Reciprocal associations among teacher-child interactions, teachers' work engagement, and children's social competence

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Reciprocal associations among teacher-child interactions, teachers' work engagement, and children's social competence © 2022 The Authors. Published by Elsevier Inc. Published version Soininen, Viola; Pakarinen, Eija; Lerkkanen, Marja-Kristiina Soininen, V., Pakarinen, E., & Lerkkanen, M.-K. (2023). Reciprocal associations among teacherchild interactions, teachers' work engagement, and children's social competence. Journal of Applied Developmental Psychology, 85, Article 101508. https://doi.org/10.1016/j.appdev.2022.101508 2023 Journal of Applied Developmental Psychology 85 (2023) 101508 Available online 4 January 2023 0193-3973/© 2022 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Reciprocal associations among teacher–child interactions, teachers’ work engagement, and children’s social competence Viola Soininen * , Eija Pakarinen , Marja-Kristiina Lerkkanen University of Jyv¨ askyl¨ a, 40014 University of Jyv¨ askyl¨ a, P. O. Box 35, Finland ARTICLE INFO Keywords: Teacher–child interaction Work engagement Social competence First grade ABSTRACT This study aimed at examining reciprocal associations among teacher–child interactions, teachers’ work engagement, and children’s social competence across grade 1. The participants were 51 Finnish teachers and 815 children. The quality of teacher–child interactions (emotional support, classroom organization, and instructional support) was assessed with the Classroom Assessment Scoring System. In addition, teachers reported their work engagement and rated children’s social competence (prosocial and antisocial behaviors). The results of the multilevel modeling indicated that high-quality instructional support was associated with more prosocial and less antisocial behavior. Prosocial behavior was associated with higher-quality instructional support. Work engagement was associated with less antisocial behavior, and prosocial behavior with higher work engagement. Finally, work engagement was associated with higher-quality teacher–child interactions. The results suggest that teachers can enhance children’s social competence with high-quality instructional support. Moreover, children’s prosocial behavior can be reflected in teacher–child interactions and in teachers’ work engagement. In their prosocial classroom model, Jennings and Greenberg (2009) proposed that teachers’ social and emotional competence and wellbeing enables them to create healthy teacher–child relationships, conduct effective classroom management, and implement socialemotional learning in the classroom. This way, they noted, a teacher creates a positive classroom climate that enhances children’s social, emotional, and academic outcomes. Thus, according to their prosocial classroom model, teachers’ occupational well-being is reflected in their interactions with children, which again is reflected in children’s outcomes, including their social competence. Based on this model, it is possible that teachers’ work engagement as part of their occupational well-being is associated with the quality of teacher–child interactions and the average social competence of the children in the classroom. However, earlier research on the associations between teachers’ work engagement and the quality of teacher–child interactions has been scarce and cross-sectional, and thus has not been able to examine directions of associations (e.g., Penttinen, Pakarinen, von Suchodoletz, & Lerkkanen, 2020). Directions of the associations between the quality of teacher–child interactions and children’s social competence also remain somewhat unclear because most of the earlier studies have only examined if the quality of teacher-child interactions enhances children’s social competence (e.g., Siekkinen et al., 2013). Recent research indicates that the associations might be reciprocal – at least between emotional support and children’s empathy (Pakarinen, Lerkkanen, & von Suchodoletz, 2020). Furthermore, although earlier research has shown that teachers’ stress is associated with children’s lower social competence (Herman, Hickmon-Rosa, & Reinke, 2018; Siekkinen et al., 2013), to date, it is not yet clear whether work engagement as a positive aspect of occupational well-being plays a role in children’s social competence. Altogether, there is a limited amount of knowledge of the possible associations among the three domains of teacher–child interactions (emotional support, classroom organization, and instructional support), teachers’ work engagement, and average social competence (prosocial and antisocial behavior) of children in the classrooms. To better understand how to support the development of children’s social competence, this study aimed to explore reciprocal associations among the quality of teacher–child interactions, teachers’ work engagement, and children’s social competence over the course of first school year. The quality of teacher–child interactions In the present study, the quality of teacher–child interactions was conceptualized with the Teaching through Interactions (TTI) framework developed by Hamre et al. (2013). The focus of the TTI framework and, * Corresponding author at: Department of Teacher education, University of Jyv¨ askyl¨ a, P.O. Box 35, 40014 University of Jyv¨ askyl¨ a, Finland. E-mail address: [email protected] (V. Soininen). Contents lists available at ScienceDirect Journal of Applied Developmental Psychology journal homepage: www.elsevier.com/locate/jappdp https://doi.org/10.1016/j.appdev.2022.101508 Received 19 January 2022; Received in revised form 12 December 2022; Accepted 28 December 2022 Journal of Applied Developmental Psychology 85 (2023) 101508 2 hence, the present study, is at teacher-child interactions at the classroom-level, not at dyadic teacher-child relationships (i.e., relationships between teacher and individual children). The TTI framework is based on Bronfenbrenner & Morris’s (2006) bioecological model, which suggests that regular interactions that children have, for example, with parents and teachers, are a central driver for their development (Hamre et al., 2013). Indeed, research has indicated that the quality of teacher-child interactions in the classroom is a