and the developmental systems model (Bowlby, 1980; Pianta et al., 2003; Verschueren & Koomen, 2012). According to attachment theorists, positive teacher-child relationships provide children with a secure base that helps them to feel confident exploring the environment, as well as a safe haven where children can seek support and comfort from their teacher in stressful times (Pianta et al., 2003). In contrast, negative teacher-child relationships tend to elicit feelings of insecurity in children and, hence, hamper their exploration of the environment. Consequently, children sharing positive relationships with teachers tend to become competent and well-adjusted at school, whereas experiences of negative teacher-child relationships may harm children’s school functioning and well-being (Pianta et al., 2003; Verschueren & Koomen, 2012). As such, previous research based on attachment theory often distinguishes between two relationship dimensions, namely, the degree of closeness and conflict in teacher-child relationships. Closeness reflects relational security and describes the degree of warmth, intimacy, and open communication between teachers and children (Pianta et al., 2003; Verschueren & Koomen, 2012). Conflict links to relational insecurity and reflects the degree of discordance, quarrels, and negativity between teachers and children (Pianta et al., 2003; Verschueren & Koomen, 2012).

A construct that is central in both attachment theory and the developmental systems model, is the idea of mental representation. Mental representation refers to the internal working model that include children’s (and adults’) thoughts, feelings, and expectations about the self (e.g., the child), the relationship partner (e.g., the adult), and their mutual relationships (Bowlby, 1980; Pianta et al., 2003). In attachment research, the idea of mental representation was first used to study parent-child relationships. From the early 1990s, the concept of mental representation has been extended to the relationship between teachers and children, and plays an important role in theory and research about teacher-child relationships (Pianta et al., 2003; Spilt et al., 2022). Teachers’ and students’ mental representations are considered to be influenced by the daily interactions between teachers and children, and also by teachers’ and children’s previous relationship histories with important others (e.g., parents, friends, previous teachers or students; Pianta et al., 2003). As teachers and children have different relationship histories, their mental representations about the mutual relationship may differ as well (Pianta et al., 2003; Zee & Koomen, 2017).

According to the developmental systems model (Pianta et al., 2003), external factors, such as cultural values, are also important to take into account when studying teacher-child relationships. How teachers and children form mutual relationships may be influenced by the cultural beliefs and values that are dominant in the country they live. Accordingly, researchers have suggested that teacher-child relationships in Eastern countries may be qualitatively different from relationships in Western countries (Beyazkurk & Kesner, 2005; Jia et al., 2009; Yang et al., 2013). To understand cross-cultural differences in teacher-child relationship quality, the dimension of individualism versus collectivism might be especially relevant (Hofstede et al., 2010). This dimension considers to what extent people in a society emphasize the independence of each person (individualism) or the interdependence between individuals (collectivism; Hofstede et al., 2010). In individualistic countries, people tend to be independent of others and guide their behaviors based on personal goals. In contrast, in collectivistic countries, the interdependency between members is regarded as important and people tend to act in accordance with group norms (Hofstede et al., 2010). These different cultural values may have consequences on how teachers and children interact with each other (Pianta et al., 2003), and therefore, affect the quality of teacher-child relationships. More specifically, teachers in Western, individualistic countries may be more inclined to keep distance in relationships with children to encourage their independent development (Rothbaum et al., 2000; Triandis, 2018). In contrast, teachers in Eastern, collectivistic countries may tend to invest in intimate contact with children to stimulate their social engagement (Rothbaum et al., 2000; Triandis, 2018). For example, teachers in Western, individualistic countries tend to motivate children to play and explore the environment independently, with less involvement in children’s activities (cf. Rothbaum et al., 2000). On the contrary, teachers in Eastern, collectivistic countries are inclined to be protective and encourage children to stay close and dependent on teachers while playing and engaging in activities (cf. X. Chen et al., 1998; Hu, 2015). As such, teachers and children in Eastern, collectivistic countries may share more close relationships with each other than teachers and children in Western, individualistic countries. Another important dimension on which people in different countries can differ is the preferred way of dealing with interpersonal conflict (X. Chen & French, 2008). In Western, individualistic countries, direct and assertive strategies (e.g., directly expressing anger and negative emotions) are more commonly used (X. Chen & French, 2008). In contrast, in Eastern, collectivistic countries, compromise and disengagement are often favored, and openly showing negative emotions toward authority figures (e.g., teachers) is often discouraged (X. Chen & French, 2008; Hofstede et al., 2010). For instance, in Chinese kindergartens, children are often learned to follow teachers’ ideas, show obedience to teachers and do not argue with teachers (Hu, 2015). As such, teachers and children in Western, individualistic countries may be more likely to develop conflictual relationships with each other than teachers and children in Eastern, collectivistic countries.

Cross-Cultural Differences in Teacher-Child Relationships

To our knowledge, only six studies have explored cross-cultural differences in affective teacher-child relationships so far (Acar et al., 2019; Bear et al., 2014; Beyazkurk & Kesner, 2005; M. Chen et al., 2019; Jia et al., 2009; Yang et al., 2013). All six studies investigated teacher-child closeness. Two of these studies were based on the relationship perceptions of older children (i.e., upper elementary school or middle school students; Jia et al., 2009; Yang et al., 2013). Both studies found that students in China experienced more closeness with their teachers than students in the United States (Jia et al., 2009; Yang et al., 2013). Studies based on teachers’ relationship perceptions found mixed results (Acar et al., 2019; Beyazkurk & Kesner, 2005). That is, Beyazkurk and Kesner (2005) showed that kindergarten and first-grade teachers in Turkey perceived higher closeness with their students than American teachers. In contrast, Acar et al. (2019) found that teachers in Turkey experienced less closeness with five-year-old children than teachers in the United States. Two studies included both teachers’ and children’s perceptions of the relationship. Bear et al. (2014) revealed that third-to-six graders in China and the United States reported equal levels of teacher-student closeness, whereas Chinese teachers perceived less closeness with their students than American teachers. In contrast, M. Chen et al. (2019) found that third-to-six graders in China experienced more teacher-child closeness than their Dutch counterparts, whereas Chinese and Dutch teachers reported equal levels of teacher-child closeness.

