A typical morning in preschool: Observations of teacher–child interactions in German preschools

Early Childhood Research Quarterly 29 (2014) 509–519

Contents lists available at ScienceDirect

Early Childhood Research Quarterly

A typical morning in preschool: Observations of teacher–child interactions in German preschools Antje von Suchodoletz a,∗ , Anika Fäsche a , Catherine Gunzenhauser a , Bridget K. Hamre b a b

University of Freiburg, Research Group “The Empirics of Education: Economic and Behavioral Perspectives”, Bismarckallee 22, 79085 Freiburg, Germany University of Virginia, Center for Advanced Studies of Teaching and Learning, Old Ivy Way 350, Charlottesville, VA 22903, USA

a r t i c l e

i n f o

Article history: Received 15 March 2013 Received in revised form 20 May 2014 Accepted 31 May 2014 Available online 17 June 2014 Keywords: Teacher–child interactions Preschool quality International child care perspectives

a b s t r a c t The study examined the applicability and generalizability of the Classroom Assessment Scoring System Pre-K (CLASS Pre-K; Pianta, La Paro, & Hamre, 2008) and the associated conceptual Teaching through Interaction framework to understand classroom processes in the German early education system. Three broad domains describe effective teacher–child interactions: Emotional Support, Classroom Organization, and Instructional Support. In the present study, we observed teacher–child interactions in 63 classrooms drawn from 26 different preschools using the CLASS Pre-K. Consistent with research from the United States, CLASS Pre-K scores demonstrated that the quality of teacher–child interactions varied widely. Data indicated that the levels of Emotional Support and Classroom Organization were moderate. In contrast, the level of Instructional Support was rather low and even decreased over the course of the morning. Furthermore, Emotional Support was found to decrease over the day in classrooms with a higher child–teacher ratio. Results have important implications for policy and practice with regard to the quality of care and education in German preschools. © 2014 Elsevier Inc. All rights reserved.

In many industrialized countries, there have been increasing concerns that not all children are well prepared to adjust successfully to the school context when they start formal schooling (Heckman, 2006; Leseman, 2009). The preschool context is, beyond the family, the most important context for learning and development in early childhood. Studies from the United States demonstrate that classroom quality, in particular the social and instructional nature of teacher–child interactions, is an important predictor of children’s academic and social development from preschool through secondary school (Hamre & Pianta, 2005; Mashburn et al., 2008; Rimm-Kaufman, Curby, Grimm, Nathanson, & Brock, 2009). State and national policies in the United States reflect these findings and focus on improving the quality of early childhood care and education. Similarly, state and federal support for early childhood education programs has been increased in Germany (Anders et al., 2012). However, there is still very limited research in Germany on the quality of early childhood education

∗ Corresponding author. Now at Department of Psychology, New York University Abu Dhabi, Saadiyat Island, Abu Dhabi, UAE. Tel.: +971 2 6285363. E-mail addresses: [email protected] (A. von Suchodoletz), [email protected] (A. Fäsche), [email protected] (C. Gunzenhauser), [email protected] (B.K. Hamre). http://dx.doi.org/10.1016/j.ecresq.2014.05.010 0885-2006/© 2014 Elsevier Inc. All rights reserved.

programs and factors that may contribute to quality (Anders et al., 2012; Roßbach, Kluczniok, & Kuger, 2008). In the United States, the Classroom Assessment Scoring System Pre-K (CLASS Pre-K; Pianta, La Paro, & Hamre, 2008) is a widely used observational measure of quality of preschool classrooms. Although a variety of contexts have been observed with the CLASS in the United States (Hamre et al., 2013; Mashburn et al., 2008; Raver et al., 2008), these findings cannot be generalized beyond the United States due to context-specific aspects, such as different foci of early childhood education systems in different countries. Only recently, research has begun to use the CLASS internationally, for example in Finland (Pakarinen et al., 2010) or Portugal (Cadima, Leal, & Burchinal, 2010). However, more research is needed to describe components and determinants of high classroom quality across international contexts and diverse early child care and educational settings. Given the growing ethnic diversity within the U.S. and around the world, such knowledge is needed to enhance the development of effective teaching and students’ learning (Hamre et al., 2013). Moreover, since the CLASS is now being used internationally, psychometric studies will need to examine the construct validity of the measure for use in early educational settings outside of the United States. In this paper, we first test the applicability and generalizability of the CLASS Pre-K and the three-domain structure of teacher–child interactions to German preschool classrooms. Second, we describe classroom quality using CLASS Pre-K ratings.

510

A. von Suchodoletz et al. / Early Childhood Research Quarterly 29 (2014) 509–519

Finally, we examine the extent to which specific teacher, classroom, and program features are related to classroom quality. Early childhood education in Germany In Germany, early child care and education is provided for children under the age of six. Similar to early childhood education programs in the United States programs in Germany are center-based, professionally delivered, and structured according to the ages of the children however without distinguishing between preschool and kindergarten levels (Leseman, 2009; Tietze, Cryer, Bairrão, Palacios, & Wetzel, 1996). Most preschools in Germany are public or church-sponsored and offer half day (with or without lunch) or full day programs (Kultusportal Baden-Württemberg, 2011). The early childhood education system in Germany is regulated at the state not the federal level. Recently core curricula for early childhood education have been recommended that emphasize children’s active role for their development and learning (Kultusportal Baden-Württemberg, 2011). As a result, free-play and child-initiated activities dominate the preschool day (König, 2009). However, in contrast to the core curricula for compulsory formal education, curricula for early childhood education are not yet mandatory. The implementation depends on stateor community-level law or even on the preschool itself, resulting in a situation in which children’s experiences in preschool differ widely. With almost 92% of all three- to six-year-old children attending preschool Bundesministerium für Bildung und Forschung [BMBF] (2010), attendance rates are comparatively high relative to the United States and other countries. However, social factors influence attendance rates (Kreyenfeld, 2004). For example, only a limited proportion of children from immigrant families (Spieß, 1998) and low-income families (Kreyenfeld & Spieß, 2002) attend preschool. Nevertheless, in Germany exists no federal or state level program that is specifically targeted at serving low-income children and their families such as the Head Start Program of the United States Department of Health and Human Services. Research on quality in early childhood education in Germany is still limited, particularly with regard to the quality of teacher–child interactions. Some of the knowledge is based on data from the 1990s (Cryer, Tietze, Burchinal, Leal, & Palacios, 1999; Tietze, Bairrão, Leal, & Roßbach, 1998) and thus might reflect educational practices that are no longer current. More recent studies (Anders et al., 2012; Kuger & Kluczniok, 2008) report variability in educational quality across classrooms. Most research in Germany has relied on the Early Childhood Environment Rating Scale (ECERS; Harms & Clifford, 1980) and the Early Childhood Environment Rating Scale-Revised (ECERS-R; Harms, Clifford, & Cryer, 1998), and has consistently revealed moderate classroom quality scores (Anders et al., 2012; Kuger & Kluczniok, 2008; Tietze et al., 1998). However, recent findings suggest that global classroom quality measured with the ECERS/ECERS-R does not sufficiently capture aspects of the preschool learning environment that predict growth in children’s learning and development (Anders et al., 2012). Teaching and learning within interactions In line with these findings, there are calls to consider additional components of quality (La Paro, Thomason, Lower, Kintner-Duffy, & Cassidy, 2012). Influential developmental theories emphasize that learning and development are influenced by interactions young children have with others, particularly with adults (Bronfenbrenner & Morris, 1998; Vygotsky, 1978). As children spend time in non-parental care arrangements, teacher–child interactions are thought to be “the primary mechanisms by which children learn in

