As mentioned in the section “The Quality of Interactions in Preschool,” there are various ways to capture interaction quality, including observation measures focusing on the level of the preschool group and the level of the individual child. An important advantage of measures focusing on the individual child compared to measures focusing on the level of the preschool group is that variations in the experiences of individual children regarding their interaction with their preschool teacher and peers can be assessed (Chien et al., 2010; Smidt and Rossbach, 2016). Therefore, interaction quality was measured at the individual child level with the observation tool “Individualized Classroom Assessment Scoring System” (inCLASS, Downer et al., 2012). This tool measures children’s interactions with teachers (e.g., the child’s communication with teachers), peers (e.g., the child’s experience of positive emotions with peers), and tasks (the degree to which the child is actively involved in tasks and activities) (Downer et al., 2010). Based on previous research (Downer et al., 2010; Slot et al., 2015; von Suchodoletz et al., 2015; Bohlmann et al., 2019; Smidt and Embacher, 2021), the 10 inCLASS dimensions were grouped into the following factors: teacher interactions (α = 0.85), which includes positive engagement with the teacher and teacher communication, peer interactions (α = 0.85), which includes peer sociability, peer communication, and peer assertiveness, task orientation (α = 0.58), which includes task engagement and self-reliance, and conflict interactions (α = 0.74), which includes teacher conflict, peer conflict, and behavior control (reverse coded). See Table 1 for descriptive statistics and Table 2 for intercorrelations. Except for task orientation, the internal consistencies of the factors are acceptable to good (Nunnally, 1978).

On one observation day (typically from 8 a.m. to 12 p.m.), up to four children from one preschool class were observed with up to four alternating observation cycles in different situations (e.g., free play, mealtime, planned activity). According to the manual, four cycles per child (a total of 16 observations for four children) can be completed in a 4-h visit (Downer et al., 2012). One cycle lasts 15 min, with 10 min for observing children and note-taking followed by 5 min for scoring in consultation with the manual (Downer et al., 2012). To determine scores, each dimension receives a code of 1 to 7 indicating a low (1 or 2), medium (3 to 5), or high (6 or 7) level, and data from each observation cycle are averaged to obtain a final score (Downer et al., 2010).

Data collection was carried out by 13 student assistants (students of educational science and psychology), who completed 2 days of training by a certified inCLASS trainer, which included a reliability test. In the first wave of the study, double coding was conducted to examine inter-rater reliability. Intraclass correlation coefficients (ICCs) of the single domains ranged between 0.75 and 0.95, indicating excellent inter-rater agreement (Cicchetti, 1994; see Smidt and Embacher, 2023a for information on the first study wave).

Preschool teachers rated their perception of their relationship with a particular child using a short form of the “Student-Teacher Relationship Scale” (STRS) (Pianta, 2001), which includes 15 items rated on a five-point scale ranging from 1 (“definitely does not apply”) to 5 (“definitely applies”) (Pianta, 1992). The German version used here was also part of the NUBBEK study (Tietze et al., 2015). The closeness subscale consists of eight items (α = 0.87, e.g., “I share an affectionate, warm relationship with this child”), and the conflict subscale consists of seven items (α = 0.89, e.g., “This child easily becomes angry at me”). For each subscale, a sum score of the individual items was calculated (Pianta, 2001).

Preschool teachers’ educational beliefs regarding their support of children were measured with a scale that was used in a German study (Schmidt and Smidt, 2021). The scale consists of 12 items assigned to subscales: beliefs on self-education (four items, α = 0.69., e.g., “When supporting children, it is important that the preschool teacher interfere as little as possible”), beliefs on co-construction (four items, α = 0.52, e.g., “When supporting children, it is important that the children are encouraged by the preschool teacher to find their own solutions”), and beliefs on instruction (four items, α = 0.79, e.g., “When supporting children, it is important that the children are taught a lot by the preschool teacher”). Each item was rated on a five-point scale ranging from 1 (“strongly disagree”) to 5 (“strongly agree”) (Schmidt and Smidt, 2021).

In contrast to the other study variables, preschool teachers’ educational beliefs regarding their support of children were partially measured during the first study wave (from April to June 2019), which took place about half a year before the second wave. Due to time and economic reasons, preschool teachers participating in the first wave of the study (71.76% of the preschool teachers in the current sample) did not have to rate their beliefs on self-education, co-construction, and instruction again in the second wave. In these cases, data from the first wave were used. For newly participating preschool teachers (28.24% of the sample), data on educational beliefs were collected during the second wave (from October to December 2019).

