CLASS Validation in Australia and the Influence of Instructional Support on Children's Cognitive Development

This paper contributes to current debate about the psychometric properties of the Classroom Assessment Scoring System (CLASS) through (1) validation of the measurement model, (2) new perspectives about sources of misfit, and (3) the predictive validity of the Instructional Support (IS) scale on children's cognitive development. (1) The traditional CLASS measurement model is assessed for the first time using both item response modelling and classical test theory (e.g., CFA). (2) Multiple indicator latent growth models (LGM) are used to account for within-classroom variation caused by the repeated-observation collection protocols. (3) LGM is used to model the influence of IS on children's cognitive development. Data are sourced from the Effective Early Educational Experiences (E4Kids) study. In 2011, the study observed 993 classrooms including centre-based child care, preschool, and school classrooms. There are opportunities to improve CLASS scale reliability by increasing the discrimination of some items, particularly in the IS Domain. Modelling the repeated-observation nature of the CLASS improves fit and reduces inter-correlations between the latent scales. The effect of IS on children's cognitive development is consistent with international studies (e.g., ES ~0.1 vs 0.3+) with higher quality IS producing gains of between one and three months on children's verbal and early mathematical abilities. There are significant opportunities to increase the level of IS in early childhood settings. The CLASS functions robustly in Australia. Family SES is a strong predictor of both CLASS ratings and child's development. This has strong implications for policy aimed at reducing inequality of outcomes related to income.