classrooms” (Bronfenbrenner & Morris, 1998, cf. in Curby, RimmKaufman, & Cameron Ponitz, 2009, p. 913). For example, research demonstrates that children who have experienced positive interactions with their teachers are more likely to be motivated to engage in learning activities and demonstrate higher task engagement than those with less positive interactions with their teachers (Howes et al., 2008; Ladd, Birch, & Buhs, 2003; Pianta, Steinberg, & Rollins, 1995). Accordingly, effective teaching that recognizes the interactive and contextual nature of children’s development and learning requires a complex set of skills that needs further conceptual clarification regarding its multiple domains and predictors (Bronfenbrenner & Morris, 1998; Curby, Rimm-Kaufman et al., 2009; Douglas, 2009; La Paro et al., 2012). Hamre et al. (2013) presented an integrated conceptual framework, the Teaching through Interaction framework that is based on findings from studies using the CLASS as an observational measure of teacher–child interactions. The CLASS organizes teacher–child interactions in three domains: Emotional Support, Classroom Organization, and Instructional Support (Pianta, La Paro, et al., 2008). Hamre et al. (2013) analyzed CLASS data from seven U.S. studies (4341 classrooms total) and compared the three-domain structure of the CLASS (i.e., Teaching through Interaction framework) against a two-domain structure (i.e., Social and Instructional Support model) and a onedomain structure (i.e., Effective Teaching model). Confirmatory factor analyses indicated that the three-domain structure best reflects the nature and quality of teacher–child interactions. Each CLASS Pre-K domain consists of several specific dimensions that are thought to reflect relevant aspects of effective teaching and are thus important for promoting students’ learning and development. The Emotional Support domain includes efforts teachers make to create a positive, supportive, safe and predictable environment in which children can take risks to explore the world and to develop autonomy and self-confidence. The dimensions are Positive Climate, Negative Climate, Teacher Sensitivity, and Regard for Student Perspectives. Previous research has shown that teachers’ ability to offer high levels of emotional support is related to higher academic achievement (Hamre & Pianta, 2005), lower levels of student aggression and higher levels of behavioral selfcontrol (Merritt, Wanless, Rimm-Kaufman, Cameron, & Peugh, 2012), higher social competence (Curby et al., 2009a; Mashburn et al., 2008), and higher peer acceptance (Hughes & Kwok, 2006). The Classroom Organization domain is concerned with how teachers organize classroom structures, routines, and activities to help students direct their attention and behavior to learning. It includes the dimensions Behavior Management, Productivity, and Instructional Learning Formats. For example, studies show that students in wellorganized classrooms show higher achievement (Cameron, Connor, Morrison, & Jewkes, 2008; Dobbs-Oates, Kaderavek, Guo, & Justice, 2011) and fewer behavioral problems (Bru, Stephens, & Torsheim, 2002). Finally, the Instructional Support domain has to do with the ways in which teachers take children’s learning to a higher level by connecting and building concepts and facts upon each other. Concept Development, Quality of Feedback, and Language Modeling are the dimensions within the Instructional Support domain. Research indicates that students’ learning gains have been associated with extensive use of scaffolding (Bogner, Raphael, & Pressley, 2002; Wharton-McDonald, Pressley, & Hampston, 1998), high-quality feedback (La Paro, Pianta, & Stuhlman, 2004), and instructional conversations (Hamre & Pianta, 2005; Justice, Mashburn, Hamre, & Pianta, 2008). However, most research with the CLASS has been done with U.S. samples while research in international educational settings is scarce. A few recent studies provide initial evidence that the Teaching through Interaction framework may reflect the structure of teacher–child interactions in other countries (Cadima et al., 2010;

A. von Suchodoletz et al. / Early Childhood Research Quarterly 29 (2014) 509–519

Pakarinen et al., 2010). A validation study indicated that the CLASS Pre-K captures teacher–child interactions in Finnish preschools within the three broad domains of Emotional Support, Classroom Organization, and Instructional Support (Pakarinen et al., 2010). In Germany, there is only one study providing preliminary data on the CLASS (Kammermeyer, Roux, & Stuck, 2011). It used video observations to determine classroom quality in preschool programs for children with special language needs indicating variability in children’s experiences. However, further research is needed to validate the three-domain structure of teacher–child interactions in German classrooms.

511

For the German Federal Ministry of Family Affairs, Senior Citizens, Women and Youth the quality of classroom experiences of children in German preschools is of major concern Bundesministerium für Familie, Senioren, Frauen und Jugend [BMFSF] (2013). A wide-ranging political and public debate concerns the factors influencing classroom quality including for example an ongoing debate on the standards of teacher education and optimal child–teacher ratios. However few studies have examined such structural aspects of quality in German preschool classrooms with even fewer studies focusing on teacher–child interactions and more research is needed to inform policy and practice. The present study

Features of the early education context and classroom quality There is an ongoing debate on the features of programs, classrooms, and teachers that contribute to quality in early childhood education (de Schipper, Riksen-Walraven, & Guerts, 2006; Pianta et al., 2005). Factors that have been found to be influential include the child–teacher ratio, group size and group composition, whether the programs are full day or half day, as well as teachers’ education and experience (Cryer et al., 1999; de Schipper et al., 2006; Pianta et al., 2005; Tietze et al., 1998). The child–teacher ratio stands out as particularly important (NICHD, 2000). Several studies found that lower ratios are related to higher classroom quality (Cryer et al., 1999; NICHD, 2000). Teachers’ interactions with children in smaller groups are more responsive, warmer, and more emotionally supportive (NICHD, 2000). Other studies, however, found no relation between the child–teacher ratio and classroom quality (Pianta et al., 2005). In an experimental study, de Schipper et al. (2006) observed interactions of teachers during episodes of structured play, once with a group of three children and once with a group of five children. For the smaller groups of children, the authors observed higher quality interactions. For example, teachers showed more support and respect for the children’s autonomy in these groups. Furthermore, the children in these groups showed more cooperative behaviors than those in the groups of five. However, the findings were strongest for younger children. The children were 34 months old on average, and it is unclear whether the effect of child–teacher ratio is similar in early educational settings with older children (for example, children between the ages of three and six). Research has also revealed mixed findings with regard to the relation between teachers’ characteristics and classroom quality. Pianta et al. (2005), for example, found that teachers’ education/training and years of experience with 4-year-olds significantly predicted classroom quality when assessed with a global measure. However, the authors found no significant relations when assessing the quality of specific aspects of teacher–child interactions. Another study including seven major reports of early childhood education programs in the United States revealed an inconsistent pattern of results with most analyses yielding null findings between teachers’ education and classroom quality (Early et al., 2007). Early et al. (2007) argue that although teacher education is important, delivering stimulating preschool education requires a great deal of skills that teachers might acquire in the preschool setting itself. Accordingly, other studies have found that classroom quality is higher in classrooms with at least moderately experienced teachers (Phillipsen, Burchinal, Howes, & Cryer, 1997). In the past decade, there have been increased political efforts to offer full day programs (Bryant et al., 2002; Riedel, 2005). Along with these efforts, concerns have been raised that the length of the day may influence the quality provided by the program. However, a study by Pianta et al. (2005) revealed no significant effect of the length of the day on classroom quality.

In the present study, we aimed to examine the applicability and generalizability of the Teaching through Interaction framework to understand classroom processes in the German early education context. Using the CLASS Pre-K in live observations, we first investigated whether the three-domain structure of teacher–child interactions sufficiently describes dimensions of classroom processes. In accordance with Hamre et al. (2013), we used confirmatory factor analyses to test several alternative models (i.e., a single domain model of Effective Teaching, a twodomain model of Social and Instructional Support, and the Teaching through Interaction three-domain model) enabling us to determine the structure of classroom interactions in the German data and allowing for a comparison across cultures. Second, we described classroom quality in German preschools using the CLASS Pre-K ratings. We also explored stability or change in classroom quality across a typical morning. Curby et al. (2011) suggest that the accuracy of interpretations of CLASS Pre-K results improves when potential influences such as time of the day are considered. On the basis of previous research (Curby, Grimm, & Pianta, 2010; Curby et al., 2011), we expected to find within-morning variability. Studies reported that as the school day progressed classroom interactions tended to become more negative and teachers more rigid, indicating lower levels of Emotional Support (Curby et al., 2011). Moreover, Curby et al. (2010) observed that changes in Emotional Support were associated by subsequent changes in Classroom Organization. Accordingly, we hypothesized both, Emotional Support and Classroom Organization, to decrease over the morning. One study has found fewer instructional strategies during the initial settlingin period when students arrive in the classroom (Curby et al., 2011). As such, we expected to find Instructional Support to increase after that period. However, we expected only a small increase due to the fact that Instructional Support is found to be generally low, particularly in preschool classrooms where educators seem to be less focused on providing rich instructional interactions (Curby et al., 2010; Hamre et al., 2013; Mashburn et al., 2008). Finally, we extended previous work by investigating whether features of the program (full- or half-day program), classroom (child–teacher ratio), and teacher (years of teaching experience) influence the variability of classroom quality over the morning. By including these variables in the analyses we aimed to determine the extent to which changes in CLASS Pre-K ratings over the morning might be a function of features of the program, the classroom, and/or the teacher (Curby et al., 2011). However given the inconsistent findings in previous work, we considered this part of the analyses as exploratory. Method Sample Teacher–child interactions were observed in 63 classrooms drawn from 26 preschools in a middle-sized city in the southwest