Only three studies investigated cross-cultural differences in teacher-child conflict. Acar et al. (2019) found Turkish teachers to report less conflict with young children than American teachers. Similarly, M. Chen et al. (2019) noted that both teachers and upper elementary students in China experienced less teacher-child conflict than teachers and upper elementary students in the Netherlands. In contrast, Beyazkurk and Kesner (2005) showed that Turkish teachers and American teachers experienced equal levels of conflict with young children.

As such, cross-cultural studies about affective teacher-child relationships are scarce and revealed inconsistent results. Among them, only two studies focused on relationships between teachers and young children (Acar et al., 2019; Beyazkurk & Kesner, 2005) and found inconsistent results, with teachers from Turkey reporting either higher levels of closeness (Beyazkurk & Kesner, 2005) or lower levels of closeness (Acar et al., 2019) than teachers in America. The mixed findings might be due to the differences in the research samples. In Beyazkurk and Kesner (2005), the American sample and the Turkish sample had age differences, with the American children being three years older (nine-year-olds) than the Turkish children (six-year-olds). As older children tend to have less close relationships with teachers, the lower levels of closeness in the American sample may simply be because the investigated children were older than the Turkish children. In contrast, Acar et al. (2019) included American and Turkish children from similar age groups (4.5 years old), and thus their results may not be biased by children’s age. However, the findings of Acar et al. (2019) seemed to contradict findings from upper elementary school (e.g., M. Chen et al., 2019). Thus, a question arises as whether previously found cross-cultural differences in the relationship between teachers and older children also apply to teachers and young children. The present study, therefore, looked at teachers and 5-to-8-year-old children in the Netherlands and China, and examined how their relationship quality may differ across countries. Furthermore, the two studies about early teacher-child relationships only focused on teachers’ reports of relationship quality but did not look at young children’s relationship perceptions. Studies with older children, however, showed that cross-cultural differences in relationship quality were sometimes found in students’ perceptions but not in teachers’ perceptions (M. Chen et al., 2019) or the other way around (Bear et al., 2014). It thus appears important to include both teachers’ and children’s relationship perceptions when investigating cross-cultural differences in teacher-child relationship quality.

Cross-Cultural Differences in Agreement in Relationship Perceptions

As mentioned above, teachers and children have different relationship histories and thus may have different mental representations of their mutual relationship (Pianta et al., 2003). Previous studies conducted in Western countries found empirical evidence for this idea (Harrison et al., 2007; Jellesma et al., 2015; Koomen & Jellesma, 2015; Mantzicopoulos & Neuharth-Pritchett, 2003; Rey et al., 2007; Spilt et al., 2010). For example, teachers and upper elementary school children often had weak to moderate agreement on their relationship perceptions, with correlations tending to be somewhat stronger for conflict (.31 < r < .59) than for closeness (.21 < r < .38; Jellesma et al., 2015; Koomen & Jellesma, 2015; Rey et al., 2007). The agreement between teachers’ and young children’s relationship perceptions seemed to be weaker than those between teachers and older children, with correlations ranging from −.02 to 25 for closeness and from .17 to .28 for conflict (Harrison et al., 2007; Mantzicopoulos & Neuharth-Pritchett, 2003; Spilt et al., 2010).

To our knowledge, studies investigating teacher-child agreement in relationship perceptions in Eastern countries are lacking. One study, however, suggested that there are cross-cultural differences in the degree to which teachers and children agree in their relationship perceptions (M. Chen et al., 2019). They found that teachers and upper elementary school students in the Netherlands had lower agreement on mutual closeness but higher agreement on conflict than teacher-child dyads in China. According to the authors, a possible explanation could be that compared with Dutch teacher-child dyads, it is more common for teachers and children in China to emphasize the harmonious part of the relationship and avoid talking about discordances (X. Chen & French, 2008; Yang et al., 2013). It is not yet known, however, whether this finding also generalizes to young children and their teachers. Therefore, the second aim of the present study was to investigate whether there were also cross-cultural differences in the degree to which teachers and young children agree on their relationship perceptions.

The Present Study

The present study aimed to advance insight into cross-cultural differences in early teacher-child relationships, and compared the quality of relationships between teachers and young children (5-to-8-year-olds) across the Netherlands and China. The Netherlands is usually considered to be a Western, individualistic country, whereas China can be seen as an Eastern, collectivistic country (Hofstede et al., 2010). These two countries were mainly chosen due to practical reasons for data collection (i.e., the first author is originally from China and works at a Dutch university). However, it is also theoretically interesting to compare the quality of teacher-child relationships between these two countries. The only two previous cross-cultural studies with young children both compared Turkey and the United States. Thus, the comparison between China and the Netherlands will provide more insight into cross-cultural differences in teacher-child relationship quality when the comparisons are made between other countries. According to the cultural dimension model of Hofstede et al. (2010), both the United States and the Netherlands score high on individualism (91 and 80 respectively, out of 100), and China (20) is rated lower on individualism than Turkey (38). As such, the discrepancy between the Netherlands and China on Individualism is somewhat larger than the discrepancy between the United States and Turkey, which also makes the comparison between the Netherlands and China interesting.

We first examined cross-cultural differences in both teachers’ and young children’s perceptions of the relationship. Based on theoretical assumptions and most previous research (Acar et al., 2019; M. Chen et al., 2019; Jia et al., 2009; Yang et al., 2013), we expected teachers and children in China to perceive higher closeness and lower conflict in their mutual relationships than teachers and children in the Netherlands. Second, we examined cross-cultural differences in the agreement between teachers’ and young children’s relationships perceptions. Based on M. Chen et al. (2019), we expected that teachers’ and children’s agreement on closeness would be higher in China than in the Netherlands, whereas their agreement on conflict would be lower in China than in the Netherlands.