critical indicator of the quality of education (Pianta, Downer, & Hamre, 2016, see also Burchinal, Vandergrift, Pianta, & Mashburn, 2010; Hu, Fan, Wu, LoCasale-Crouch, & Song, 2019). In TTI, interactions in the classroom are conceptualized under three domains: emotional support, classroom organization, and instructional support. The quality of these three domains can be assessed with an observational tool, the Classroom Assessment Scoring System (CLASS; Pianta, La Paro, & Hamre, 2008). The domain of emotional support has its roots in attachment theory (Ainsworth, Belehar, Waters, & Wall, 1978; Bowlby, 1969) and selfdetermination theory (Ryan & Deci, 2000; Skinner & Belmont, 1993), which underline the importance of sensitive and responsive interactions in supporting learning and motivation. According to the TTI framework, emotionally supportive teachers can create and maintain a positive climate in their classroom, they are sensitive to children’s individual needs, and they consider children’s perspectives. In these classrooms, children enjoy being with each other and with the teachers; they approach the teacher freely and express their ideas. The second domain, classroom organization, is grounded on earlier classroom management (see Emmer & Stough, 2001) and self-regulation skill research (e.g., Paris & Paris, 2001; Raver, 2004; Rimm-Kaufman et al., 2002). Teachers with high-quality classroom organization manage children’s behavior and time efficiently and make the lessons interesting for the children. In well-organized classrooms, children behave well and are interested and involved in the activities and learning. The last domain, instructional support, is based on studies in cognitive and language development and in how adults can support this development with, for example, scaffolding (e.g., Catts, Fey, Zhang, & Tomblin, 1999; Skibbe, Behnke, & Justice, 2004). In instructionally supportive classrooms, teachers support children’s learning in several ways: they give children specific feedback that is targeted on enhancing the learning, they use versatile language to support language development, and they ask open-ended questions such as “why” and “how” to encourage conversation and to support children’s deeper understanding of concepts rather than memorizing facts. In these classrooms, children actively take part in conversations, and activities are related to their own experiences and lives. Teachers’ work engagement and the quality of teacher–child interactions Teachers’ occupational well-being has received growing interest in the field of educational research during the last decade (see Cumming, 2017). Research in occupational well-being has first focused on challenges in well-being such as stress and burnout whereas recent research has also noticed the positive aspects of well-being such as work engagement and job satisfaction (see e.g., Sepp¨ al¨ a et al., 2009). However, the relation between teachers’ occupational well-being and the quality of teacher–child interactions (e.g., Friedman-Krauss, Raver, Morris, & Jones, 2014; Hoglund, Klingle, & Hosan, 2015) and children’s social competence (e.g., Herman et al., 2018; Siekkinen et al., 2013) has mainly been examined in terms of stress and burnout. Teachers’ stress refers to negative emotions such as anxiety, frustration, and tension (Kyriacou, 2001) and its associations with the quality of teacher-child interactions (e.g., Jennings, 2015; Penttinen et al., 2020) and child outcomes (e.g., Siekkinen et al., 2013) have been negative. Since work engagement refers to positive emotions (see below), we expect that it has positive associations with the quality of teacher-child interactions and the average social competence of the children in the classroom. Work engagement is characterized by three components: vigor, dedication, and absorption (Schaufeli, Salanova, Gonz´ alez-rom´ a, & Bakker, 2002). Teachers with a high level of vigor have a lot of energy and are ready to invest in their work. Highly dedicated teachers are proud of their work, consider it meaningful, and feel enthusiasm and inspiration while working. Finally, teachers with a high absorption are highly concentrated while working and might feel that “time flies” while working. Compared with those in other professions, teachers usually experience relatively high work engagement (Hakanen, Ropponen, Schaufeli, & De Witte, 2019). High work engagement can be considered favorable, as it is related to higher job satisfaction (Høigaard, Giske, & Sundsli, 2012; Klassen et al., 2012), lower levels of burnout (Høigaard et al., 2012), and lower intention to leave the profession (Høigaard et al., 2012; Skaalvik & Skaalvik, 2016). Bakker and Demerouti (2008) introduced a model of work engagement, in which they proposed that people’s performance at work is predicted by their work engagement, which is predicted by their resources. The first studies that examined the relation between work engagement and teaching performance reported teachers’ work engagement as being positively associated with child-rated quality of instruction in secondary schools (Klusmann, Kunter, Trautwein, Lüdtke, & Baumert, 2008) and teachers’ self-rated job performance among beginning teachers (Bakker & Bal, 2010). However, to date, there has been a limited amount of research conducted concerning the associations between teachers’ work engagement and the observed quality of teacher–child interactions. In kindergarten classrooms, Penttinen et al. (2020) reported that teachers experiencing a high level of work engagement also provided high-quality instructional support. However, teachers’ work engagement was not associated with the quality of emotional support or classroom organization. To the best of our knowledge, no research examining the relation between work engagement and the observed quality of teacher–child interactions in elementary school classrooms has been conducted. Moreover, the previous research conducted in kindergarten classrooms has been cross-sectional; thus, it is not clear if high-quality interactions with children enhance teachers’ occupational well-being or vice versa. Therefore, in the present study, we examined also if the quality of teacher–child interactions is associated with teachers’ level of work engagement. Children’s social competence and the quality of teacher–child