512

A. von Suchodoletz et al. / Early Childhood Research Quarterly 29 (2014) 509–519

of Germany (in the state of Baden-Württemberg). Initially, a letter with information about the primary goals of the study was sent to the principals of 38 local preschools, representing public/private, full-day/half-day, and one-group/multi-group preschools in the area. Principals of 12 preschools declined to participate, in most instances due to a conflict with other current studies in their institution. There were no demographic differences in the number of enrolled children and percentage of children from immigrant and low-income families between the preschools that participated and those that refused to participate. In preschools where principals consented to participate (68% of all contacted), the study procedures were explained in more detail and teachers were selected to participate in the observation on a voluntary basis. In one preschool two out of four teachers chose not to participate in the classroom observations; all other teachers agreed to participate resulting in an overall teachers’ consent rate of 97%. The final sample largely reflected the proportion of public/private, full-day/half-day, and small/large preschools (i.e., preschools with only one classroom and preschools with multiple classrooms) as well as children’s nationality and families’ socio-economic background of the region in Germany where the data was collected. All teachers participating in the study were females. For each classroom, the lead teacher reported on teacher, classroom, and program characteristics. Teachers were on average 36 years old (SD = 9.72) with experience working in early child care and education settings ranging from one year to 34 years (M = 11.70 years, SD = 8.01). Each observed classroom included 21 children on average (minimum 11 children, maximum 28 children) and had an average of 50% boys. Class size (i.e., total number of children enrolled in the class) and the number of teaching assistants in the classroom were used to calculate the child–teacher ratio. The ratio ranged from 2.71 to 21.00 (M = 7.70, SD = 3.21). The vast majority of classrooms served children between the ages of three and six (in Germany, kindergarten) and all were situated in early child care centers. Only one classroom also included children younger than the age of two and served children between one and six years of age. All classrooms were German speaking. With regard to the length of the day (i.e., the facility’s opening hours), we created a binary dummy-coded variable indicating half-day (0 = open in the morning only, 19% of the classrooms) and full-day program (1 = open all day, 81% of the classrooms). Procedure In spring 2012, each classroom was observed four times during a regular morning. The observations began when the children arrived and ended when they started nap time or left for the day. For each of the four cycles, trained observers observed all ongoing classroom activities and interactions while taking notes on indicators and behavioral markers for each of the 10 CLASS Pre-K dimensions. Subsequently, the observers left the classroom and made their coding decision for each dimension in a quiet area in the preschool for approximately 10 min (before beginning the next observation cycle). An observation cycle lasted an average of 18 min (M = 18.21, SD = 1.87). In total, this amounted to 53–81 min of observation per classroom (M = 72.56, SD = 6.19). The classrooms were observed by the first author and three female university students in the university’s bachelor of psychology program, all of whom spoke German as their native language. CLASS Pre-K observation training included a two-day intensive course and a reliability test prior to data collection. All observers met or exceeded the recommended level of reliability (i.e., 80% of codes within and between dimensions within one scale point of agreement; the actual average reliability score of the four observers was 86%, ranging between 80% and 94%). Moreover, during data collection, all observers did one calibration session. The lead observer additionally took part in a regular

calibration organized by the Center for Advanced Studies in Teaching and Learning CASTL at the University of Virginia. Lead teachers were asked to complete a questionnaire on teacher and classroom demographics. Questionnaires were given to the teachers at the end of the observation day and returned by mail. Each teacher received a gift certificate for a local bookstore for their participation. Measures Classroom quality. We used the Classroom Assessment Scoring System Pre-K (CLASS Pre-K; Pianta, La Paro, et al., 2008) to assess the quality of teacher–child interactions in the preschool classrooms. The focus of observation was the classroom as well as the average experience of a child in that classroom. Trained observers rated 10 specific dimensions of teacher–child interactions on a 7-point Likert scale ranging from 1(low) to 7 (high). We determined codes in keeping with the manual by Pianta, La Paro et al. (2008) which provides detailed information on the dimensions, the specific indicators, and observable behavioral markers as well as anchor examples. In 15% of the classrooms, we had two independent observes double code one observation cycle to ensure inter-rater agreement (10 cycles in total). Following Pianta, La Paro, et al. (2008), we considered ratings within one point of each other to reflect an acceptable degree of accuracy. Overall, 86% of the ratings met this criterion. In addition, we examined the intra-class correlations (ICCs) across the three domains to estimate the level of agreement between raters. ICCs reflect the nature of the CLASS data better than kappa values and are more commonly used with CLASS data (Hamre, Hatfield, Pianta, & Jamil, 2014). The ICCs were .82 for Emotional Support, .67 for Classroom Organization, and .73 for Instructional Support. We created the variables for the analyses by averaging the scores for each classroom in each of the 10 dimensions from the multiple observation cycles. We then computed domain mean scores that represent the average quality of teacher–child interactions provided in that classroom. Data analysis approach In order to test whether a nested structure of the data (i.e., classrooms nested in preschools) had to be taken into account for estimation, we calculated intra-class correlations for the three CLASS Pre-K domain scores across and within the four observational cycles using Mplus 7 (Muthén & Muthén, 2012). Further, we computed design effects as functions of the intra-class correlations and the average cluster size (with on average 2.41 classrooms per preschool). The design effects are approximately equal to 1 + (average cluster size − 1) × intraclass correlation, while values > 2 indicate that the clustering in the data needs to be considered during estimation (Muthén & Satorra, 1995). The intra-class correlations of the three CLASS domains ranged between .18 and .42, yielding design effects <1.60. Thus, we conducted the following analyses using a one-level design.1 A test of multivariate skewness (Srivastava’s b1p = 19.97 (10), p < .05) indicated that the data deviated from the normal distribution. We used a Bayesian estimator with non-informative priors to estimate all models in Mplus 7 (Muthén & Muthén, 2012). It is robust to distributional assumptions of the estimated parameters of interest and to relatively small sample sizes and thus provides more trustworthy results than the traditional maximum likelihood estimator (Lee & Song, 2004; Muthén, 2010; Song & Lee,

1 All models were also estimated in a multilevel framework in order to account for clutering of the data (classrooms within preschools) and thus to check whether this changes the results.

A. von Suchodoletz et al. / Early Childhood Research Quarterly 29 (2014) 509–519

2008). The estimation compares an obtained value with a posterior probability distribution of predicted values (Kruschke, 2011). A posterior predictive p-value as an indicator for model fit is considered extreme when it is close to 0 or to 1, indicating that the data have a low probability of occurring under the postulated statistical model (Kacker, Forbes, Kessel, & Sommer, 2008). Additionally, we checked for convergence in order to ensure model fit. The model converged within 10,000 iterations. The trace plots of all parameters showed appropriate mixing. Significant effects were indicated when the conditional confidence intervals of the fixed posterior distribution for the estimates did not include zero. To examine the three-domain structure of teacher–child interactions, we used confirmatory factor analyses to test different models of classroom interactions (i.e., a one-domain model of Effective Teaching, a two-domain model of Social and Instructional Support, and a three-domain model reflecting the Teaching through Interaction framework). The Bayesian approach currently does not provide an economic way to test model fit or to test different models against each other. Thus, we additionally report a maximum-likelihood estimation with restricted standard errors (MLR) and Satorra-Bentler (S.-B.) scaled 2 difference tests to compare the different models of classroom interactions (Satorra, 2000). This allows our results to be compared to previous studies using this very same approach. We evaluated global model fit with the root mean square error of approximation (RMSEA), comparative fit index (CFI), and standardized root-mean square residual (SRMR). Following the analytic strategy by Hamre et al. (2013) with U.S. CLASS data, we accepted models when RMSEA was <.10. For SRMR, we used a cut-off value of <.08 (Hu & Bentler, 1999; Kelloway, 1998). In the case of CFI, we took values >.95 to indicate a good model fit (Hu & Bentler, 1999; O’Boyle & Williams, 2011). To explore variability in classroom quality over a typical morning and to examine whether teacher, classroom, and program features influence stability or change in classroom quality, we conducted linear growth models. For this purpose, we used the four observation cycles as repeated measurement points for the three CLASS Pre-K domains. First, we conducted a baseline linear growth model with equidistant time points (i.e., four cycles) separately for each domain (i.e., Emotional Support, Classroom Organization, Instructional Support), each of which contained two individual growth parameters: (a) an intercept parameter representing the initial value for the specific CLASS Pre-K domain score and (b) a slope parameter representing the rate of change of the specific CLASS Pre-K domain score over time (i.e., across the observation time). The intercept and slope growth factors were allowed to covary. In a second step, we added the teachers’ years of experience working in early child care settings, child–teacher ratio, and length of the day to the growth models as time-invariant covariates in order to test whether they predicted variability in the CLASS Pre-K domain scores. Results We first examined the structure of the CLASS Pre-K using confirmatory factor analyses. Next, we describe classroom quality in German preschools using CLASS Pre-K ratings. Finally, we report the results of growth curve models analyzing variability in classroom quality over a typical morning. Structure of the CLASS Pre-K CFA models. In accordance with Hamre et al. (2013), we tested three models. Model 1 was a one-domain model with all 10 CLASS Pre-K dimensions loading on a single latent variable that could be conceptualized as general teacher effectiveness (Hamre et al.,