Methods

Sample and Selection Procedure

Data collection in the Netherlands was approved by the ethical committee of the University of Amsterdam (file number: 2018-CDE-9440). At the time of data collection in China, there was no official ethical committee in the region where the data were collected. We thus invited a senior researcher to read the proposal and confirm that our data collection procedure complied with the laws and rules in China. The Dutch data were collected by trained bachelor and master students in the second half of the 2018–2019 and 2019–2020 school year. All the data were collected before schools were suspended due to the COVID-19 pandemic. The first author and trained assistants collected data in Zhejiang, China in the second half of the 2018–2019 school year. Research assistants and the first author contacted schools from address lists and from their personal networks via phone and e-mail. Once schools agreed to participate, parents and teachers received an information letter about the aims of the study. In the end, we got consent from 94.73% of the invited teachers and 61.59% of the invited parents. After informed consent was obtained from parents and teachers, the first author and research assistants visited schools to collect data.

In each participating classroom, we strived to include eight randomly selected children (4 boys, 4 girls) for participation. The decision to include eight children per class was made because completing questionnaires about more children would be overly burdensome for teachers and would make them less motivated to participate (cf. M. Chen et al., 2019; Zee et al., 2018). Both in the Netherlands and China, we started with randomly selecting ten children (five boys, five girls) per classroom for the child-report. We selected a bit more children, in order to prevent the examiner from having to go back to the school if one or two of the children were absent. Subsequently, eight of these ten children were randomly selected for the teacher to complete the questionnaire about. As a result, the number of child reports (N_Dutch = 234; N_Chinese = 190) was somewhat higher than the number of teacher reports (N_Dutch = 201; N_Chinese = 152). In classrooms where parental permission was available for fewer than eight children, teachers provided reports for all children with parental permission.

In the Netherlands, elementary education consists of eight years, with two years of kindergarten and six years of elementary school (Luijkx & de Heus, 2008). Children start kindergarten when they turn four years old and are often around six years old when they go to first grade. As kindergarten and grade 1 are both part of elementary education and are located in the same building, kindergartners don’t change schools and usually do not experience huge differences when they make the transition to first grade. In Zhejiang, China, children also start kindergarten at four years old but they stay in kindergarten for three years. Thus, Chinese children are often around 5 years old in the second year of kindergarten, and around 6 years old in the third year of kindergarten (see Hu et al., 2014 for an overview of the kindergarten system in Zhejiang, China). As 4-year-olds may still have insufficient cognitive skills to respond to the used child instrument in an adequate way (Spilt et al., 2010), we focused on young children who were at least five years old. In the Netherlands, children are thus either in the second year of kindergarten or first grade. Although we include children both from kindergarten and elementary school, it may not be much of a problem as kindergarten classes and elementary classes are within the same system in the Netherlands. In China, children are in the second year or third year of kindergarten. We selected children based on their age instead of their grade level, as age has been found to be associated with the affective quality of teacher-child relationships (see McGrath & van Bergen for a review).

In the Netherlands, first graders go to school five days a week, while kindergartners go either four days or five days a week, depending on the school. In China, children usually go to school five days a week. As for teachers, in the Netherlands, teachers in elementary education basically teach all subjects. However, a lot of teachers work part-time. When there are part-time teachers, a class is often taught by two part-time teachers. In this case, we focused on the teacher who spend more time with the children. Only if this teacher did not want to participate, we asked the teacher with fewer working days to complete the questionnaire. To make sure that teachers spent enough time with children, only teachers who taught the class for at least two days a week were asked to participate. In China, kindergarten teachers usually work full-time. Each class often has two teachers, who teach all subjects and share the teaching tasks. However, one teacher is called the head teacher and takes more responsibilities than the other teacher, whose job is mainly assisting the head teacher. As the head teacher usually spends more time with children and has a role that is more similar to kindergarten teachers in the Netherlands, we decided to focus on head teachers in China.

The Dutch sample included 234 children (47.7% girls) from 38 classes in 15 schools across the Netherlands. These children were either in the second year of kindergarten (n = 138) or in first grade (n = 96). Children had a mean age of 5.82 years (SD = 0.74; range = 5–8 years). According to their teachers, most children (81.2%) belonged to the majority ethnic group (i.e., Dutch), whereas other children (18.8%) had a minority ethnic background (e.g., Turkish, Surinamese). Out of the 38 teachers invited for participation, three teachers did not provide teacher-report. Hence, the Dutch sample included 35 teachers (97.1% female), with on average 16.31 years of teaching experience (SD = 13.41, range = 0.5–40 years). Most of the teachers worked full time (n = 17), with other teachers working either four days (n = 8), three days (n = 8), or two days (n = 2) a week. On average, each teacher in the Netherlands provided reports for 5.74 children (SD = 2.01, range = 1–8). In total, teacher-reports were available for 201 children (49% girls).

The Chinese sample consisted of 190 kindergartners (50.5% girls) from 19 classes in three schools in Zhejiang, China. These children were from the second (n = 70) or third (n = 120) year of kindergarten, with a mean age of 5.60 years (SD = 0.51, range = 5–7 years). Most children (96.8%) belonged to the ethnic majority group (i.e., Han Zu), whereas a small proportion of children (3.2%) had an ethnic minority background (e.g., She Zu). The Chinese sample included 19 teachers (100% female), with on average 10.47 years of teaching experience (SD = 5.15, range = 4–25 years). All teachers worked full time. All Chinese teachers provided reports for eight children. Thus, teacher reports were available for 152 children (50% girls).

Procedure

The data collection procedure was the same across the two countries. The examiners got a short training where they learned how to administer the instrument to assess children’s relationship perceptions, that is, the Young Children’s Appraisals of Teacher Support (Y-CATS). The Y-CATS is an age-appropriate questionnaire for young children that is administered individually by the researcher. More information about the content of the Y-CATS can be found in the Measurements section. The Y-CATS was then administered to the selected children individually, in a separate room, with only the examiner being present. Children responded to the Y-CATS on two occasions, one to two weeks apart. This was done because young children may have less stable relationship perceptions (c.f. Spilt et al., 2010), and administering the Y-CATS twice would help to get a more reliable measure of children’s relationship perceptions. The administration of the Y-CATS took approximately 10 minutes per occasion for each child.