interactions Social competence can be defined as an ability to use a variety of skills in interactions with others to achieve meaningful goals in social groups (Waters & Sroufe, 1983). Sheridan and Walker (1999), and Merrell and Gimpel (1998) have suggested that these skills can be conceptualized under two separate aspects. The first aspect reflects the ability to learn context-appropriate social skills needed in social interaction whereas the second aspect reflects the ability to behave in a socially acceptable manner. When developing the Multisource Assessment of Social Competence Scale, Junttila, Voeten, Kaukiainen, and Vauras (2006) named the two aspects as prosocial behavior and antisocial behavior. The first aspect, prosocial behavior, refers to behaviors that people usually see as desirable and therefore encourage children to display (Junttila et al., 2006). These behaviors include cooperative behaviors (e.g., offering help, starting conversations, participating in group activities) and behaviors that show empathy (e.g., sensitive and accepting behaviors, showing, and communicating positive emotions). Thus, socially competent children have social skills that can be used to successfully interact with other people (Junttila et al., 2006; Merrell & Gimpel, 1998; Sheridan & Walker, 1999). In contrast with prosocial behavior, antisocial behavior refers to behaviors that people usually see as undesirable (Junttila et al., 2006). These behaviors include impulsive behaviors (e.g., tantrums, getting easily irritated) and disruptive behaviors (e.g., teasing, quarreling, acting without thinking). Impulsive children have challenges in V. Soininen et al. Journal of Applied Developmental Psychology 85 (2023) 101508 3 restraining immediate responses, waiting, and planning (Baer & Nietzel, 1991) whereas disruptive children direct their disruptive behaviors at other people, often intentionally to annoy them (Kaplan, Gheen, & Midgley, 2002). Thus, socially competent children can inhibit impulsivity and disruptiveness (i.e., antisocial behavior) and behave in a way that is acceptable to other people (Junttila et al., 2006; Merrell & Gimpel, 1998; Sheridan & Walker, 1999). When defining antisocial behavior, it is important to consider also externalizing behavior which shares some similarities with antisocial behavior. Externalizing behavior refers to variety of behaviors that are against social norms and/or are harmful to other people (Kauten & Barry, 2020). These behaviors include both hyperactive and aggressive behaviors (Hinshaw, 1987; Liu, 2004). Externalizing behavior is sometimes used as a synonym for antisocial behavior (Hinshaw, 1987; Liu, 2004) but as Liu (2004) states, there are distinctions between the two concepts. For example, externalizing behavior includes hyperactive behaviors but not all hyperactive children are antisocial (see Liu, 2004). Moreover, externalizing behavior can include behaviors such as substance use (Kauten & Barry, 2020) or delinquency (Liu, 2004) which are not part of antisocial behavior as defined above. To conclude, children considered as highly socially competent display high levels of prosocial behavior in combination with low levels of antisocial behavior. A child with lower social competence can display either a mixture of prosocial and antisocial behaviors or a lack of prosocial behaviors and a presence of antisocial behaviors. Thus, it is also possible, that children display both high prosocial behavior and high antisocial behavior or low levels of both types of behavior (Junttila et al., 2006). Studies have highlighted the importance of children’s social competence by showing its relationship with several positive outcomes. For example, in kindergarten, a link has been found between children’s social competence and preliteracy skills (Pakarinen, Salminen, Lerkkanen, & von Suchodoletz, 2018), the number of friends, and the level of peer acceptance (Ladd, Birch, & Buhs, 1999). In elementary school, longitudinal studies have reported that children’s social competence predicts their later academic skills (Caprara, Barbaranelli, Pastorelli, Bandura, & Zimbardo, 2000; Legkauskas & Magelinskait˙ eLegkauskien˙ e, 2021), level of anxiety (Junttila, Vauras, Niemi, & Laakkonen, 2012; Legkauskas & Magelinskait˙ e-Legkauskien˙ e, 2021), loneliness (Junttila et al., 2012), and peer preference (Caprara et al., 2000). Furthermore, children with high social competence in kindergarten are more likely to obtain a college degree and a full-time job in early adulthood and less likely to be engaged in criminal activity (Jones, Greenberg, & Crowley, 2015). The role of teacher-child interactions in children’s social competence has been examined rather notably in different educational settings. A meta-analysis by Perlman et al. (2016) showed that in preschool classrooms where the quality of instructional support was high, children’s social competence was also high. However, the quality of emotional support or classroom organization was not related with children’s social competence (Perlman et al., 2016). Similarly, in Finnish kindergarten classrooms where the quality of instructional support was high in fall, children displayed more empathy and less disruptiveness in spring (Siekkinen et al., 2013). In contrast, in a study by Broekhuizen et al. (2016), high-quality emotional support and classroom organization – but not instructional support – in prekindergarten and kindergarten classrooms predicted children’s high social competence in first grade. Moreover, it has been shown in middle school data that high emotional support was associated with more prosocial behavior and high classroom organization with less aggression (Luckner & Pianta, 2011). However, there has been less research examining reciprocal associations or if children’s social competence predicts the quality of teacher-child interactions. As an exception, one recent study reported a reciprocal link between kindergarteners’ empathy and the quality of emotional support indicating that the quality of teacher-child interactions and children’s social competence might be reciprocally associated. If there was high-quality emotional support in the classroom in fall, children displayed more empathy in spring and vice versa: if children displayed more empathy in the classroom in fall, the