513

2013). Model 2 was a two-domain model with the latent factors social support and instructional support, reflecting two major areas of student adjustment, namely social and academic skills (Hamre et al., 2013). Social support was represented by the dimensions Positive Climate, Negative Climate, Teacher Sensitivity, Regard for Student Perspectives, Behavior Management, Productivity, and Instructional Learning Formats. The remaining dimensions represented instructional support. Model 3 was the three-domain model as conceptualized in the Teaching through Interaction framework comprising the domains Emotional Support, Classroom Organization, and Instructional Support, which has been empirically confirmed in U.S. samples (Hamre et al., 2013; Pianta, La Paro, et al., 2008). The factor loadings indicated the exact same item structure for each CLASS Pre-K dimension when Bayesian and MLR estimation were used (Table 1). In terms of relative fit, the Satorra-Bentler scaled 2 difference tests comparing models 1 and 2 and 2 and 3, respectively, became significant (model 1 vs. model 2: 2 (1) = 23.56, p < .001; model 2 vs. model 3: 2 (2) = 9.59, p < .01). This indicates that model 3 provided the best relative fit to the data. Absolute model fit statistics suggest that all three models had less than adequate fit. Across all of the fit statistics, the model fit of model 3 (the Teaching through Interaction model), however, was better than that of the models 1 and 2 (Table 2). CLASS data from Finland (Pakarinen et al., 2010) suggested that excluding the dimension Negative Climate might improve model fit. A closer inspection of factor loadings revealed that the loading of the Negative Climate dimension on its hypothesized domain of Emotional Support was only .52 (Bayesian estimation), while all other dimensions loaded on their respective domains above .64 (Table 1). Furthermore, correlations with the Negative Climate dimension were lowest within the Emotional Support domain (rs = <.50; Table 3). We therefore repeated the CFA without the dimension Negative Climate. However, excluding this dimension did not significantly improve model fit with the present data. Thus, to allow for comparison of the results with previous studies from the United States where the CLASS Pre-K was developed, we included all 10 dimensions in the subsequent analyses using model 3 with three correlated factors. Internal Consistency: In the present study, Cronbach’s alpha coefficients indicated high internal consistency of the CLASS domain scores, aggregated over the four observation cycles (i.e., over the morning; Table 1). In order to account for the multiple observations of each indicator of the three CLASS domains, we additionally calculated internal composite consistency coefficients for the three domain scores across the four cycles within the confirmatory factor model (Hamre et al., 2013). The composite reliability of Emotional Support was xx = .89, the composite reliability of Classroom Organization was xx = .81, and the composite reliability of Instructional Support was xx = .85, all confirming the high reliability of the CLASS domain scores on a composite level.

Describing classroom quality We investigated the range and distribution of scores in each of the 10 dimensions. Examination of the descriptive statistics for classroom quality suggested substantial variability across classrooms (Table 1). At the domain-level, the overall level of Emotional Support (M = 5.54, SD = 0.65) and Classroom Organization (M = 4.82, SD = 0.87) was moderate. In contrast, the overall level of Instructional Support was rather low (M = 2.47, SD = 0.68). Correlations between dimensions and between domains are presented in Table 3. Some dimensions demonstrated higher convergence among each other than other dimensions. Moreover, correlations among domains were high (rs > .60) indicating that although the three-domain factor model showed better model fit than the

514

A. von Suchodoletz et al. / Early Childhood Research Quarterly 29 (2014) 509–519

Table 1 Descriptive statistics and factor loadings for the CLASS Pre-K Dimensions. M

SD

Emotional support Positive climate Negative climate Teacher sensitivity Regard for student perspective

5.38 6.02 5.04 4.86

0.95 0.27 0.89 0.90

Classroom organization Behavior management Productivity Instructional learning formats

5.30 4.92 4.23

0.99 1.10 0.97

Instructional support Concept development Quality of feedback Language modeling

2.17 2.52 2.73

0.78 0.81 0.76

a b

˛

S.E.a

S.E.b

2–7 6–7 2–7 3–7

.86 (.05) −.52 (.11) .86(.05) .81 (.05)

.86 (.04) −.51 (.18) .86 (.04) .81 (.06)

3–7 2–7 2–6

.64 (.08) .79 (.07) .85 (.05)

.64 (.09) .80 (.04) .86 (.04)

1–4 1–5 1–4

.70 (.08) .90 (.06) .80 (.06)

.70 (.06) .91 (.05) .79 (.07)

Range

.82

.81

.83

Standardized estimates of factor loadings using Bayes estimation. All estimates were significant (indicated by a 95%-Bayesian Credibility Interval not containing zero). Standardized estimates of factor loadings using MLR estimation. All estimates were significant p < .001.

Table 2 Fit statistics for the different CFAs. Model

2

df

CFI

TLI

RMSEA

1. Effective teaching (1-domain structure) 2. Social and instructional support (2-domain structure) 3. Teaching through interactions (3-domain structure)

89.91*** 66.41*** 56.94**

35 34 32

.85 .91 .93

.81 .89 .91

.16 .12 .11

90% CI

SRMR

(.12, .20) (.08, .17) (.06, .16)

.07 .06 .06

Note: *p < .05. ** p < .01. *** p < .001. The pattern of results was replicated in multilevel analyses.

two-domain and the one-domain models there is still a notable overlap among domains.

but then showed a very slight increase during Cycle 4. All covariates accounted for 26% of variance in the intercept (R2 = .26, 95 BCI .057, .560) and 37% of variance in the slope (R2 = .37, 95 BCI .093, .834). Child–teacher ratio was negatively associated with the intercept and slope of Emotional Support, indicating that in classrooms with a higher number of children per teacher Emotional Support was lower at the beginning of the morning and that it significantly decreased as the morning progressed. No significant associations were found between Emotional Support and teachers’ years of experience or length of the school day. In the Classroom Organization baseline model the average intercept was significantly different from zero and remained significant when adding the time-invariant covariates to the model. However, neither the slope nor the predictors were significant in this model. With regard to the Instructional Support baseline model the average intercept and the average slope were significantly different from zero (Table 4). Within teacher–child interactions, Instructional Support began to decrease after Cycle 2 (Fig. 1). Both the average intercept and average slope remained significantly different from zero when adding the time-invariant covariates to the model (Table 5). In addition,

Within-morning variability of classroom quality We then examined variability of classroom quality over the morning and whether features of the teacher, the classroom, and the program predicted stability or change. All growth models fit the data well, indicated by the posterior predictive p-values (Table 4 for the baseline models and Table 5 for the models with time-invariant covariates) and by good convergence reflected in mixed trace plots for each parameter. For the Emotional Support baseline model, the average intercept was significantly different from zero (Table 4) and remained significantly different from zero when adding the time-invariant covariates (i.e., teaching experience, child–teacher ratio, length of the day) to the model. In addition in the extended model the average slope was also significantly different from zero (Table 5). As indicated in Fig. 1 Emotional Support showed a decrease during Cycles 1 and 2, remained relatively stable during Cycle 3, Table 3 Correlations among the CLASS Pre-K domains and dimensions. Classroom organization Emotional support 1. Positive climate 2. Negative climate 3. Teacher sensitivity 4. Regard for student perspectives

.76**

Classroom organization 5. Behavior management 6. Productivity 7. Instructional learning formats

–

Instructional support 8. Concept development 9. Quality of feedback 10. Language modeling

–

*

p < .05.** p < .01.

Instructional support

1

2

3

4

5

6

7

– −.42** .75** .71**

– −.45** −.35**

– .69**

–

.46** .59** .68**

−.51** −.41** −.48**

.55** .51** .68**

.42** .54** .69**

– .58** .45**

– .71**

–

.40** .49** .48**

−.38** −.30* −.21

.44** .56** .62**

.49** .49** .45**

.26* .52** .56**

.48** .62** .54**

8

9

.63**

.71**

– .59** .57** .54**

– .65** .48**

– .73**

A. von Suchodoletz et al. / Early Childhood Research Quarterly 29 (2014) 509–519

515

Table 4 Estimated parameters for the different growth models without predictors of intercept and linear growth. Model

Emotional support Classroom organization Instructional support

Intercept

Slope

B (posterior SD)

One-tailed p-value

95% BCI

B (posterior SD)

One-tailed p-value

95% BCI

5.67 (0.12) 4.98 (0.15) 2.76 (0.14)

.00 .00 .00

5.43, 5.90 4.69, 5.27 2.52, 3.06

−0.06 (0.05) −0.09 (0.05) −0.08 (0.06)

.08 .04 .00

−0.15, 0.03 −0.20, 0.010 −0.31, −0.12

Note: Posterior SD = posterior standard deviation of unstandardized parameter estimate. Posterior predictive p-values were .09 for the Emotional Support model, .43 for the Classroom Organization model, and .43 for the Instructional Support model. The pattern of results was replicated in multilevel analyses.