After children finished the first occasion of the Y-CATS, eight children were randomly selected for the teacher reports. Each teacher then filled out a questionnaire about their relationships with the selected children and some other topics beyond the scope of the present study. The teacher questionnaire took approximately 30 minutes to complete for eight children.

Measurements

Teachers’ Relationship Perceptions

Teachers’ perceptions about their relationship with individual children were measured with the Closeness (five items) and Conflict subscales (five items) from the validated Dutch short version of the Student Teacher Relationship Scale (STRS; Koomen et al., 2012; Zee & Koomen, 2017). The Closeness subscale taps the degree of warmth, intimacy, and open communication between a teacher and a specific child (e.g., “I share an affectionate, warm relationship with this child” and “This child seems to feel secure with me”). The Conflict subscale describes the degree of discordance and negativity in the relationship (e.g., “This child and I always seem to be struggling with each other” and “This child feels that I treat him/her unfairly”). Teachers indicated to what extent each statement applied to their relationship with a certain child on a five-point scale, from 1 (Definitely does not apply) to 5 (Definitely applies). For full information about the item content, please refer to Table A1 in the Appendix.

The reliability and validity of the short version of the STRS have been supported both in the Netherlands and in China before (M. Chen et al., 2019; Zee & Koomen, 2017; Zee et al., 2013, 2017). In the present study, the STRS showed satisfactory reliability both in the Dutch sample (McDonald’s omega is .75 for Closeness and .82 for Conflict) and in the Chinese sample (McDonald’s omega is .82 for Closeness and .81 for Conflict). Previous research has found measurement invariance of the STRS for cross-cultural comparisons between the Netherlands and China in an upper elementary school sample (M. Chen et al., 2019). In the present study, evidence for partially strong measurement invariance across China and the Netherlands was found as well.¹ Partially strong invariance is sufficient for making meaningful cross-cultural comparisons (Byrne et al., 1989).

Children’s Relationship Perceptions

Children’s relationship perceptions were assessed by the Warmth and Conflict subscales of the Young Children’s Appraisals of Teacher Support (Y-CATS; Mantzicopoulos & Neuharth-Pritchett, 2003). The Warmth subscale (10 items) describes whether a child feels supported, encouraged, and accepted in the relationship with a teacher (e.g., “Miss/Mr. [name of the teacher] is nice to you” and “Miss/Mr. [name of the teacher] often smiles to you”). The Conflict subscale (10 items) asks children about the degree of discordance, quarrels, and negativity in the relationship (e.g., “Miss/Mr. [name of the teacher] often says that you are bad” and “Miss/Mr. [name of the teacher] easily gets angry with you”). The Warmth subscale parallels the Closeness subscale of the STRS and the Conflict subscale of the Y-CATS is comparable with the Conflict subscale of the STRS (Mantzicopoulos & Neuharth-Pritchett, 2003).

The Y-CATS items were printed on small cards and were read by the examiner one by one to the child. The child could choose to put the printed card into a safe to indicate that the read statement was “true” (scored as 2), or into a trashcan to indicate that it was “untrue” (scored as 1). To encourage children’s honest responses, they were assured that their answers to the questions would not be shared with their teachers. Every child finished two practice questions before responding to the Y-CATS questions. The practice questions included two easily verifiable statements, with one describing a true situation (Miss/Mr. [name of the teacher] is taller than you) and the other describing an untrue situation (Miss/Mr. [name of the teacher] has blue hair).

A validated Dutch version of the Y-CATS was available (Spilt et al., 2010). The English version of the Y-CATS was translated into Chinese by the first author (native Chinese), and was translated back into English by two master students who are native Chinese and fluent in English. The back-translated versions were then checked by the third author (native Dutch) to see if the translation was correct. As children completed the Y-CATS two times, we aggregated the answers on both occasions per item, and recoded the answer options into three categories. (1 = “untrue” on both occasions, 2 = “untrue” on one occasion, “true” on the other occasion, 3 = “true” on both occasions).

Analysis

Data analysis was conducted in Mplus version 7.32 (Muthén & Muthén, 1998–2012). To answer the first research question, we used multiple group models to compare the latent means of teachers’ and children’s relationship perceptions across countries. However, meaningful comparisons of latent means can only be made when sufficient measurement invariance has been obtained. Although partially strong invariance across China and the Netherlands has been established for teachers’ relationship perceptions (M. Chen et al., 2019), the Y-CATS has not been used for cross-cultural comparisons before. We therefore first checked the measurement invariance of children’s relationship perceptions before proceeding to the actual analyses.

The procedure of testing measurement invariance with categorical data were used to test the measurement invariance of the Y-CATS (Bowen & Masa, 2015; H. Wu & Estabrook, 2016). As models with categorical data are likely to encounter convergence problems (cf. P. W. Lei & Shiverdecker, 2020), we fitted separate models for Warmth and Conflict to reach model convergence. For both models, we started with a configural model where all the factor loadings and thresholds were freely estimated across the Dutch and Chinese samples. In the standard procedure of testing measurement invariance of continuous indicators, the common practice is to test weak invariance first and then strong invariance. Different from this procedure, for models with categorical indicators, the configural model is directly compared to a strong invariance model where all the loadings and thresholds are constrained to be equal across samples (Bowen & Masa, 2015). When strong invariance could not be established, each indicator (both loadings and thresholds) was freed one by one to locate non-invariant indicators (Bowen & Masa, 2015; Raykov et al., 2018). As multiple tests were performed simultaneously in this step, a Bonferroni correction was applied to get a corrected p-value for model comparisons (i.e., .05 divided by the number of comparisons; e.g., .05/8= .006; Raykov et al., 2018). In the end, all non-invariant indicators suggested by model comparisons were freed in a partially strong invariance model (Raykov et al., 2018). Meaningful cross-cultural comparisons can still be made when at least two indicators are invariant across groups (Byrne et al., 1989).

After sufficient measurement invariance was established for the Y-CATS, mean comparisons of teachers’ and children’s relationship perceptions were made based on the final (partially) strong invariance model. The means in the Dutch group were fixed to zero whereas the means in the Chinese group were freely estimated to indicate the mean differences. We used Z tests to examine whether mean differences differed significantly from zero. Standardized latent mean scores (Cohen’s d) were reported to indicate whether a significant difference was small (.20 ≤ d < .50), medium (.50 ≤ d < .80), or large (.80 ≤ d; Cohen, 1988).