Preschool teachers’ work engagement was assessed with the “Utrecht Work Engagement Scale” (UWES-9) (Schaufeli et al., 2006). This short form consists of nine items (α = 0.94, e.g., “At my work, I feel bursting with energy”), which are rated on a seven-point scale ranging from 0 (“never”) to 6 (“always”) (Schaufeli et al., 2006). Preschool teachers’ self-efficacy was captured with the “Kurzskala zur Erfassung allgemeiner Selbstwirksamkeitserwartungen (short scale for measuring general self-efficacy beliefs)” (ASKU) (Beierlein et al., 2013), which measures general self-efficacy beliefs with three items (α = 0.86, e.g., “I can rely on my own abilities in difficult situations”) that are rated on a five-point scale ranging from 1 (“strongly disagree”) to 5 (“strongly agree”).

Based on previous study findings (e.g., Justice et al., 2008; Rudasill and Rimm-Kaufman, 2009; Downer et al., 2012; Linberg and Kluczniok, 2020; Ramirez and Linberg, 2022; Smidt and Embacher, 2023a,b), child characteristics (age, gender, language skills, personality types) and structural characteristics (child–staff ratio, preschool teachers’ work experience, number of children with immigration background per preschool class, and adequate equipment in the preschool class) were considered as control variables. The age of the children and their gender were captured through paper–pencil interviews or telephone interviews with their parents. Children’s language skills were measured with a mean score (T-value) of three subtests (understanding sentences, morphological rule formation, and phonological working memory, α = 0.73) of the “Sprachentwicklungstest für drei- bis fünfjährige Kinder (language development test for 3 to 5-year-old children)” (SETK 3–5) (Grimm, 2015).

Child personality was measured using the “Fünf Faktoren Fragebogen für Kinder−Kurzform (Five Factor questionnaire for children – short version)” (FFFK-K) (Asendorpf, 2007), which was developed within the framework of the Socio-Economic Panel (SOEP). Preschool teachers assessed children’s personality with 10 bipolar items (originally rated on an 11-point scale ranging from 0 to 10), which were assigned to the following factors: neuroticism (α = 0.71; e.g., worried−calm), extraversion (α = 0.85; e.g., talkative−quiet), intellect (α = 0.73; e.g., interested−uninterested), agreeableness (α = 0.73; e.g., gentle−stubborn), and conscientiousness (α = 0.60; e.g., orderly−disorderly) (Asendorpf, 2007; see Smidt and Embacher, 2023a for further information on the use of the questionnaire in the first study wave).

Three personality types have been shown to be predictive of several social outcomes and problem behaviors among children: resilients, overcontrollers, and undercontrollers (e.g., Asendorpf and van Aken, 1999; Hart et al., 2003; van den Akker et al., 2013). To derive these types, we followed the method of Asendorpf et al. (2001) and conducted a two-step clustering procedure (Ward, followed by k-means) (for a detailed description, see Smidt and Embacher, 2023a). The “Big Five” cluster profiles show that resilients are characterized by below-average scores on neuroticism and above-average scores on extraversion, intellect, agreeableness, and conscientiousness. Overcontrollers are described by above-average scores on neuroticism and below-average scores on extraversion, intellect, agreeableness, and conscientiousness. Undercontrollers score above average on extraversion and below average on neuroticism, intellect, agreeableness, and conscientiousness.

Referring to previous studies (e.g., Asendorpf et al., 2001; Barbaranelli, 2002), a double-cross validation was applied to evaluate the consistency of the three-cluster solution across 10 different subsamples (see Smidt and Embacher, 2023a). The median κ-value was κ = 0.81, which is considered acceptable (Asendorpf et al., 2001). Regarding structural characteristics, the child–staff ratio, preschool teachers’ work experience, and the number of children with an immigration background per preschool class were determined from the information provided by the preschool teachers. The child–staff ratio was calculated by dividing the number of children in the preschool group by the number of staff (preschool teachers and assistants) in the group. The resulting number indicates how many children are cared for per staff member. Work experience was measured by asking preschool teachers how many years in total they had been working in the profession. The number of children with an immigration background per preschool class was computed on the basis of the information provided by the preschool teachers. Preschool teachers’ perception of adequate equipment in the preschool class (e.g., sufficient writing materials, books) was captured with seven items (α = 0.65) from a scale used in the German BiKS study (based on the Home Observation for Measurement of the Environment, Caldwell and Bradley, 1984). Preschool teachers had three answer options (The following tools are available so often that. 1 = some children can play with it, 2 = about half of all children can play with it, 3 = almost all children can play with it).

