Poster #183 - The network approach to unpack the multicausal system of early care and education quality

• In Event: 3-106 - Poster Session 11
  In Poster Session: PS 11 Section - School Readiness/Childcare

Integrative Statement

What are the central features of early childhood education and care (ECEC) quality? Despite extensive research on the reductionist structural quality → process quality → child outcomes model, fundamental questions on the best ways to define ECEC quality remain debatable. If researchers thus have difficulties making recommendations on how to define and promote high ECEC quality (OECD, 2018), what to expect from policy and practice? At the core of the current model are aggregated mean scores of teacher-child interaction quality ratings (labeled as process quality) justified by means of traditional psychometrics (i.e., factor analysis). Recent findings have put into questions these quality scores because they allow distinguishing only between few classrooms of low and high quality (Burchinal, 2018). In this paper we introduce as an alternative a network-theory-inspired approach that is gaining popularity in other areas of science (Barabási, 2011; Borsboom, 2017). In contrast to previous research, overall levels of ECEC quality may result from complex causal links between single indicators (e.g., teacher-child-ratio→ teachers’ responsiveness→ language modeling→ scaffolding learning) that may be bidirectional. For example, sensitive teachers may engage in conversations with children more often, and this would further enhance their sensitivity to the children's needs.
We examined network structures of teacher-child interactions (operationalized by CLASS Pre-K: The Classroom Assessment Scoring System; Pianta, La Paro, and Hamre, 2008), structural quality, and classroom composition variables (e.g., proportion of disadvantaged children) in a sample of 177 German preschools. On a descriptive level of analysis, univariate latent growth models showed a decrease in interaction quality and an increase in children-to-teacher ratios over the course of one typical morning. Multivariate network analysis (regularized EBIC graphical LASSO estimation of partial correlations; Epskamp & Fried, 2018) was conducted on both between-classrooms (Figure 1) and within-classrooms levels (Figure 2). In the within-analysis (Figure 2), longitudinal SEM fixed effects models (Bollen & Brand, 2010) were used for deriving partial correlations –this allowed us to control for potential unobserved and time stable confounders (e.g., classroom curricula or teacher/child traits). Children-to-teacher ratios were associated directly only on the between level with classroom behavioral management and teachers’ facilitation of child engagement (Figure 1), but not with longitudinal changes in teacher-child interactions (Figure 2). This suggests that teacher-child ratios are an important precondition, but not a decisive factor, for effective interactions. Taken together, the network analyses revealed that teacher’s overall involvement, teacher’s sensitivity, language modeling, and the organization of learning formats were key predictors for effective classroom interactions by centrality network indices. This is the first study to explore the network structures between ECEC process quality, structural quality, and classroom composition variables. We discuss how these findings might help future research and policy interventions to identify central aspects of ECEC quality that can activate effective teaching and learning processes.

Authors

• Alexandru Agache, Ruhr University Bochum
  Presenting Author

• Lilly Bihler, Ruhr-Universität Bochum
  Non-Presenting Author

• Katharina Kohl, Ruhr-Universität Bochum
  Non-Presenting Author

• Jessica Willard, Ruhr-Universität Bochum
  Non-Presenting Author

• Birgit Leyendecker, Ruhr-Universität Bochum
  Non-Presenting Author