To answer the second research question, we compared the strength of correlations between teacher-reported and child-reported relationship quality across the Dutch sample and the Chinese sample. In both samples, each teacher-child dyad received a unique ID number, to match teacher-report and child-report for specific teacher-child dyads. To calculate correlations between teachers’ and children’s relationship perceptions, we first saved the factors score from the models in the previous step (research question 1) and paired teacher-report with child-report based on the ID numbers. Next, correlations were calculated for each sample separately and were transformed into Z-scores using Fisher’s r-to-z transformation (Fisher, 1915, 1921). Finally, Wald’s Z tests were used to examine whether correlations between teacher-report and child-report differed across the Dutch and Chinese samples.

As children’s Gender and Age, and teachers’ Teaching Experience were frequently found to affect the quality of teacher-student relationships (McGrath & van Bergen, 2015; Zee & Koomen, 2017), we also tested models including these variables as covariates. Given that some teachers in the Netherlands worked part-time whereas all Chinese teachers worked full-time, we also included teachers’ Teaching Days as a covariate. After including these covariates, the results appeared to be the same as in the models without covariates. Due to model parsimony consideration, we reported results from models without covariates.

Model Estimation and Model Fit

Models testing the measurement invariance of children’s relationship perceptions were estimated with weighted least squares means and variance adjusted (WLSMV) estimation. Model fit was indicated by chi-square tests and two fit indices, Root Mean Square Error of Approximation (RMSEA) and Comparative Fit Index (CFI). As the cutoff values of fit indices developed for continuous data tend to be too lenient for models with categorical data, we adopted more stringent cutoff values (Bowen & Masa, 2015; Monroe & Cai, 2015). A model can be considered as satisfactory with a non-significant chi-square value, and CFI >.95 and RMSEA <.06 (Bowen & Masa, 2015). Nested models for testing measurement invariance were compared by conducting WLSMV chi-square difference tests in Mplus. Model equivalence can be indicated by a non-significant chi-square difference value.

For all multiple group models, missing values (<2.3% are missing) were handled by full maximum likelihood estimation (FIML) in Mplus. Our data had a multi-level structure, as children were nested within teachers (up to 20.73% of the variances in relationship perceptions were at the teacher level). Hence, we used the “Type = Complex” option in Mplus to compute standard errors with a sandwich estimator that takes into account the nestedness in data (Muthén & Muthén, 1998–2012).

Results

Measurement Invariance of Children’s Relationship Perceptions

The configural invariance model of child-perceived Warmth showed a satisfactory model fit, χ² (70) = 77.86, p = .243, RMSEA = .023, 90% CI = [.000, .048], CFI = .984. However, two items (Y-CATS19 and Y-CATS21; see item content in Table 1) showed low factor loadings (λ < .20) and were thus excluded from further analysis. The reliability of the eight-item Warmth subscale was acceptable in the Dutch sample (McDonald’s omega = .60) and high in the Chinese sample (McDonald’s omega = .80). Therefore, we continued testing measurement invariance with eight items.

Table 1. Components and measurement invariance of children’s relationship perceptions.

Item	Content	Loading	Thresholds(1)	Thresholds(2)
Warmth
Y-CATS1	Miss/Mr. … likes your family	.50	−1.64	−1.17
Y-CATS3	Miss/Mr. … tells you you’re smart	.39/.97	-1.77/-.1.85	-0.99/-1.23
Y-CATS7	Miss/Mr. … likes you	.76	−2.32	−1.55
Y-CATS10	Miss/Mr. … is nice to you	.78	−2.49	−1.70
Y-CATS13	Miss/Mr. … says nice things about your work	.68	−1.75	−1.22
Y-CATS16	Miss/Mr. … helps you when you find something hard	.53	−1.55	−0.95
Y-CATS25	Miss/Mr. … listens if you ask something	.45	−1.78	−1.08
Y-CATS27	Miss/Mr. … often smiles to you	.56/.97	-1.00/-1.97	-0.35/-1.36
Conflict
Y-CATS2	Miss/Mr. … often says that you are bad	.88/.81	1.10/1.79	1.52/2.64
Y-CATS5	Miss/Mr. … gives you work that is too hard for you	.55/.52	0.09/0.79	0.74/1.72
Y-CATS8	Miss/Mr. … tells you that you are doing something wrong a lot	.69	0.72	1.45
Y-CATS11	Miss/Mr. … tells you to stop work that you like doing	.65/.65	0.13/0.62	0.75/1.53
Y-CATS14	Miss/Mr. … easily gets angry with you	.86	1.20	1.87
Y-CATS17	Miss/Mr. … is sometimes mean	.41	0.55	1.11
Y-CATS22	Miss/Mr….has too many rules for our class	.69	0.14	0.85
Y-CATS24	Miss/Mr. … tells you that you do not listen	.84	0.93	1.60
Y-CATS26	Miss/Mr. … tells you to do work you don’t want to do	.68/.73	0.19/0.80	0.79/1.77

Note. Items that are invariant across countries had the same loading and thresholds across samples. For non-invariant items, loadings and thresholds for both samples (Dutch sample/Chinese sample) are provided in bold. Y-CATS19 (Miss/Mr … .tells good stories), Y-CATS21(Miss/Mr…asks how things are at home), and Y-CATS20 (Miss/Mr … tells you to stop work that you like doing) were excluded from the main analysis due to low factor loadings.

The configural invariance model with the eight Warmth indicators had a satisfactory fit, χ² (40) = 44.87, p = .275, RMSEA = .024, 90% CI = [.000, .055], CFI = .990. The strong invariance model, however, showed a significantly worse fit than the configural model, $\mathrm{\Delta }\chi$² (14) = 32.47, p = .003. Thus, indicators were freed one by one to find out which indicators should be freed. Model comparisons suggested freeing two items (Y-CATS3 and Y-CATS27; see item content in Table 1). The loadings and thresholds of these items were thus freed across countries. This partially strong invariance model did not fit worse than the configural model, $\mathrm{\Delta }$χ² (10) = 10.17, p = .425. The model also showed a satisfactory fit, χ² (50) = 54.26, p = .315, RMSEA = .020, 90% CI = [.000, .050], CFI = .991, and was therefore taken as the final model for mean comparisons.