quality of emotional support was higher in spring (Pakarinen et al., 2020). Furthermore, if there was more impulsivity in the classroom in fall, the quality of emotional support was lower in spring (Pakarinen et al., 2020). To conclude, many studies have shown that high-quality emotional support (Broekhuizen et al., 2016; Pakarinen et al., 2020), classroom organization (Broekhuizen et al., 2016; Luckner & Pianta, 2011), and instructional support (Perlman et al., 2016; Siekkinen et al., 2013) are related with higher average social competence of the children in the classroom. However, only few studies have examined if children’s social competence plays a role in teacher-child interactions and not all domains of teacher-child interactions have yet been examined reciprocally with social competence (see Pakarinen et al., 2020). For this reason, more research is needed to fully understand the dynamics between the quality of teacher-child interactions and average social competence of the children in the classroom. Children’s social competence and teachers’ work engagement Earlier research on the association between teachers’ occupational well-being and children’s social competence has focused on the role of teachers’ stress in children’s social competence. In Finnish kindergarten classrooms, children displayed more impulsivity and disruptiveness, less empathy, and lower cooperation skills if the teacher experienced higher stress (Siekkinen et al., 2013). Similarly, in US kindergarten to 4th grade classrooms, there was more disruptive behavior and less prosocial behavior in the classroom if the teacher experienced high levels of stress and burnout together with low levels of coping (Herman et al., 2018). In turn, research on the associations between teachers’ work engagement and children’s social competence remains limited. However, Jennings and Greenberg’s (2009) prosocial classroom model suggests that teachers’ social and emotional competence and well-being enhance their capability to provide high-quality interactions which support children’s social, emotional, and academic skill development. At the same time, the model indicates that if teachers have challenges to their well-being, it may be difficult for them to respond to children’s individual needs, be proactive, and manage child behavior, which is reflected in child outcomes such as child behavior in the classroom. Thus, based on this model, it is possible, that teachers’ work engagement as one aspect of teachers’ occupational well-being is associated with both the quality of teacher–child interactions and the average social competence of the children in the classroom. Education system in Finland The current study was conducted in Finland, where nine years of comprehensive school begin in the year when the child turns seven years old. Primary school is preceded by a one-year, free of charge, and mandatory kindergarten year, and followed by three years of secondary education in either upper secondary school or vocational school. Primary and secondary education in Finland is also free of charge. Elementary school teachers (grades 1–6) are required to have at least a master’s degree in education, and an average class size in elementary school is 20 children (Organisation for Economic Co-operation and Development, 2018). Class sizes are not nationally regulated. In first grade, one school week consists of a minimum of 20 h of teaching. First grade teaching in Finland is focused on the development of children’s academic, social, and motor skills (Finnish National Agency for Education, 2016). In terms of social competence, development of cooperating skills is important as children learn to work together in groups. The aims of the present study There has been a limited amount of research on the role of teachers’ work engagement in the observed quality of teacher–child interactions V. Soininen et al. Journal of Applied Developmental Psychology 85 (2023) 101508 4 (with the exception of a cross-sectional study by Penttinen et al., 2020). Moreover, previous studies examining the associations between teacher–child interactions and children’s social competence have mainly examined if the quality of teacher–child interactions enhance children’s average social competence in the classrooms (e.g., Siekkinen et al., 2013) and less attention has been given to reciprocal associations (see Pakarinen et al., 2020). Finally, the associations between teachers’ work engagement and children’s social competence remain unstudied. Since little is known about the associations and direction of effect among the variables, this study examined reciprocal associations of the quality of teacher–child interactions, teachers’ work engagement, and children’s social competence between classrooms. More specifically, the following research question was investigated: To what extent are the quality of teacher–child interactions (emotional support, classroom organization, and instructional support), teachers’ level of work engagement, and average social competence (prosocial and antisocial behavior) of the children in the classrooms associated across grade 1? Moreover, the following hypotheses were set: First, based on the model of work engagement by Bakker and Demerouti (2008), we expected that teachers’ level of work engagement is positively associated with subsequent quality of teacher-child interactions in the classrooms (Hypothesis 1). Second, because recent research has suggested that the quality of teacher-child interactions and children’s social competence might be reciprocally associated (see Pakarinen et al., 2020), we assumed that the quality of teacher-child interactions is positively associated with subsequent social competence of the children in the classrooms (Hypothesis 2a) and that average social competence of the children in the classrooms is positively associated with subsequent quality of teacher-child interactions (Hypothesis 2b). Third, although earlier research has not examined the associations between teachers’ work engagement and average social competence of the children in the classroom, based on the prosocial classroom model by Jennings and Greenberg (2009) and earlier research on the associations between teachers’ stress and children’s social competence (Siekkinen et al., 2013), we hypothesized that teachers’ level of work