Table 5 Estimated Parameters for the Different Growth Models with Predictors of Intercept and Linear Growth. Model

R2

1. Emotional support Child–teacher ratio Teaching experience Length of day

One-tailed p-value

95% BCI

B (Posterior SD)

One-tailed p-value

95% BCI

6.47 (0.41) −0.10 (0.04) −0.01 (0.02) 0.08 (0.29)

.00 .00 .30 .40

5.67, 7.26 −0.17, −0.03 −0.04, 0.02 −0.51, 0.67 0.06, 0.56

−0.38 (0.16) 0.04 (0.01) −0.00 (0.01) 0.06 (0.10)

.01 .00 .25 .29

−0.69, −0.08 0.01, 0.07 −0.02, 0.01 −0.16, 0.29 0.09, 0.83

5.58 (0.56) −0.05 (0.05) −0.02 (0.02) 0.04 (0.40)

.00 .15 .17 .45

4.47, 6.69 −0.15, 0.05 −0.06, 0.02 −0.76, 0.83 0.01, 0.29

−0.15 (0.21) 0.01 (0.02) 0.00 (0.01) 0.02 (0.15)

.23 .35 .50 .46

−0.57, 0.25 −0.03, 0.04 −0.02, 0.02 −0.27, 0.31 0.01, 0.24

3.77 (0.49) −0.06 (0.04) −0.01 (0.02) −0.70 (0.35)

.00 .07 .39 .03

2.79, 4.72 −0.15, 0.02 −0.03, 0.04 −1.39, 0.02 0.04, 0.60

−0.57 (0.17) 0.02 (0.01) −0.00 (0.01) 0.35 (0.12)

.00 .10 .25 .00

−0.92, −0.25 −0.01, 0.05 −0.02, 0.01 0.11, 0.59 0.12, 0.87

.09 3. Instructional support Child–teacher ratio Teaching experience Length of day

Slope

B (posterior SD)

.26 2. Classroom organization Child–teacher ratio Teaching experience Length of day

R2

Intercept

.26

.37

.06

.45

Note: Posterior SD = posterior standard deviation of unstandardized parameter estimate. Posterior predictive p-values were .12 for the Emotional Support model, .41 for the Classroom Organization model, and .38 for the Instructional Support model. Length of day is dummy coded (0 = half-day program, 1 = full-day program). The pattern of results was replicated in multilevel analyses.

the decline over the course of the morning was partially explained by the length of the school day. Full day programs started the morning with lower levels of Instructional Support but decreased less over the morning than half day programs. All covariates accounted for 26% of the variance in the intercept (R2 = .26, 95 BCI .044, .598) and 45% of the variance in the slope (R2 = .45, 95 BCI .118, .869). No significant associations were found between Instructional Support and teachers’ years of experience or child–teacher ratio.

Discussion The present study investigated the Teaching through Interaction framework (Hamre et al., 2013) in German preschool classrooms using the CLASS Pre-K (Pianta, La Paro, et al., 2008). Our findings provide preliminary evidence for acceptable psychometric qualities of the CLASS Pre-K as a standardized observational measure of teacher–child interactions in Germany. Specifically, our results provide moderate support for a three-domain structure of teacher–child interactions that is consistent with the Teaching through Interaction framework and previous research on classroom quality in the United States (Curby, Rimm-Kaufman et al., 2009; Hamre et al., 2013; Mashburn et al., 2008; Raver et al., 2008). Overall, the scores demonstrated variability among classrooms. With regard to Emotional Support and Classroom Organization, many classrooms scored in the four to five (or “good”) range. Of particular concern, however, is the finding that children are unlikely to experience high-quality learning support, such as concept development, feedback, or language modeling, as indicated by the low levels of Instructional Support. Furthermore, our findings suggest that classroom quality changes over the course of the morning. Child–teacher ratio was found to be important in explaining changes in teachers’ Emotional Support, whereas length of day explained decreases in teachers’ Instructional Support. No significant associations with quality were found for teachers’ years of experience working in early educational settings. Applicability of the CLASS Pre-K in Germany

Fig. 1. Growth trajectories of the CLASS Pre-K domain scores over the preschool morning. Points represent the mean CLASS Pre-K domain scores per cycle. For the Emotional Support scores, the Negative Climate scores were reversed before calculation.

Examination of the structure of the CLASS Pre-K indicates that the three-domain model (i.e., the Teaching through Interaction

516

A. von Suchodoletz et al. / Early Childhood Research Quarterly 29 (2014) 509–519

framework; Hamre et al., 2013) sufficiently describes classroom quality in German preschool classrooms. Overall, the results of our confirmatory factor analyses suggest that teacher–child interactions in German preschool classrooms are organized in three broad domains, Emotional Support, Classroom Organization, and Instructional Support. We also tested a one-domain model of global teaching effectiveness (i.e., Effective Teaching) and a twodomain model of Social and Instructional Support. Consistent with other studies using the CLASS in different samples (Hamre et al., 2013; Pakarinen et al., 2010; Pianta, La Paro, et al., 2008), our results indicated that the Teaching through Interaction framework provides the best fit for describing teacher–child interactions in German preschool classrooms. High standardized factor loadings of the CLASS Pre-K dimensions suggest construct validity of the domains Emotional Support, Classroom Organization, and Instructional Support. Reliability indicators confirmed internal consistency of the domains, thus providing further support that the Teaching through Interaction framework describes the reality of teacher–child interactions in preschool classrooms. Examination of correlations between the three domains, however, suggests considerable multicollinearity that might limit the interpretation of results using the three-domain approach. There is indeed preliminary evidence for a revised Teaching through Interaction framework (Hamre et al., 2014). Hamre et al. (2014) suggest a bifactor model with general and domain-specific elements of effective teacher–child interactions. Future studies might take this approach when studying teacher–child interactions and how teachers contribute to children’s development. Minor differences in the structure of the CLASS are suggested by recent international studies. In the Finnish data, an exclusion of the dimension Negative Climate from the three-domain model improved model fit (Pakarinen et al., 2010). In our data however, omitting the dimension Negative Climate did not significantly improve model fit statistics. Negative Climate is typically very low with limited variability across studies and samples (Hamre et al., 2013; Pakarinen et al., 2010). Thus the low Negative Climate ratings in our data might rather be a reflection of the instrument and early child care in general than of specific aspects of the German context. However, more research is needed to clarify potential differences in the structure and function of classroom interactions across international early childhood education programs and contexts. Children’s experiences in German preschool classrooms The present findings indicate that children in the observed classrooms experienced moderately high levels of Emotional Support and Classroom Organization. That is, the classrooms were generally warm, positive, and well-organized places for children to be. According to a number of studies, a positive, emotionally supportive climate encourages children’s learning and development (Hamre & Pianta, 2005; Mashburn et al., 2008; Rimm-Kaufman, La Paro, Downer, & Pianta, 2005). In contrast, the provision of Instructional Support was low. For example, preschool teachers provided more general praise and feedback but did not offer encouragement of students’ efforts that expands learning and understanding. Instructional strategies were focused on the outcome of an activity but not on the processes of problem solving and learning. Our results reflect previous findings of less optimal instructional support in preschool classrooms particularly in the United States where the mean levels of this domain are consistently low (Curby et al., 2010; La Paro, Pianta, & Stuhlman, 2004). The findings might also reflect preschool teachers’ emphasis on rather providing a caring environment than rich instructional interactions (Curby et al., 2010). We also found that children’s experiences differ across classrooms, with only a few classrooms providing children with high-quality interactions. These findings are consistent with other

recent studies on classroom quality in the German early education and child care context, which found that less than 10% of classrooms match high-quality standards (Anders et al., 2012; Kluczniok, Sechtig, & Roßbach, 2012; Kuger & Kluczniok, 2008; Wertfein, Kofler, & Müller, 2011). A comparison with studies using the CLASS in Finland (Pakarinen et al., 2010) and the United States (Hamre et al., 2013) indicates some cross-cultural differences in classroom quality. For example, the amount of Emotional Support provided in German preschool classrooms is similar to that in Finnish classrooms and only somewhat higher than that in classrooms in the United States (Hamre et al., 2013; Pakarinen et al., 2010). Classroom Organization is lower in German classrooms than in Finnish classrooms (Pakarinen et al., 2010) but is in the medium range and comparable to U.S. samples (Hamre et al., 2013). In particular, productivity in German preschool classrooms, i.e., the provision of activities, organization of routines and transitions, teachers’ preparation of materials, etc., seems lower. For example, our observation protocols indicate that teachers often prepare materials for the next activity during the day. Also, in many cases, children in our study were observed wandering around the classroom without being engaged by the teacher in new activities. This could be due to the predominance of free-play and child-initiated activities in German preschools (König, 2009). As a consequence, productivity as specified in the CLASS Pre-K might be less observable in everyday routines in German preschool classrooms. With regard to Instructional Support, German preschool classrooms score similarly to classrooms in the United States but considerably lower than classrooms in Finland. Similar to Finland, but in contrast to the United States, there are no direct instructions concerning academic skills or extensive whole-group activities in German preschool classrooms. Both, the German and Finnish early educational systems support children’s learning and skill development by providing time for child-initiated activities based on the basis of the assumption that children learn at their own speed and according to their own interests (for Germany see for example: Kultusportal Baden-Württemberg, 2011; for Finland: Pakarinen et al., 2010). Given these similarities, one could expect equal levels of Instructional Support in these countries. Pakarinen et al. (2010) argued that the higher level of Instructional Support in Finland can be explained by the high standards and level of preschool teacher education. In Germany, preschool teacher education is not a homogeneous training program and does not necessarily require a university degree. Different public and private institutions, colleges, and universities offer training programs organized as vocational training, bachelor’s, or master’s programs. Thus, German preschool teachers may have acquired lower levels of knowledge and skills about interacting with children in cognitively supportive ways than their colleagues in Finland. Particularly within the context of free play, the dominant activity in German preschools, additional education might help teachers to provide and/or facilitate opportunities for learning. Contrary to our hypothesis and to findings from U.S. pre-K samples (Curby et al., 2010), our findings suggest that Instructional Support decreases over the course of the day. It might be that instructional activities decrease naturally through the course of the day as both teachers’ and students’ attentional resources deplete (Curby et al., 2011). It is also likely that changes in the types of activities during the day might be related to the decline in Instructional Support (Curby et al., 2011). However given that Instructional Support was already low at the beginning of the observations, the present findings require further eploration of patterns of variability in classroom quality across the day. As expected, Emotional Support decreased during the first few hours of a preschool day, however, only in classrooms with more children per teacher. These findings are consistent with findings

