LEVELS OF QUALITY AND CHILDREN'S DEVELOPMENT ACROSS

University Of Nebraska Lincoln, Lincoln NE

Investigators

Abstract

The overall goal of the project is to identify thresholds of early childhood education quality in predicting social-emotional, cognitive, and language outcomes in multiple secondary data sets that can inform national and state policies that promote optimal child development through early childhood education and child care settings. The project's three objectives are to: (1) compare different analytic strategies for identifying thresholds of quality; (2) replicate the analytic strategy with multiple national and state data sets to determine if thresholds are similar or different across data sets and to examine convergent findings across data sets; and (3) examine subgroup differences to determine whether minimum levels of quality necessary to promote positive development differ based on family resources (e.g., family income, parent education); child characteristics (child sex, age); child minority status and cultural background (child language, race, ethnicity); or program context (geographic setting/program auspice). Data sets from the Early Head Start Research and Evaluation Project (EHSREP), Quality Interventions in Early Care and Education (QUINCE), Early Head Start Family and Child Experiences Survey (Baby FACES), Early Childhood Longitudinal Study-Birth Cohort (ECLSB), Study of Statewide Early Education Programs (SWEEP), Educare Learning Network (Educare), and Nebraska Student and Staff Record System in preschool and infant toddler databases (NSSRS) will be used. The research team's previous ACF-funded research examined thresholds or active ranges of quality that influenced children's development using Generalized Additive Modeling analysis paired with linear spline modeling. The study will build on this work by incorporating two complementary yet distinct methods. Generalized Additive Mixed Modeling will be implemented to account for the nested structure many of the data sets offer. Multivariate Adaptive Regression Splines provide a means for efficiently examining a large number of variables and determining the existence of key interaction effects (i.e., moderators or sub-group differences in the relationship between quality and outcomes). Results from this investigation will provide a much-needed empirical basis for the high-stakes cut points now in use in Quality Rating and Improvement Systems and tiered reimbursement systems. Identifying converging results across multiple data sets will increase confidence in the validity of thresholds identified.