engagement is positively associated with subsequent social competence of the children in the classrooms (Hypothesis 3). Method Participants and procedures Participants of the present study were 51 first grade teachers and 815 children from their classrooms who were participating in a follow-up study (Lerkkanen & Pakarinen, 2016-2022) in the 2017–2018 school year. Participation in the study was voluntary, and teachers and children’s guardians gave written consent for their or their child’s participation. The ethical statement from the ethical committee of the university was received prior to commencing the study. Before data collection, all 70 elementary schools in eight municipalities in Central Finland were contacted to ask permission from the principal to conduct the study in the school. All schools provided permission. Next, all 136 first grade teachers in the schools were contacted and invited to participate in the study. Finally, 54 teachers from 36 schools agreed to participate in the study. Number of participating classrooms per school varied between one and four and 88.5% of the children from the classrooms participated in the study based on the guardians’ consents. Participating schools were located in both rural and suburban areas. Because there was missing data from three of the participating 54 classrooms, only 51 classrooms were included in the analyses of the present study. The mean age of the participating teachers was 45 years (SD =8.85, min. 27 years, max. 62 years), and their average work experience at schools was 16.54 years (SD =9.31, min. 0.5 years, max. 39 years). All teachers were Finnish-speaking Caucasians and most of them (94.1%) were female. In the participating classrooms, class sizes varied from seven to 25 (M =19.25, SD =4.40). Children’s (50.1% girls) mean age at the beginning of the school year was 7.19 years (SD =0.34, min. 4.83 years, max. 9.28 years). The highest education of parents varied as follows: no vocational education (2.8%), vocational courses (min. 4 months; 1.2%), vocational school degree (33.1%), college-level training (7%), polytechnic degree (23.2%), university/college degree (22.8%), and licentiate or doctoral degree (4.3%). However, parental education information was available only for 67% of the participating children. Data of the present study were collected twice during the first school year: in fall (2017; Time 1 [T1]) and spring (2018; Time 2 [T2]). Data consisted of video recordings from the classrooms, teacher questionnaires on their work engagement, and teacher ratings on children’s social competence. Video recordings of the classes were collected to assess the quality of teacher–child interactions in the classroom. Video recordings were conducted during one regular school day, and three to four lessons were recorded in each classroom in fall (T1) and again in spring (T2). Similarly, the teacher questionnaires and their ratings on children’s social competence were collected first in fall (T1) and again in spring (T2). Reliability information, that is, Cronbach’s alphas for all measures, is presented in Table 2. Measures The quality of teacher–child interactions The quality of teacher–child interactions was assessed with the Classroom Assessment Scoring System (CLASS K-3: Pianta et al., 2008). Validity of the CLASS in the Finnish data has been reported by Pakarinen et al. (2010). The measure has been widely used to measure the quality of teacher-child interactions in Finland (e.g., Pakarinen et al., 2020; Penttinen et al., 2020; Salminen, Pakarinen, Poikkeus, Laakso, & Lerkkanen, 2022; Siekkinen et al., 2013) and around the world (e.g., Canada: Hoglund et al., 2015; Chile: Leyva et al., 2015; China: Hu et al., 2019). CLASS conceptualizes teacher–child interactions under three domains: emotional support, classroom organization, and instructional support. Each of the domains consists of three to four dimensions, which again are measured with more specific behavior indicators. Indicators, dimensions, and domains are all described in the coding manual (Pianta et al., 2008), which guides the assessment of the quality of interactions. In the present study, approximately seven (M =6.82, SD =1.18, min. 5, max. 9) cycles per classroom in fall and approximately six (M =5.90, SD =0.73, min. 5, max. 8) cycles per classroom in spring were rated by six certified coders. The average length of one cycle was 18.5 min (SD =3.0, min. 10.5., max. 26.5). The mean score of cycles for each domain at one time point was used in the analyses. To assess the inter-rater reliability of the CLASS ratings, 26% of the cycles were double coded, and the adjacent agreements and intraclass correlation coefficients (ICCs) for the ratings were calculated (see Table 1). Adjacent agreement refers to the percentage of ratings that are within one point of each other (Pianta et al., 2008). Developers of the CLASS measure (Pianta et al., 2008) use this measure to indicate sufficient inter-rater reliability of the ratings. In the present study, adjacent agreements varied from 77.2% to 100% (see Table 1). Thus, the level of adjacent agreement was acceptable for all ten dimensions at both timepoints (Pianta et al., 2008). ICCs were calculated using two-way random effects model with absolute agreement (Landers, 2015). For most of the dimensions, ICCs indicated good (0.60–0.74) or excellent (0.75–1.00) reliability (Cicchetti & Sparrow, 1981; see Table 1). As exceptions, the level of agreement was fair for the ratings of Regard for student perspectives in the spring data and poor for spring ratings of Negative climate in the spring data. As ICC as a measure is subject to a variety of statistical assumptions (e.g., normality, stable variance), low ICC for negative climate might be caused by the nonnormality of the distribution and low variance in the ratings of this dimension. As shown in Table 1, adjacent agreement for the ratings of negative climate in spring was very and over 90% of the ratings were within one point of each other. V. Soininen et al. Journal of Applied Developmental Psychology 85 (2023) 101508 5 Teachers’ work engagement Teachers’ work engagement was measured with the Utrecht Work Engagement Scale (UWES; Schaufeli et al., 2002). Validity of the UWES measure in the Finnish data has been reported by Sepp¨ al¨ a et al. (2009). The measure