The psychometric properties of child care quality measures in Ecuadord : lessons for monitoring process variables at scale

This paper analyzes four instruments that are widely used to measure the quality of centers serving children ages 0 to 36 months - the CLASS, the ITERS-R, the CC-IT-HOME, and the MITRCC - and that were administered to a sample of 404 child care centers in Ecuador. We first assess the psychometric properties of these instruments in their first application in Ecuador. Specifically, we examine their internal consistency, test the underlying subscale structure by means of confirmatory factor analysis (CFA), verify construct validity by testing associations with quality-related factors (e.g., child-caregiver ratio), and check concurrent validity of the instruments' total scores. We then explore how we can use the data from these instruments to inform the development of a simple, less costly checklist that programs in Latin America and the Caribbean (LAC) can employ to monitor service quality. To that end, we look at the correlation of the separate subscales from the simpler measures (ITERSR, CC-IT-HOME, and MITRCC) with each of the dimensions of the CLASS. In addition, we map the subscales of all of the instruments to a set of child-caregiver interaction constructs identified in the literature as core for the healthy development of children, and used CFA to test how well the instrument subscales loaded onto each latent construct separately. Overall, the instruments showed excellent consistency. The first CFA also demonstrated that the data were a good fit to the published structure of each instrument. Associations with qualityrelated factors exhibited the expected signs, and concurrent validity across instruments revealed low correlations between overall scores on CLASS and the other instruments. Moderate correlations were found between certain dimensions of CLASS and the subscales of other instruments that reflected a priori similar constructs (in particular the Listening and Talking, Interaction, and Personal Care Routines subscales of the ITERS-R). Finally, the second CFA revealed that dimensions of CLASS had the highest loadings to three of the theoretical constructs for child-caregiver interaction: Sensitivity/Responsiveness, Positive Regard/Warmth and Joint Attention. However, several of the subscales of the ITERS-R and CC-IT-HOME, as well as the MITRCC showed encouraging associations with theoretical constructs for process quality that were not as strongly captured by CLASS.