A. von Suchodoletz et al. / Early Childhood Research Quarterly 29 (2014) 509–519

from other studies suggesting that teachers in these classrooms are less able to meet the individual students’ needs (Curby et al., 2011). Moreover, in our study classrooms with more children per teacher already started with lower levels of Emotional Support than classrooms with fewer children per teacher. The findings suggest that teachers are less able to create positive social and emotional interactions with children as the number of children per teacher increases. Litjens and Taguma (2010) argue that a ratio of four children per teacher allows teachers to effectively interact with each child. Only a few countries in Europe, such as Finland, Iceland, and Sweden meet the recommended low ratio (Moser & Bennett, 2006). Consistent with Moser and Bennett’s (2006) findings for Germany, the child–teacher ratio was found to be higher in the present study. Furthermore, the child–teacher ratio might also indirectly influence the quality of interactions through working conditions, teachers’ job satisfaction, and stress Organization for Economic Cooperation and Development [OECD] (2012). Teachers in classrooms with more children per teacher tend to spend more time in restrictive communication redirecting children’s behavior (Litjens & Taguma, 2010). As a result, teachers may perceive managing their classroom as more stressful and be less able to engage in positive interactions with children. Thus, every additional adult in the classroom may not only improve classroom quality but also the teacher’s well being.

Practical implications The CLASS Pre-K was developed with U.S. samples to describe various aspects of classroom quality in early childhood education settings (Pianta, La Paro, et al., 2008). The present study adds to a growing body of international research providing evidence that the CLASS Pre-K instrument can be used to reliably assess the threedomain structure of effective teacher–child interactions in settings other then the United States. However, the CLASS Pre-K training and reliability testing still relies solely on examples and training videos from the U.S. preschool context. Given the differences in early childhood education programs across countries discussed above, providing additional international examples and training videos might help to improve the CLASS Pre-K training and make it better fit for international use. For example, the training for the German version of the Attachment Story Completion Task successfully relies on German videos as indicated by good psychometric properties of the measure (Gloger-Tippelt, König, Zweyer, & Lahl, 2007). In line with previous national and international studies (Anders et al., 2012; NICHD, 2005; Pianta & Hamre, 2009; Raver et al., 2008; Tietze et al., 1998), experiences of children in preschool do not appear consistent with high-quality standards of teaching and vary in quality. In combination with our findings on withinmorning variability and how teacher and classroom characteristics contribute to within-morning variability the study has important implications for policy and practice concerned with regard to quality of care and education in German preschools. Anders et al. (2012, p. 242) point out that “preschool education may only be an effective means of promoting the development of cognitive skills if it is of high quality.” Intervention studies have shown that preschool teachers can be trained to provide better emotional support and classroom management (Pianta, Mashburn, Downer, Hamre, & Justice, 2008; Raver et al., 2008). For example, the Chicago School Readiness Project (CSRP; Raver et al., 2008) was successful in improving teachers’ ability to provide high-quality standards to their classrooms. However, the study only reported the short-run impact of the intervention in Head-Start Program settings (Raver et al., 2008). Pianta, Mashburn, et al. (2008) found that only more intensive professional development support for teachers lead to

517

improvements in classroom quality. However, the extent to which such interventions can work in the German context still needs to be examined. In Germany, interventions would need to include the promotion of teachers’ abilities to provide high-level Instructional Support within child-initiated activities, given the dominance of free-play time in these settings. Limitations First, the sample reflects the diversity of early education programs in the area where the data was collected but cannot be considered representative of German programs nationally. For example, the sample did not include preschools in large urban centers with concentrated poverty. Previous research has revealed that children living in poverty have a higher risk for developing externalizing and internalizing problems (Li-Grining, Votruba-Drzal, Bachman, & Chase-Landsdale, 2006). Teachers in classrooms serving children from families with low socioeconomic backgrounds may face different challenges in ensuring high levels of classroom quality than teachers in our sample (Li-Grining et al., 2010). Second, RMSEA in our three-domain model was higher than what is typically accepted for good model fit. However, RMSEA is sensitive to sample size (Kenny, 2012; Pakarinen et al., 2010). It will be important to replicate the findings with both larger and more diverse samples recruited in different regions of Germany, including urban and rural areas and centers with concentrated poverty. Moreover, this study does not provide any data on predictive validity of the CLASS Pre-K. This remains a very important unanswered question. In addition, future studies should double-code a higher percentage of observation cycles to ensure inter-rater reliability. The data also revealed potential differences in inter-rater agreement across CLASS domains that is consistent with other studies (Hamre et al., 2014): While the interrater agreement for Emotional Support and Instructional Support was acceptable, the ICC for Classroom Organization was rather low. Another notable limitation is related to our selection of covariates. Although variables included in the present analyses were chosen on the basis of the literature, classroom quality is also influenced by other potentially important factors, such as teachers’ psychosocial characteristics and changes in types of activities during the day. The main reason we did not consider additional covariates was the small sample size, which limited the statistical power of our analyses. Conclusion The present study contributes to the literature by showing that the Teaching through Interaction framework reflects the reality of teacher–child interactions in German preschool classrooms. An evaluation of different models suggests that teacher–child interactions are best described in three broad domains: Emotional Support, Classroom Organization, and Instructional Support. The results provide evidence that the CLASS Pre-K (Pianta, La Paro, et al., 2008) can be used to reliably assess these components of teaching in the German early educational system and thus may be used to evaluate quality in preschool classrooms and to inform intervention policies and practice. Acknowledgements The research reported here was supported by Grant 22682167.4/2010 from the German-Israeli Foundation for Scientific Research and Development (GIF). We wish to thank all preschool teachers and principals who participated in or contributed to this project.