We did not find statistically significant associations between preschool teachers’ beliefs on self-education, co-construction and instruction, and teacher interactions. Although educational beliefs are considered to filter, frame, and guide pedagogical activities (Fives and Buehl, 2012), it must be also noted that research on educational beliefs and interaction quality is not consistent with studies revealing no influence of preschool teachers’ educational beliefs on interactional processes in preschools (Kluczniok and Roßbach, 2014). In this regard, the results of the present study fit with the inconsistent research landscape. We also did not find associations between preschool teachers’ work engagement, self-efficacy, and teacher interactions. This pattern of findings deviates from previous research, which found positive associations between preschool teachers’ work engagement (Penttinen et al., 2020; Soininen et al., 2023) and self-efficacy (Jennings, 2015; Hu et al., 2021; Wolstein et al., 2021) and the quality of interactions in preschools. An explanation for the null findings may lie in the inCLASS, which primarily focuses on children’s interactions with preschool teachers and not vice versa as many other instruments focusing on interaction quality do (Schmidt et al., 2018). This could probably explain why no associations were found.

We did not find any associations between preschool teachers’ educational beliefs, work engagement, self-efficacy, and the quality of peer interactions. It could be argued that null findings are probably not surprising because children’s interactions with peers are in focus rather than interactions with preschool teachers. Furthermore, reference could be made to the specificities of the inCLASS, which focuses on the children’s perspective on interaction quality and less on preschool teachers. However, this does not mean that preschool teachers do not play an important role in fostering children’s interaction with peers. Research suggests that preschool teachers should use scaffolding (based on Vygotsky; see Bodrova and Leong, 2018) to facilitate children’s interactions with peers (Acar et al., 2017) and should engage children in cognitively stretching “high-yield” activities (Bruner, 1980; Kontos and Wilcox-Herzog, 1997) to promote peer interactions (Smidt and Embacher, 2020). Previous findings from Austria indicate that the potential in this respect has not yet been fully tapped (Smidt and Embacher, 2020).

Preschool teachers’ endorsement of instruction was lower on average compared with self-education and co-construction, as Table 1 illustrates (see Schmidt and Smidt, 2021 and Mackowiak et al., 2022 for similar findings). However, it was positively related to children’s task orientation, whereas educational beliefs on co-construction revealed a negative association with task orientation. No effects were found for beliefs on children’s self-education. The positive effect of beliefs on instruction is reasonable because it can be assumed that an endorsement of instruction corresponds with preschool teacher-directed activities with clear learning goals (Schmidt and Smidt, 2021). Preschool teacher-led activities were found to be positively related to the development of children’s academic skills (de Haan et al., 2014; Goble and Pianta, 2017), and previous research also indicates that preschool teacher-led activities may reduce children’s off-task behavior (Rimm-Kaufman et al., 2005), a finding that fits with the present study. This is also true for the null finding regarding the beliefs on self-education, which corresponds with the expectation that preschool teachers endorsing self-education would tend to be reserved in practicing teacher-directed activities with clear learning goals (Schmidt and Smidt, 2021).

Some researchers suggest that a strong focus on children’s self-education and reservation of preschool teachers can be an issue because concerns have been expressed that the children who are most in need of support, targeted stimulation, and active assistance could be disadvantaged (Grell, 2010). The interpretation of the negative association between co-construction beliefs and children’s task orientation is much less clear as one would expect the opposite pattern of results (Winsler and Carlton, 2003; Bodrova and Leong, 2018). An explanation may lie in a lack of clarity on the part of the preschool teachers on how co-construction can be implemented to support children’s task orientation. Some research has revealed a high overlap (r = 0.59) between beliefs on co-construction and self-education (Schmidt and Smidt, 2021), whereas other findings suggest that preschool teachers do not rely on co-construction but on self-education (Mackowiak et al., 2022). These somewhat “hazy” findings might indicate a conceptual ambiguity of co-construction and self-education that preschool teachers probably face. Findings showing that cognitively stimulating learning support for children by preschool teachers is hardly implemented in everyday preschool life (Schmidt and Smidt, 2021) can also be interpreted in this direction.

We did not find any associations between preschool teachers’ work engagement and self-efficacy and the quality of interactions in terms of children’s task orientation. These patterns of findings deviate from previous research, which found positive associations between preschool teachers’ work engagement and interaction quality (Penttinen et al., 2020; Soininen et al., 2023) and preschool teachers’ self-efficacy and interaction quality (Jennings, 2015; Hu et al., 2021; Wolstein et al., 2021). However, in all of these studies, interaction quality was examined with the Classroom Assessment Scoring System (CLASS; Pianta et al., 2008), which also comprises interaction quality in terms of time on tasks (classroom organization) but focuses much more strongly on the preschool teacher. Thus, reasons for the lack of findings may lie in the measurement of the inCLASS.