The configural invariance model of child-perceived Conflict showed satisfactory model fit, χ² (70) = 84.88, p = .109, RMSEA = .032, 90% CI = [.000, .053], CFI =.986. However, one item (Y-CATS20) had a low factor loading (λ = .10) and was therefore excluded from further analyses. The nine-item Conflict subscale showed high reliability both in the Dutch sample (McDonald’s omega = .81) and in the Chinese sample (McDonald’s omega = .79).

The configural invariance model with the nine Conflict indicators had a satisfactory fit, χ² (54) = 59.34, p = .287, RMSEA = .022, 90% CI = [.000, .050], CFI = .995. Compared with this model, the strong invariance model fitted significantly worse, $\mathrm{\Delta }\chi$² (16) = 93.74, p < .001. Freeing indicators one by one suggested four non-invariant items (Y-CATS2, Y-CATS5, Y-CATS11, and Y-CATS26; see Table 1). We thus freed loadings and thresholds of these items in a partially strong invariance model, which did not fit significantly worse than the configural model, $\mathrm{\Delta }$χ² (8) = 14.54, p = .069. This model also showed a satisfactory model fit, χ² (62) = 73.84, p = .144, RMSEA = .030, 90% CI = [.000, .053], CFI = .989, and was thus taken as the final model for mean comparisons

Cultural Differences in Teachers’ and Children’ Relationship Perceptions

For teachers’ relationship perceptions, the multiple group comparison of latent means showed that Dutch teachers and Chinese teachers did not differ in the mean of Closeness, M_diff  = 0.001, Z = .002, p = .998, 95% CI = [−0.57, 0.57], or in the mean of Conflict, M_diff  = 0.19, Z = 0.77, p = .445, 95% CI = [−0.30, 0.68]. These results indicated that Dutch and Chinese teachers experienced equal levels of Closeness and Conflict in their relationships with individual children.

As for children’s relationship perceptions, the Chinese and Dutch samples differed significantly, both in the mean of Warmth, M_diff  = −0.88, Z = −3.21, p = .001, 95% CI = [−1.42, −0.34], and Conflict, M_diff  = 0.97, Z = 7.16, p < .001, 95% CI = [0.70, 1.23]. Chinese children reported lower Warmth and higher Conflict than Dutch children. These differences could be considered as large for both Warmth, Cohen’s d = −1.09, and Conflict, Cohen’s d = 1.23.

Agreement Between Teachers’ and Children’s Relationship Perceptions

Correlations between the factor scores of teachers’ and children’s relationship perceptions are presented in Table 2, for the Dutch sample and Chinese sample separately. Teacher-perceived Closeness was not significantly correlated with child-perceived Warmth, neither in the Dutch sample, r =.14, p = .056, nor in the Chinese sample, r = −.01, p = .912. The strength of these correlations did not differ significantly across countries, Z = 1.26, p = .104. Teacher-perceived Conflict was positively correlated with child-perceived Conflict both in the Dutch sample, r = .25, p < .001, and in the Chinese sample, r = .21, p = .009. However, the strength of these associations did not differ across countries, Z = 0.39, p = .348.

Table 2. Correlations between latent factors scores of teachers’ and children’s relationship perceptions.

	1	2	3	4
1. Teacher-reported Closeness	-	−78**	−.01	−.17*
2. Teacher-reported Conflict	−.55**	-	−.02	.21**
3. Child-reported Warmth	.14	−.17*	-	−.14*
4. Child-reported Conflict	−.19**	.25**	−.29**	-

Note. Correlations are presented for the Dutch sample (below diagonal) and Chinese sample (above diagonal), separately.

Discussion

The present study is one of the first studies to investigate cross-cultural differences in both teachers’ and young children’s perceptions of teacher-child relationship quality. We are also among the earliest to explore cross-cultural differences in the agreement between teachers’ and young children’s perceptions of mutual relationships. Four main findings emerged from this study. First, we found sufficient measurement invariance (partial scalar invariance; Byrne et al., 1989) for the instrument used to measure young children’s relationship perceptions (i.e., the Y-CATS) across the Netherlands and China. In other words, the Y-CATS measures children’s relationship perceptions in comparable ways in the Netherlands and China, and thus we are not running the risk of “comparing chopsticks with forks” (F. F. Chen, 2008). These findings suggest that the Y-CATS can be used to make meaningful cross-cultural comparisons of young children’s relationships perceptions across the Netherlands and China. Furthermore, previous studies have successfully used the Y-CATS both in individualistic countries (i.e., the U.S. and the Netherlands, Mantzicopoulos & Neuharth-Pritchett, 2003; Spilt et al., 2010) and in more collectivistic countries (i.e., Italy and Greece; Longobardi et al., 2016; Papadopoulou & Gregoriadis, 2017). These findings further provide evidence that this instrument can be used in different cultural contexts. As these previous studies did not test measurement invariance across different countries, more research is needed to find out whether the Y-CATS can also be used to make meaningful cross-cultural comparisons across other countries.