has been widely used in Finland (e.g., Hakanen, Bakker, & Schafeli, 2006; Penttinen et al., 2020; Vuorio, Suominen, Kautiainen, & Korhonen, 2019) and around the world (e.g., European countries: Hakanen et al., 2019; India: Kataria, Garg, & Rastogi, 2013; Serbia: Petrovi´ c, Vukeli´ c, & ˇ Cizmi´ c, 2017). UWES consists of nine items that the teachers answered on a scale from 1 (never) to 7 (daily). These items fall under three domains: vigor (e.g., “At my work, I feel bursting with energy”), dedication (e.g., “I am enthusiastic about my job”), and absorption (e.g., “I get carried away when I am working”). The mean score of the nine items was used in the analysis. Children’s social competence Teachers rated children’s social competence using the Multisource Assessment of Children’s Social Competence Scale (MASCS; Junttila et al., 2006). Validity of the MASCS measure in the Finnish data has been reported by Junttila et al. (2006). The measure has been recently used to measure children’s social competence for example in studies by Panula, Junttila, Aromaa, Rautava, and R¨ aih¨ a (2020), Pakarinen et al. (2020), and Salminen et al. (2022). MASCS consists of 15 items and four subscales: cooperating skills (5 items, e.g., “offers help to other children”), empathy (3 items, e.g., “is sensitive to the feelings of others”), impulsivity (3 items, e.g., “is easily irritated”), and disruptiveness (4 items, e. g., “bothers and annoys other children”). All 15 items were rated by teachers on a scale from 1 (never) to 4 (very frequently). Of the subscales, cooperating skills and empathy reflect prosocial behavior, whereas impulsivity and disruptiveness reflect antisocial behavior. Mean scores for prosocial behavior and antisocial behavior were used in the analysis. Control variables Teachers reported their work experience and class size on the questionnaire. Work experience and class size were controlled for in the analyses because earlier research has indicated that these teacher and classroom characteristics are related with the quality of teacher-child interactions (e.g., Friedman-Krauss et al., 2014; Slot, Lerkkanen, & Leseman, 2015). Children’s gender was controlled for as previous studies have shown that girls’ social competence is usually higher than boys’ (e.g., Junttila et al., 2006; Pakarinen et al., 2018; Siekkinen et al., 2013). Data analyses First, as preliminary analysis, correlations among the variables were conducted with Mplus version 8 (Muth´ en & Muth´ en, 1998-2017) (see Table 2). Moreover, to investigate if there were differences among classrooms in children’s social competence or gender, intraclass correlation coefficients (ICCs; see Table 2) were calculated. Third, to examine to what extent are quality of teacher-child interactions, teachers’ level of work engagement, and children’s average social competence in classrooms associated across grade 1 fall (T1) and grade 1 spring (T2), multilevel models were conducted with Mplus. In the models, the social competence T1 scores were group-mean centered at the within-level. At the between-level, social competence at T1 was used as a classroommean averaged variable (cluster-mean). Multilevel modeling is an ideal analysis strategy for nested data because it enables the variance in the observed variables to be divided into the variation caused by the membership in a certain classroom (between-level variation) and variation due to individual children’s differences, after taking into account classroom membership (within-level variation) (Hox, 2010). At the between-level of the models (see schematic representation of the multilevel models, Fig. 1), stability and cross-lagged paths were estimated between the quality of teacher–child interactions, teacher’s level of work engagement, and children’s average social competence in the classroom at T1 and T2. All study variables were allowed to correlate with each other. Moreover, at the between-level of the models, teachers’ work experience and class size were controlled for. However, these two between-level control variables were excluded from the final models due to model parsimony. Excluding the between-level control variables from the final models did not influence the results. At the individual level of the models (within-level), group-mean centered social competence at T1 predicted social competence at T2. Social competence at T1 was predicted by gender. Multilevel-models were conducted separately for the three domains of teacher–child interactions (i.e., emotional support, classroom organization, and instructional support) and two domains of children’s social competence (i.e., prosocial and antisocial behavior) because both the three CLASS domains and two MASCS domains were highly correlated. Consequently, six models were conducted (Figs. 2–7). Goodness-of-fit for the models was evaluated with a χ 2 test, comparative fit index (CFI), Tucker Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). Model fits for the investigated models are presented in Table 3. Results Descriptive results, correlations among the study variables, and intraclass correlation coefficients (ICCs) are shown in Table 2. ICCs were estimated using the classroom as a cluster variable. ICCs were statistically significant for both dimensions of children’s social competence (i. e., prosocial, and antisocial behavior) at both time points (i.e., first grade fall and first grade spring), indicating that there was statistically significant variation in children’s social competence among classrooms. ICC for children’s gender was not statistically significant, indicating that there was no statistically significant variation in children’s gender among the classrooms. Consequently, children’s gender was treated as a within-level variable (variation only within classrooms) in further analysis, whereas children’s social competence was treated as both a withinand between-level variable. Table 1 Inter-rater reliabilities of the classroom assessment scoring system ratings. PC NC TS RSP BM PR ILF CD QF LM First grade fall Adjacent agreement 97.4% 100% 94.7% 93.4% 98.7% 97.4% 96.1% 96.1% 96.1% 97.4% ICC 0.82 0.88 0.84 0.86 0.86 0.81 0.78 0.84 0.85 0.89 First grade spring . Adjacent agreement 88.3% 93.2% 88.3% 77.7% 93.2% 94.2% 93.2% 87.4% 92.2% 91.3% ICC 0.70 0.24 0.67 0.54 0.71 0.65 0.73 0.72 0.69 0.70 Note: PC =Positive climate; NC =Negative climate; TS =Teacher sensitivity; RSP =Regard for student perspectives; BM =Behavior management; PR =Productivity; ILF =Instructional learning formats; CD =Concept development; QF =Quality of feedback; LM =Language modeling; ICC =Intraclass correlation coefficient. V. Soininen et al. Journal of Applied Developmental Psychology 85 (2023) 101508 6 The quality of emotional support The results first showed that at the between-level of the model 1, quality of emotional support, teachers’ work engagement, and children’s prosocial behavior were associated with their preceding values (Fig. 2: Model 1). Moreover, average prosocial behavior of the children in the classroom in fall was marginally significantly associated with the quality of emotional support, and statistically significantly associated with the level of teachers’ work engagement in spring. Thus, if children displayed more prosocial behavior in the classroom at the beginning of grade 1, there was a higher quality of emotional support in the classroom at the end of grade 1, although the association was rather small. Furthermore, if there was more prosocial behavior in the classroom in fall, teachers reported higher levels of work engagement in spring. At the within-level of the model, children’s prosocial behavior at the beginning of grade 1 was positively associated with their prosocial behavior at the end of grade 1, and girls displayed more prosocial behavior than boys. Results further showed that at the between-level of model 2, average antisocial behavior of the children in the classroom was stable across grade 1 (Fig. 3: Model 2). Moreover, teachers’ work engagement was associated with the subsequent quality of emotional support; if teachers reported higher work engagement in fall, the observed quality of emotional support in the classroom was higher in spring. At the withinlevel of the model, children’s antisocial behavior in fall was associated with their antisocial behavior in spring. Moreover, boys displayed more antisocial behavior than girls. The quality of classroom organization Results showed that at the between-level of model 3, classroom organization was stable across grade 1 (Fig. 4: Model 3). Moreover, teachers’ work engagement was positively associated with subsequent quality of classroom organization. Thus, if teachers reported higher work engagement in the fall, the quality of classroom organization in the classroom was higher in the spring. Furthermore, classroom organization was positively related to subsequent prosocial behavior of the children in the classroom, albeit marginally significantly. Thus, if the quality of classroom organization in the classroom was higher in fall, there was on average more prosocial behavior in the classroom in spring, although the association was rather small. The results further showed that teachers’ work engagement was negatively associated with subsequent antisocial behavior of the children in the classroom (Fig. 5: Model 4). Thus, if teachers reported higher work engagement in the fall, there was on average less antisocial behavior in the classroom in spring. The quality of instructional support Instructional support showed stability across the school year (Fig. 6: Model 5). Moreover, there was a reciprocal association between the quality of instructional support and children’s prosocial behavior; First, instructional support was positively and statistically significantly associated with subsequent prosocial behavior of the children in the classroom. Thus, in classrooms where the quality of instructional support was higher in fall, children displayed on average more prosocial behavior in spring. Second, average prosocial behavior of the children in the classroom in fall was positively associated with subsequent quality of instructional support in spring. Thus, in classrooms where children displayed on average more prosocial behavior in the fall, the quality of instructional support was higher in spring. Finally, the results showed that the quality of instructional support was negatively associated with subsequent antisocial behavior of the children in the classroom (Fig. 7: Model 6). Thus, if the quality of instructional support was higher in fall, there was on average less antisocial behavior in the classroom in spring. Moreover, teachers’ work engagement was positively associated with subsequent instructional Table 2 Descriptive statistics and bivariate correlations between study variables (within-level correlations below the diagonal and between-level correlations above the diagonal). 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. First grade fall 1. Emotional sup. T1 1 0.72*** 0.54*** 0.41*** 0.51*** −0.08 0.70*** 0.43*** 0.43*** 0.35** 0.46** −0.17 – 0.08 −0.28* 2. Classroom org. T1 – 1 0.46*** 0.43*** 0.46*** 0.06 0.56*** 0.57*** 0.40*** 0.35** 0.49** −0.02 – 0.03 −0.06 3. Instructional sup. T1 – – 1 0.18 0.29 † −0.25 † 0.26* 0.16 0.45*** 0.10 0.40** −0.41** – 0.20 −0.21 4. Work engag. T1 – – – 1 0.40** −0.22 0.46*** 0.47*** 0.34** 0.64*** 0.29 † −0.30 – 0.19 −0.05 5. Prosocial beh. T1 – – – – 1 −0.10 0.58*** 0.48*** 0.37** 0.51*** 0.65*** −0.17 – 0.23 −0.23 6. Antisocial beh. T1 – – – – −0.32*** 1 −0.13 −0.20 −0.23 † −0.19 0.03 0.82*** – −0.35* −0.06 First grade spring 7. Emotional sup. T2 – – – – – – 1 0.73*** 0.51*** 0.45*** 0.39** −0.27 † – 0.08 −0.25* 8. Classroom org. T2 – – – – – – – 1 0.52*** 0.42*** 0.38** −0.18 – 0.05 −0.08 9. Instructional sup. T2 – – – – – – – – 1 0.23** 0.23 † −0.28* – 0.10 0.14 10. Work engag. T2 – – – – – – – – – 1 0.11 −0.09 – 0.01 −0.24* 11. Prosocial beh. T2 – – – – 0.75*** −0.36*** – – – – 1 −0.11 – 0.08 −0.31 † 12. Antisocial beh. T2 – – – – −0.26*** 0.84*** – – – – −0.42*** 1 – −0.42* 0.03 Control variables 13. Child gender 1 – – – – −0.16** 0.29*** – – – – −0.19*** 0.28*** 1 – – 14.Class size – – – – – – – – – – – – – 1 −0.06 15. Work experience – – – – – – – – – – – – – – 1 Descriptive statistics Mean 5.42 5.62 2.55 6.08 3.04 1.64 5.36 5.68 2.60 6.16 3.02 1.70 1.50 19.25 16.54 Std. deviation 0.46 0.45 0.44 0.75 0.30 0.33 0.40 0.43 0.46 0.83 0.26 0.34 0.50 4.39 9.31 Minimum 4.14 4.15 1.78 4.13 2.13 1.07 4.38 4.44 1.87 2.67 2.28 1.09 1 7 0.5 Maximum 6.40 6.47 3.53 7.00 3.65 2.58 6.20 6.50 3.73 7.00 3.53 2.83 2 25 39 α 0.79 0.74 0.82 0.87 0.89 0.92 0.74 0.75 0.79 