Regarding conflict interactions, the only statistically significant association was found with preschool teachers in terms of work engagement. There were no statistically significant findings regarding preschool teachers’ beliefs and self-efficacy. The most obvious explanation for the extensive lack of associations is probably the small variance of the conflict interaction domain (see Table 1). There is simply very little variance that can be explained at all. Indeed, the inCLASS factor conflict interactions is viewed critically due to the lack of variance and other measurement problems (von Suchodoletz et al., 2015; Slot and Bleses, 2018; Smidt and Embacher, 2021). There is one exception: we found a positive association between work engagement and conflict interactions (higher extent of conflicts). One possible explanation could involve “absorption,” which is a characteristic of work engagement and means that people are immersed in their work (Bakker, 2011). The findings from our study could imply that preschool teachers with higher work engagement are probably less aware of the (few) conflictive interactions of the children with their peers, with themselves, and in terms of difficulties in behavior control as they are immersed and engaged in their work and different tasks. This should be explored in more detail in future studies.

Although educational beliefs of teachers have been considered to be important for the formation of teacher–student relationships (Myers and Pianta, 2008), empirical research is sparse and provides no clear evidence for associations between preschool teachers’ educational beliefs and relationship quality (Mashburn et al., 2006; Hamre et al., 2008). The present study points in the same direction since preschool teachers’ professional competencies in terms of beliefs on self-education, co-construction, and instruction did not predict teacher–child closeness. One explanation for the lack of findings in the present study and in previous research may lie in the nature of educational beliefs, which have been referred to as a “messy” construct (Fives and Buehl, 2012, p. 471) because they are difficult to define and conceptualize.

Preschool teachers’ work engagement was positively associated with teacher–child closeness. Although referring to interactions instead of relations, previous research indicating positive associations between work engagement and interaction quality between preschool teachers and children (Penttinen et al., 2020; Soininen et al., 2023) could probably be tentatively connected to the findings of the present study. Following the job demands–resources model (e.g., Bakker, 2011; Bakker et al., 2014; Bakker and Demerouti, 2017), associations can also be theoretically expected as it is assumed that higher work engagement corresponds with positive emotions such as joy and enthusiasm, which may positively affect relationships with children. However, empirical findings on relations between preschool teachers’ work engagement and teacher–child closeness seem to be sparse. In addition, preschool teachers’ self-efficacy had a significant positive association with teacher–child closeness (only p < 0.10). This finding is generally in line with previous research (Mashburn et al., 2006; Hamre et al., 2008) and underpins the still quite limited empirical findings on the importance of preschool teachers’ self-efficacy for the quality of teacher–child relationships (e.g., Chen and Phillips, 2018).

A higher endorsement of educational beliefs on instruction corresponded with a perception of preschool teacher–child relationships as more conflicted. A higher endorsement of beliefs on co-construction corresponded with a perception of these relationships as less conflicted. One explanation could be that resentment and resistance among children could arise from an endorsement of instruction, which corresponds with teacher-directed activities in which the child has little opportunity to have a voice (Schmidt and Smidt, 2021). This resentment and resistance might be manifested in preschool teachers’ perceptions of relationships with children as more conflicted. If this interpretation is true, the effect of beliefs on co-construction also seems reasonable because in educational practice, children might be more actively involved, they might have more opportunities to have a voice, and preschool teachers might be in a dialogue with the children and provide them with targeted assistance (“scaffolding”) (Bodrova and Leong, 2018; Schmidt and Smidt, 2021).

There were no significant associations between preschool teachers’ work engagement and teacher–child conflict. This null finding seems surprising because according to the job demands–resources model (e.g., Bakker, 2011; Bakker et al., 2014; Bakker and Demerouti, 2017), one would expect that high work engagement, which is related to positive emotions such as joy and enthusiasm, would lead to less conflictive relationships between preschool teachers and children, at least in the perception of the preschool teacher. In any case, this finding cannot be readily explained and would need further investigation in additional studies. A different picture emerges with regard to preschool teachers’ self-efficacy, which was negatively related to teacher–child conflict (i.e., higher self-efficacy led to less conflicted relationships between preschool teachers and children in the perception of preschool teachers). This finding fits with previous research where the same pattern of results was reported (Hamre et al., 2008). It also adds additional evidence to the research on preschool teachers’ self-efficacy for the quality of teacher–child relationships (e.g., Chen and Phillips, 2018).