518

A. von Suchodoletz et al. / Early Childhood Research Quarterly 29 (2014) 509–519

References Anders, Y., Roßbach, H.-G., Weinert, S., Ebert, S., Kuger, S., Lehrl, S., et al. (2012). Home and preschool learning environments and their relations to the development of early numeracy skills. Early Childhood Research Quarterly, 27, 231–244. http://dx.doi.org/10.1016/j.ecresq.2011.08.003 Bundesministerium für Bildung und Forschung (BMBF). (2010). (Education in Germany 2010. An indicator-based report and analyses of educational perspectives under demographic changes) Bildung in Deutschland 2010. Ein indikatorengestützter Bericht mit einer Analyse zu Perspektiven des Bildungswesens im demografischen Wandel. Retrieved from http://www.bildungsbericht.de/ daten2010/bb 2010.pdf Bundesministerium für Familie, Senioren, Frauen und Jugend (BMFSF). (2013). (High-quality child care) Gute Kinderbetreuung. Retrieved from http://www.bmfsfj.de/BMFSFJ/Kinder-und-Jugend/kinderbetreuung.html Bogner, K., Raphael, L., & Pressley, M. (2002). How grade 1 teachers motivate literate activity by their students. Scientific Studies of Reading, 6, 135–165. http://dx.doi.org/10.1207/S1532799XSSR0602 02 Bronfenbrenner, U., & Morris, P. A. (1998). The ecology of developmental processes. In W. Damon, & R. M. Lerner (Eds.), Handbook of child psychology: Theoretical models of human development (Vol. 1) (5th edition, Vol. 1, pp. 993–1029). New York, NY: John Wiley and Sons. Bru, E., Stephens, P., & Torsheim, T. (2002). Students’ perceptions of class management and reports of their own misbehavior. Journal of School Psychology, 40, 287–307. http://dx.doi.org/10.1016/S0022-4405(02)00104-8 Bryant, D., Clifford, R. M., Saluja, G., Pianta, R., Early, D., Barbarin, O., et al. (2002). Diversity and directions in state pre-kindergarten programs. Retrieved from http:// fpg.unc.edu/resources/diversity-and-directions-state-pre-kindergartenprograms Cadima, J., Leal, T., & Burchinal, M. (2010). The quality of teacher–student interactions: Associations with first graders’ academic and behavioral outcomes. Journal of School Psychology, 48, 457–482. http://dx.doi.org/ 10.1016/j.jsp.2010.09.001 Cameron, C. E., Connor, C. M., Morrison, F. J., & Jewkes, A. M. (2008). Effects of classroom organization on letter–word reading in first grade. Journal of School Psychology, 46, 173–192. http://dx.doi.org/10.1016/j.jsp.2007.03.002 Cryer, D., Tietze, W., Burchinal, M., Leal, T., & Palacios, J. (1999). Predicting process quality from structural quality in preschool programs: A cross-country comparison. Early Childhood Research Quarterly, 14, 339–361. http://dx.doi.org/10.1016/S0885-2006(99)00017-4 Curby, T. W., LoCasale-Crouch, J., Konold, T. R., Pianta, R. C., Howes, C., Burchinal, M., et al. (2009). The relations of observed pre-K classroom quality profiles to children’s achievement and social competence. Early Education & Development, 20, 346–372. http://dx.doi.org/10.1080/10409280802581284 Curby, T. W., Rimm-Kaufman, S. E., & Cameron Ponitz, C. (2009). Teacher–child interactions and children’s achievement trajectories across kindergarten and first grade. Journal of Educational Psychology, 101, 912–925. http://dx.doi.org/ 10.1037/a0016647 Curby, T. W., Grimm, K., & Pianta, R. C. (2010). Stability and change in early childhood classroom interactions during the first two hours of a day. Early Childhood Research Quarterly, 25, 373–384. http://dx.doi.org/10.1016/j.ecresq.2010.02.004 Curby, T. W., Stuhlmann, M., Grimm, K., Mashburn, A., Chomat-Mooney, L., Downer, et al. (2011). Within-day variability in the quality of classroom interactions during third and fifth grade: Implications for children’s experiences and conducting classroom observations. The Elementary School Journal, 112, 16–36. http://dx.doi.org/10.1086/660682 de Schipper, E. J., Riksen-Walraven, M., & Guerts, S. A. E. (2006). Effects of child–caregiver ratio on the interactions between caregivers and children in child-care centers: An experimental study. Child Development, 77, 861–874. http://dx.doi.org/10.1111/j.1467-8624.2006.00907.x Dobbs-Oates, J., Kaderavek, J. N., Guo, Y., & Justice, L. M. (2011). Effective behavior management in preschool classrooms and children’s task orientation: Enhancing emergent literacy and language development. Early Childhood Research Quarterly, 26, 420–429. http://dx.doi.org/10.1016/j.ecresq.2011.02.003 Douglas, K. (2009). Sharpening our focus in measuring classroom instruction. Educational Researcher, 38, 518–521. http://dx.doi.org/10.3102/0013189X09350881 Early, D. M., Maxwell, K. L., Burchinal, M., Alva, S., Bender, R. H., Bryant, D., et al. (2007). Teachers’ education, classroom quality, and young children’s academic skills: Results from seven studies of preschool programs. Child Development, 78, 558–580. http://dx.doi.org/10.1111/j.1467-8624.2007.01014.x Gloger-Tippelt, G., König, L., Zweyer, K., & Lahl, O. (2007). Bindung und Problemverhalten bei fünf und sechs Jahre alten Kindern [Attachment and conduct problems in 5- and 6-year-old children]. Kindheit und Entwicklung, 16, 209–219. http://dx.doi.org/10.1026/0942-5403.16.4.209 Hamre, B. K., & Pianta, R. C. (2005). Can instructional and emotional support in the first-grade classroom make a difference for children at risk of school failure? Child Development, 76, 949–967. http://dx.doi.org/ 10.1111/j.1467-8624.2005.00889.x Hamre, B. K., Pianta, R. C., Downer, J. T., DeCoster, J., Mashburn, A. J., Jones, S. M., et al. (2013). Teaching through interactions: Testing a developmental framework of teacher effectiveness in over 4000 classrooms. Elementary School Journal, 113, 461–487. http://dx.doi.org/10.1086/669616 Hamre, B. K., Hatfield, B., Pianta, R. C., & Jamil, F. (2014). Evidence for general and domain specific elements of teacher–child interactions: Associations with preschool children’s development. Child Development, 85, 1257–1274. http://dx.doi.org/10.1111/cdev.12184

Harms, T., & Clifford, R. M. (1980). Early childhood environment rating scale. New York, NY: Teachers College Press. Harms, T., Clifford, R. M., & Cryer, D. (1998). Early childhood environment rating scale (Rev. ed.). New York, NY: Teachers College Press. Heckman, J. J. (2006). Skill formation and the economics of investing in disadvantaged children. Science, 312, 1900–1902. http://dx.doi.org/10.1126/ science.1128898 Howes, C., Burchinal, M., Pianta, R. C., Bryant, D., Early, D., Clifford, R., et al. (2008). Ready to learn? Children’s pre-academic achievement in pre-kindergarten programs. Early Childhood Research Quarterly, 23, 27–50. http://dx.doi.org/10.1016/j.ecresq.2007.05.002 Hu, L. T., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. http://dx.doi.org/10.1080/10705519909540118 Hughes, J. N., & Kwok, O. (2006). Classroom engagement mediates the effect of teacher–student support on elementary students’ peer acceptance: A prospective analysis. Journal of School Psychology, 43, 465–480. http://dx.doi.org/ 10.1016/j.jsp.2005.10.001 Justice, L. M., Mashburn, A. J., Hamre, B. K., & Pianta, R. C. (2008). Quality of language and literacy instruction in preschool classrooms serving atrisk pupils. Early Childhood Research Quarterly, 23, 51–68. http://dx.doi.org/ 10.1016/j.ecresq.2007.09.004 Kacker, R. N., Forbes, A., Kessel, R., & Sommer, K.-D. (2008). Bayesian posterior predictive p-value of statistical consistency in interlaboratory evaluations. Metrologia, 45, 512–523. http://dx.doi.org/10.1088/0026-1394/45/5/004 Kammermeyer, G., Roux, S., & Stuck, A. (2011). Additive Sprachförderung in Kindertagesstätten–Welche Sprachfördergruppen sind erfolgreich? [Language promotion in early childhood education–which language promotion groups are successful?]. Zeitschrift für Empirische Pädagogik, 25(4), 423–438. Kelloway, E. K. (1998). Using LISREL for structural equation modeling: A researcher’s guide. Thousand Oaks, CA: Sage. Kenny, D. A. (2012). Structural equation modeling. Retrieved from http:// davidakenny.net/cm/fit.htm Kluczniok, K., Sechtig, J., & Roßbach, H.-G. (2012). Qualität im Kindergarten Wie gut ist das Niveau der Kindestagesbetreuung in Deutschland und wie wird es gemessen? [Quality in the kindergarten. How high is the child care standard in Germany and how is it measured?]. DJI Impulse, 98, 33–36. König, A. (2009). Observed classroom interaction processes between pre-school teachers and children: Results of a video study during free-play in German pre-schools. Educational and Child Psychology, 26(2), 53–65. Kreyenfeld, M. (2004). (Social structure and child care: An analysis of social and economic determinants for utilization of child care facilities (MPIDR Working Paper 2004-009)) Sozialstruktur und Kinderbetreuung: Eine Analyse der sozialen und ökonomischen Determinanten der Nutzung von Kindertageseinrichtungen. Retrieved from http://www.demogr.mpg.de/papers/working/wp-2004-009.pdf Kreyenfeld, M., & Spieß, K. (2002). Die Förderung von Kindertageseinrichtungen – eine Analyse ihrer verteilungspolitischen Konsequenzen. In G. Neubauer, J. Fromme, & A. Engelbrecht (Eds.), Ökonomisierung der Kindheit. Sozialpolitische Entwicklungen und ihre Folgen (pp. 59–76). Opladen, Germany: Leske + Budrich. Kruschke, J. K. (2011). Bayesian assessment of null values via parameter estimation and model comparison. Perspectives on Psychological Science, 6, 299–312. http://dx.doi.org/10.1177/1745691611406925 Kuger, S., & Kluczniok, K. (2008). Prozessqualität im Kindergarten – Konzept, Umsetzung und Befunde [Process quality in the kindergarten – concept, implementation and findings]. Zeitschrift für Erziehungswissenschaft, Sonderheft, 11, 159–178. http://dx.doi.org/10.1007/978-3-531-91452-7 11 Kultusportal Baden-Württemberg. (2011). (Orientation for learning and education in kindergarten and other daycare centers in Baden-Württemberg) Orientierungsplan für Bildung und Erziehung in baden-württembergischen Kindergärten und weiteren Kindertageseinrichtungen. Retrieved from http://www.kultusportal-bw.de/site/ pbs-bw/get/documents/KULTUS. Dachmanda nt/KULTUS/import/pb5start/pdf/ KM KIGA Orientierungsplan 2011.pdf La Paro, K. M., Pianta, R. C., & Stuhlman, M. (2004). The Classroom Assessment Scoring System: Findings from the prekindergarten year. The Elementary School Journal, 104(5), 409–426. Retrieved from http://www.jstor.org/stable/3202821 La Paro, K. M., Thomason, A. C., Lower, J. K., Kintner-Duffy, V. L., & Cassidy, D. J. (2012). Examining the definition and measurement of quality in early childhood education: A review of studies using the ECERS-R from 2003–2010. Early Childhood Research & Practice, 14, 14. Ladd, G. W., Birch, S. H., & Buhs, E. S. (2003). Children’s social and scholastic lives in kindergarten: Related spheres of influence? Child Development, 70, 1373–1400. http://dx.doi.org/10.1111/1467-8624.00101 Lee, S., & Song, X. (2004). Evaluation of the Bayesian and maximum likelihood approaches in analyzing structural equation models with small sample sizes. Multivariate Behavioral Research, 39, 653–686. http://dx.doi.org/10.1207/ s15327906mbr3904 4 Leseman, P. (2009). Tackling social and cultural inequalities through early childhood education and care in Europe. Brussels, Belgium: The Education, Audiovisual and Culture Executive Agency. Li-Grining, C. P., Votruba-Drzal, E., Bachman, H. J., & Chase-Landsdale, P. L. (2006). Are certain preschoolers at risk in the era of welfare reform? Children’s effortful control and negative emotionality as moderators. Children and Youth Services Review, 28, 1102–1123. http://dx.doi.org/10.1016/j.childyouth.2005.10.016 Li-Grining, C. P., Raver, C. C., Champion, K., Sardin, L., Metzger, M., & Jones, S. M. (2010). Understanding and improving classroom emotional climate and behavior management in the “real world”: The role of Head Start