0.93 0.88 0.92 – – – ICC 2 – – – – 0.24*** 0.18*** – – – – 0.18*** 0.13*** 0.00 – – Note: T1 =first grade fall, T2 =first grade spring; N =51; ***p <.001, **p <.01, *p <.05; † p <.10; 1 Child gender: 1 =female, 2 =male; 2 Intraclass correlation coefficients. V. Soininen et al. Journal of Applied Developmental Psychology 85 (2023) 101508 7 support; if the teacher reported higher work engagement in fall, the quality of instructional support in the classroom was higher in spring. Discussion The aim of this study was to examine reciprocal associations among the quality of teacher–child interactions, teachers’ level of work engagement, and average social competence of the children in the classroom during first grade. The results showed reciprocal associations between the quality of instructional support and children’s average prosocial behavior in the classrooms. Other associations were in one direction: Teachers’ work engagement was positively associated with subsequent quality of teacher–child interactions (emotional support, classroom organization, and instructional support). Children’s average prosocial behavior in the classroom was positively associated with subsequent quality of emotional support whereas the quality of classroom organization was positively associated with subsequent prosocial behavior in the classroom. However, these two associations were rather small and therefore only marginally significant. Finally, teachers’ high levels of work engagement in fall were associated with less antisocial behavior in the classrooms in spring whereas children’s average prosocial behavior in the classroom in fall was associated with higher work engagement for teachers in spring. Together these results reveal new information on the diverse associations among teacher-child interactions, teachers’ occupational well-being, and children’s prosocial and antisocial behaviors in first grade classrooms. The role of work engagement in the quality of teacher–child interactions As hypothesized (Hypothesis 1) based on the model of work engagement (Bakker & Demerouti, 2008), teachers’ work engagement was positively associated with subsequent quality of teacher–child Fig. 1. Schematic Representation of the Models. Note: T1 =first grade fall; T2 =first grade spring Fig. 2. Model 1: Emotional Support, Work Engagement, and Prosocial Behavior. Note: T1 =first grade fall; T2 =first grade spring; † p<.10, *p<.05, **p<.01, ***p<.001. Non-significant associations were estimated but not displayed due to reasons of model clarity. T1 social competence at the within-level was group-mean centered, and at the between-level, the mean of children’s social competence in the classroom was used (cluster-mean). V. Soininen et al. Journal of Applied Developmental Psychology 85 (2023) 101508 8 interactions. Associations between observed quality of teacher–child interactions and teachers’ work engagement have been studied earlier only cross-sectionally (see Penttinen et al., 2020). Thus, the results of the present study expand our understanding of the role of teachers’ work engagement in the quality of teacher–child interactions by showing that teachers’ work engagement at the beginning of grade 1 is related with the quality of emotional support, classroom organization, and instructional support at the end of grade 1. When teachers experience high work engagement, they have energy to work, they find their job as being meaningful, and it is easy for them to concentrate on their work (Schaufeli et al., 2002). This enjoyment and dedication might be reflected in teachers’ motivation to perform in their job as well as possible (see Bakker & Demerouti, 2008). The present study further suggests that the energy, dedication, and concentration that teachers feel when experiencing work engagement can be seen in all three domains of teacher-child interactions: in the warm and supportive interactions (emotional support), in proactive behavior management and efficient time management (classroom organization), and in highquality instructions, conversations, and feedback (instructional support). Altogether, the results of this study, together with the earlier results (Penttinen et al., 2020) and theoretical models (Bakker & Demerouti, 2008; Jennings & Greenberg, 2009), highlight the importance of teachers’ occupational well-being in the quality of teacher–- child interactions. The interplay between social competence and the quality of teacher–child interactions The results of the present study supported the Hypothesis 2a and 2b by showing that there were reciprocal associations between the quality of instructional support and average prosocial behavior of the children in the classroom. More precisely, the results indicated that in line with Hypothesis 2a, the quality of instructional support was associated with subsequent prosocial and antisocial behaviors in the classroom; if there was high-quality instructional support evident in the classroom in fall, children showed on average more prosocial behavior and less antisocial behavior in the classroom in spring. Moreover, in line with the Hypothesis 2b, in classrooms where children displayed more prosocial behavior in fall, the quality of instructional support was higher in spring. Associations between the remaining two domains of teacher-child interactions and children’s social competence were not reciprocal but the direction of the association was different for emotional support and Fig. 3. Model 2: Emotional Support, Work Engagement and Antisocial Behavior. Note: T1 =first grade fall; T2 =first grade spring; † p<.10, *p<.05, **p<.01, ***p<.001. Non-significant associations were estimated but not displayed due to reasons of model clarity. T1 social competence at the within-level was group-mean centered, and at the between-level, the mean of children’s social competence in the classroom was used (cluster-mean). Fig. 4. Model 3: Classroom Organization, Work Engagement, and Prosocial Behavior. Note: T1 =first grade fall; T2 =first grade spring; † p<.10, *p<.05, **p<.01, ***p<.001. Non-significant associations were estimated but not displayed due to reasons of model clarity. T1 social competence at the within-level was group-mean centered, and at the between-level, the mean of children’s social competence in the classroom was used (cluster-mean). V. Soininen et al.