Second, in contrast to our hypothesis (Acar et al., 2019; M. Chen et al., 2019), Dutch and Chinese teachers appeared to experience equal levels of closeness and conflict in their relationships with young children. Although not in line with our expectations, this finding is not very surprising, given the inconsistent findings from previous studies using teachers’ relationship perceptions. Teachers in Eastern, collectivistic countries were found to report either more closeness (Beyazkurk & Kesner, 2005), less closeness (Acar et al., 2019; Bear et al., 2014), or equal levels of closeness (M. Chen et al., 2019) than teachers in Western, individualistic countries. These inconsistent findings are hard to explain, as studies comparing the same two countries (e.g., Turkey versus the United States; Acar et al., 2019; Beyazkurk & Kesner, 2005) or focusing on the same grade levels (e.g., upper elementary school; Bear et al., 2014; M. Chen et al., 2019) also showed contradictory results. Therefore, inconsistencies in findings may not simply be due to the specific countries that were compared or the grade levels on which these studies have focused. As a possible explanation for our finding, the degree of individualism/collectivism at the country level may not necessarily reflect each individual teachers’ cultural values. For instance, although the Netherlands is more individualistic than China at the country level, some Chinese teachers may hold more individualistic values than Dutch teachers. As we recruited teachers in China from a relatively developed area (Zhejiang province), these teachers may be more influenced by Western values and could be more individualistic than teachers in other areas of China (X. Chen & French, 2008; Y. Wu et al., 2022). As such, Chinese teachers in our study may have cultural values that are more comparable to Dutch teachers, and this might explain why we did not find teachers’ relationship perceptions to differ across countries. To further explore why teachers’ relationship perceptions may (and may not) differ across countries, it could be beneficial to measure each individual teachers’ cultural values, for example, by using the Individualism and Collectivism Scale (Triandis & Gelfland, 1998).

Third, in contrast to our expectations (M. Chen et al., 2019; Jia et al., 2009; Yang et al., 2013), we found that Chinese children perceived both less warmth and more conflict with teachers than Dutch children. These findings are surprising given that previous research based on older children’s relationship perceptions consistently revealed that students in Eastern, collectivistic countries experienced more close and less conflictual relationships with teachers than students in Western, individualistic countries (M. Chen et al., 2019; Jia et al., 2009; Yang et al., 2013). The surprising results may be related to another important cultural dimension known as power distance, which reflects whether the relationship between teachers and children is more hierarchical or more equal (Hofstede et al., 2010). Different from the collectivism/individualism dimension (i.e., the degree to which teachers tend to stay closely connected with children), power distance shows the degree to which teachers are more dominant, superior, and more powerful than children (Hofstede et al., 2010). As power distance is higher in China than in the Netherlands, Chinese teachers may tend to pay more attention to children’s behavioral management, be more strict with children’s misbehaviors, and tend to ask children to be obedient, compliant, and staying quiet (Hu et al., 2016). This hierarchical relationship may especially make young children in China feel vulnerable in relationships with teachers, in contrast to upper elementary or middle school students (M. Chen et al., 2019; Jia et al., 2009; Yang et al., 2013). As young children still lack sufficient self-regulation skills to manage their behaviors (Montroy et al., 2016), Chinese teachers’ requests for obedience and compliance could elicit feelings of insecurity and experiences of confrontations and negativity with teachers at this age (X. Chen et al., 2000; Portilla et al., 2014). In contrast, older children have developed better self-regulation skills (Montroy et al., 2016) and have a longer period to internalize the cultural expectations of showing conformity to teachers. Thus, older children in China may become able to comply with teachers’ requests for obedience, and hence, experience their relationships with teachers as even more close and less conflictual than Dutch students (M. Chen et al., 2019). To examine the tenability of this explanation, future cross-cultural studies may follow students from kindergarten to upper elementary school and investigate the changes in students’ relationship perceptions. As it is now, our findings seem to imply that results from cross-cultural comparisons of older children’s relationship perceptions cannot simply be generalized to younger children’s relationship perceptions.

Fourth, in contrast to our hypothesis (M. Chen et al., 2019), we did not find cross-cultural differences in the agreement between teachers’ and children’s relationship perceptions. This was an unexpected finding, as M. Chen et al. (2019) did find cross-cultural differences in the degree of agreement between teachers’ and upper elementary students’ relationship perceptions. Again, a possible explanation may be that the present study focused on 5-to-8-year-old children, whereas M. Chen et al. (2019) investigated third-to-sixth graders. M. Chen et al. (2019) suggested that upper elementary students and teachers in China were more likely to focus on relational warmth and connectedness and avoid acknowledging the conflict and negativities in the relationship, as compared to their Dutch counterparts. Consequently, Chinese teacher-child dyads appeared to have higher agreement on closeness and lower agreement on conflict than Dutch teacher-child dyads. However, for young children, the levels of their communication skills and cognitive abilities may be more influential to their agreement with teachers on relationship perceptions. More specifically, young children may still lack sufficient communication skills and abilities for abstract reasoning, which may especially affect their agreement with teachers on the degree of warmth/closeness in the relationship (Mantzicopoulos & Neuharth-Pritchett, 2003). Indications of warmth are often subtle, nuanced, and more difficult for young children to capture, and thus young children often have relatively low agreement with teachers on this dimension (cf. Jellesma et al., 2015; Koomen & Jellesma, 2015; Mantzicopoulos & Neuharth-Pritchett, 2003; Spilt et al., 2010). As this phenomenon concerns basic developmental traits, it may apply to young children both in the Netherlands and China and hence, no cross-cultural differences were found in the agreement on warmth/closeness. In line with this idea, young children in both countries appeared to lack agreement with their teachers on the degree of closeness in the relationship. Nevertheless, in both countries, young children and teachers did show a significant agreement on relational conflict. As indicators of conflict are often more salient and observable, it seems that young children can reach a certain degree of agreement with teachers when it comes to the quarrels and negativity in the relationship (Hughes, 2011; Zee & Koomen, 2017). Against expectations, agreement on conflict also appeared to be equally strong across the Netherlands and China. It could be that young children in China do not yet have the tendency to avoid acknowledging conflict with teachers, resulting in finding similar agreement in conflict in China as in the Netherlands (cf. M. Chen et al., 2019). More support for this explanation, however, is needed.

In sum, the present study supported the theoretical assumption that teachers and children have different perceptions of their mutual relationship (Pianta et al., 2003). Together with the findings of previous studies (Bear et al., 2014; M. Chen et al., 2019), our findings suggest that including young children’s perceptions provides additional information about cross-cultural differences in teacher-child relationships that is not revealed by results based on teachers’ relationship perceptions. Thus, it seems important to include both teachers’ and young children’s relationship perceptions in cross-cultural research. As findings seem to be different for young children as for older children, it would be interesting for future research to follow children and their teachers for a long period to allow the examination of possible developmental differences in teacher-child relationships across countries.