A. von Suchodoletz et al. / Early Childhood Research Quarterly 29 (2014) 509–519 teachers’ psychosocial stressors. Early Education and Development, 21, 65–94. http://dx.doi.org/10.1080/10409280902783509 Litjens, I., & Taguma, M. (2010). Literature overview for the 7th meeting of the OECD network on early childhood education and care. Paris, France: OECD Publishing. Mashburn, A. J., Pianta, R. C., Hamre, B. K., Downer, J. T., Barbarin, O. A., Bryant, D., et al. (2008). Measures of classroom quality in prekindergarten and children’s development of academic, language, and social skills. Child Development, 79, 732–749. http://dx.doi.org/10.1111/j.1467-8624.2008.01154.x Merritt, E. G., Wanless, S. B., Rimm-Kaufman, S. E., Cameron, C., & Peugh, J. L. (2012). The contribution of emotional support to children’s social behaviors and selfregulatory skills in first grade. School Psychology Review, 41, 141–159. Moser, T., & Bennett, J. (2006, August). The Nordic Early Childhood Education and Care-systems (ECEC) in a pan-European perspective. In Paper presented at the 16th Conference of the European Early Childhood Education Research Association (EECERA), “Democracy and Culture in Early Childhood Education” Reykjavik, Iceland, 31.8.06. Muthén, B. (2010). Bayesian analysis in Mplus: A brief introduction (Technical Report Version 3). Retrieved from Statmodel https://www.statmodel.com/ download/IntroBayesVersion%203.pdf Muthén, B., & Muthén, L. (2012). Mplus (Version 7) [Computer software]. Los Angeles, CA: Authors. Muthén, B. O., & Satorra, A. (1995). Complex sample data in structural equation modeling. Sociological Methodology, 25, 267–316. http://dx.doi.org/10.2307/271070 Early Child Care Research Network NICHD. (2000). Characteristics and quality of child care for toddlers and preschoolers. Applied Developmental Science, 4, 116–135. http://dx.doi.org/10.1207/S1532480XADS0403 2 Early Child Care Research Network NICHD. (2005). A day in third grade: A large-scale study of classroom quality and teacher and student behavior. The Elementary School Journal, 105, 305–323. http://dx.doi.org/10.1086/428746 O’Boyle, E. H., & Williams, L. J. (2011). Decomposing model fit: Measurement vs. theory in organizational research using latent variables. Journal of Applied Psychology, 96, 1–12. http://dx.doi.org/10.1037/a0020539 Organization for Economic Cooperation Development (OECD). (2012). Starting strong III: A quality toolbox for early childhood education and care. Paris, France: OECD. http://dx.doi.org/10.1787/9789264123564-en Pakarinen, E., Lerkkanen, M.-K., Poikkeus, A.-M., Kiuru, N., Siekkinen, M., RaskuPuttonen, H., et al. (2010). A validation of the Classroom Assessment Scoring System in Finnish kindergartens. Early Education and Development, 21, 95–124. http://dx.doi.org/10.1080/10409280902858764 Phillipsen, L. C., Burchinal, M., Howes, C., & Cryer, D. (1997). The prediction of process quality from structural features of child care. Early Childhood Research Quarterly, 12, 281–303. http://dx.doi.org/10.1016/S0885-2006(97)90004-1 Pianta, R. C., & Hamre, B. K. (2009). Conceptualization, measurement, and improvement of classroom processes: Standardized observation can leverage capacity. Educational Researcher, 38, 109–119. http://dx.doi.org/ 10.3102/0013189X09332374 Pianta, R. C., Steinberg, M. S., & Rollins, K. B. (1995). The first two years of school: Teacher–child relationships and deflections in children’s classroom adjustment. Development and Psychopathology, 7, 295–312. http://dx.doi.org/ 10.1017/S0954579400006519 Pianta, R. C., Howes, C., Burchinal, M., Bryant, D., Clifford, R., Early, C., et al. (2005). Features of pre-kindergarten programs, classrooms, and teachers: Do they predict observed classroom quality and child–teacher interactions? Applied Developmental Science, 9, 144–159. http://dx.doi.org/10.1207/s1532480xads0903 2 Pianta, R. C., La Paro, K., & Hamre, B. K. (2008). Classroom Assessment Scoring System – CLASS. Baltimore, MD: Paul H. Brookes.

519

Pianta, R. C., Mashburn, A. J., Downer, J. T., Hamre, B. K., & Justice, L. (2008). Effects of web-mediated professional development resources on teacher–child interactions in pre-kindergarten classrooms. Early Childhood Research Quarterly, 23, 431–451. http://dx.doi.org/10.1016/j.ecresq.2008.02.001 Raver, C. C., Jones, S. M., Li-Grining, C. P., Metzger, M., Champion, K. M., & Sardin, L. (2008). Improving preschool classroom processes: Preliminary findings from a randomized trial implemented in Head Start settings. Early Childhood Research Quarterly, 23, 10–26. http://dx.doi.org/10.1016/j.ecresq.2007.09.001 Riedel, B. (2005). (The institutional offer for children from 3 years to school entry (kindergarten age)) Das institutionelle Angebot für Kinder von 3 Jahren bis zum Schuleintritt (Kindergartenalter). Retrieved from http://www.bmfsfj.de/doku/ Publikationen/zahlenspiegel2005/01-Redaktion/PDF-Anlagen/Gesamt dokument,property=pdf,bereich=zahlenspiegel2005,sprache=de,rwb=true.pdf Rimm-Kaufman, S. E., La Paro, K. M., Downer, J. T., & Pianta, R. C. (2005). The contribution of classroom setting and quality of instruction to children’s behavior in kindergarten classrooms. The Elementary School Journal, 105, 377–395. http://dx.doi.org/10.1086/429948 Rimm-Kaufman, S. E., Curby, T. W., Grimm, K. J., Nathanson, L., & Brock, L. L. (2009). The contribution of children’s self-regulation and classroom quality to children’s adaptive behaviors in the kindergarten classroom. Developmental Psychology, 45, 958–972. http://dx.doi.org/10.1037/a0015861 Roßbach, H.-G., Kluczniok, K., & Kuger, S. (2008). Auswirkungen eines Kindergartenbesuchs auf den kognitiv-leistungsbezogenen Entwicklungsstand von Kindern [The effects of attending kindergarten on children’s cognitive and performance-related development]. Zeitschrift für Erziehungswissenschaft, 11, 139–158. http://dx.doi.org/10.1007/978-3-531-91452-7 10 Satorra, A. (2000). Scaled and adjusted restricted tests in multisample analysis of moment structures. In R. D. H. Heijmans, D. S. G. Pollock, & A. Satorra (Eds.), Innovations in multivariate statistical analysis: A Festschrift for Heinz Neudecker (pp. 233–247). London, UK: Kluwer. Song, X.-Y., & Lee, S.-Y. (2008). A Bayesian approach for analyzing hierarchical data with missing outcomes through structural equation models. Structural Equation Modeling, 15, 272–300. http://dx.doi.org/10.1080/10705510801922472 Spieß, K. (1998). (Governmental engagement in the child care markets. Analyses with German–American comparison) Staatliche Eingriffe in Märkte fuür Kinderbetreuung. Analysen im deutsch-amerikanischen Vergleich. Frankfurt am Main: Campus Verlag. Tietze, W., Cryer, D., Bairrão, J., Palacios, J., & Wetzel, G. (1996). Comparisons of observed process quality in early child care and education programs in five countries. Early Childhood Research Quarterly, 11, 447–475. http://dx.doi.org/10.1016/S0885-2006(96)90017-4 Tietze, W., Bairrão, J., Leal, T., & Roßbach, H.-G. (1998). Assessing quality characteristics of center-based early childhood environments in Germany and Portugal: A cross-national study. European Journal of Educational Psychology, 13, 283–298. http://dx.doi.org/10.1007/BF03173094 Vygotsky, L. S. (1978). Mind in society. Cambridge, MA: Harvard University Press. Wertfein, M., Kofler, A., & Müller, K. (2011). (Small children – high demands 2010! Selected results of the 2nd IFP-day nursery study) Kleine Kinder–großer Anspruch 2010! Ausgewählte Ergebnisse der 2 IFP-Krippenstudie. In In Staatsinstitut für Frühpädagogik & IFP-Infodienst Bildung, Erziehung, Betreuung von Kindern in Bayern. Munich, Germany: IFP. Wharton-McDonald, R., Pressley, M., & Hampston, J. M. (1998). Literacy instruction in nine first-grade classrooms: Teacher characteristics and student achievement. The Elementary School Journal, 99(2), 101–128.