Psychometric properties of the classroom assessment scoring system (Pre-K): implications for measuring interaction quality in diverse early childhood settings

The Classroom Assessment Scoring System (CLASS) is an observational instrument assessing the nature of everyday interactions in educational settings. The instrument has strong theoretical groundings; however, prior empirical validation of the CLASS has exposed some psychometric weaknesses. Further the instrument has not been the subject of psychometric analysis at the indicator level. Using a large dataset including observations of 993 Australian classrooms, confirmatory factor analysis is used to replicate findings from the few existing validation studies. Item response modelling is used to examine individual indicator behaviour. Latent growth models are used to produce new findings about estimating factor scores. Findings show that the CLASS exhibits stable psychometric properties within classrooms over repeated observations. Model fit is improved and factor scores are more reliable when the repeated-observations made in administering the CLASS are accounted for statistically. It is recommended that researchers enforce a fixed number of repeated observations to minimise bias