Limitations

This study is among the first cross-cultural studies that focused on early teacher-child relationships and compared young children’s relationship perceptions across countries. Nevertheless, this study still has limitations. First, we focused on teachers and children in the Netherlands and China, and thus our findings may not necessarily generalize to comparisons between other countries. Second, we included a Chinese sample from a relatively developed area in China to make it more comparable to the Dutch sample in terms of social-economic status. In rural or less developed areas of China, the traditional collectivistic culture may be even more influential (Y. Li et al., 2015). Thus, it may be interesting to investigate whether our findings apply to these areas as well. Third, we did not focus on another important indicator of affective teacher-child relationships, that is, real-time, observable interactions between teachers and children (Pianta et al., 2003). Using observations to examine teachers’ and children’s interactive behaviors might help to gain even more advanced insights into cross-cultural differences in teacher-child relationships.

Fourth, the warmth scale of children’s relationship perceptions turned out to have lower reliability. In previous studies, the reliability of the warmth scale also appeared to be lower than that of the conflict scale (Mantzicopoulos & Neuharth-Pritchett, 2003; Papadopoulou & Gregoriadis, 2017; Spilt et al., 2010). A possible reason may be that indicators of warmth are often subtle and less observable than indicators of conflict and, hence, warmth can be more difficult for young children to capture. Given the lower reliability, the present study employed latent models and thus some of the measurement errors were already separated from the true scores. Nevertheless, the lower reliability may indicate that young children sometimes have difficulties to report about the degree of warmth in their relationships with teachers. Future research may thus consider making the items less abstract, or even using visual clues to help children understanding the content.

Finally, although the warmth scale of the Y-CATS (child perception) was developed to parallel the closeness dimension of the STRS (teacher perception; Mantzicopoulos & Neuharth-Pritchett, 2003), there are some differences in the specific items included in the two subscales (see Tables 1 and A1). This might explain the low agreement between child-reported warmth and teacher-reported closeness in both countries. However, some previous studies did find significant associations between the warmth scale of the Y-CATS and the closeness scale of the STRS (Longobardi et al., 2016; Mantzicopoulos & Neuharth-Pritchett, 2003; Spilt et al., 2010). To further explore whether the lower agreement in warmth/closeness was due to measurement issues, other child measures may also be used in future research to replicate our results.

Implications and Conclusions

Despite the limitations, the present study may provide helpful suggestions for researchers and school practitioners. First, our findings showed that young children’s perceptions of teacher-child relationships may differ across countries. As such, it is important for researchers and school practitioners to be aware that research findings based on child reports in Western, individualistic countries may not simply generalize to children in Eastern, individualistic countries. For example, relationship-focused programs that are developed to intervene in problematic teacher-child relationships in Western contexts (e.g., Playing-2-gether; Vancraeyveldt et al., 2015) may need to be adjusted when applied to tackle problematic relationships in Eastern countries like China.

Second, our results confirmed the theoretical assumption that children have different perceptions of their mutual relationship than teachers (Pianta et al., 2003). As cross-cultural differences were only found in children’s relationship perceptions but not in teachers’ perceptions, it seems that cross-cultural differences in teacher-child relationship quality may depend on whether teachers or children report about the relationship (cf. Bear et al., 2014; M. Chen et al., 2019). Future cross-cultural research is thus encouraged to include both teachers’ and children’s relationship perceptions, also in research with young children. For teachers and school practitioners, our findings indicate that it is important to take young children’s relationship perceptions into account when identifying children who share unfavorable relationships with their teachers.

In conclusion, the present study found that young children in China perceived their relationships with teachers as less close and more conflictual than young children in the Netherlands. Their teachers, however, reported an equal level of closeness and conflict in the relationship. In both countries, teachers and children had a significant agreement on relational conflict but lacked agreement on relational closeness. Teacher-child agreement on both closeness and conflict was equally strong across countries. As such, it seems important to also include young children’s relationship perceptions in cross-cultural comparisons of early teacher-child relationships.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Data Availability Statement

The data used in the research cannot be publically shared because data contain confidential information about the participants, whereas data and used materials are available upon request, by emailing: Mengdi Chen ([email protected]), or Dr. Debora Roorda ([email protected]).

Additional information

Funding

This work was supported by the CSC & UvA Scholarship as provided by the China Scholarship Council and the Research Institute of Child Development and Education of University of Amsterdam [Grant number: 201807720058].

Notes

1. More specifically, the configural model had a good fit, χ2 (66) = 115.04, p < .001, RMSEA = .065, 90% CI = [.045, .084], CFI = .921, SRMR = .057. The weak invariance model did not fit significantly worse than the configural model, ∆χ2 (8) = 4.98, p = .760, ΔCFI = .010, ΔRMSEA = −.008, ΔSRMR = .008. However, the strong invariance model had a significantly worse fit than the weak invariance model, ∆χ2 (8) = 41.95, p < .001, ΔCFI = −.084, ΔRMSEA = .016, ΔSRMR = .020. After freeing three intercepts, the partially strong measurement invariance had a comparable model fit as the weak invariance model, ∆χ2 (5) = 8.08, p = .152, ΔCFI = −.007, ΔRMSEA = .001, ΔSRMR = .000. The partially strong invariance model also showed a good fit, χ² (79) = 124.73, p < .001, RMSEA = .057, 90% CI = [.037, .076], CFI = .926, SRMR = .064.

Appendix

Table A1. Item content of the closeness and conflict subscale of the student teacher relationship sclae (STRS).

Item	Content
Closeness
Item1	This child openly shares his/her feelings and experiences with me.
Item2	If upset, this child will seek comfort from me.
Item3	I share an affectionate, warm relationship with this child
Item4	This child seems to feel secure with me
Item5	My interactions with this child make me feel effective and confident
Conflict
Item1	This child and I always seem to be struggling with each other.
Item2	This child feels that I treat him/her unfairly.
Item3	When this child is in a bad mood, I know we are in for a long and difficult day
Item4	This child’s feelings toward me can be unpredictable or can change suddenly
Item5	Dealing with this child drains my energy.