I. QUALITY THRESHOLDS, FEATURES, AND DOSAGE IN EARLY CARE AND EDUCATION: INTRODUCTION AND LITERATURE REVIEW.

This monograph addresses the hypotheses that preschool children benefit most strongly when early care and education (ECE) is at or above a threshold of quality, has specific quality features, and/or is of longer duration. These issues are pivotal in recent policies designed to improve the quality of ECE, especially for children from low-income families. Evidence of quality thresholds in which ECE quality has stronger impacts in settings with moderate to high levels of quality than in settings with low quality would inform policy initiatives in which monetary incentives or consequences are allocated to ECE settings based on their level of quality. Evidence that specific features of quality, such as quality of teacher-child interactions and of literacy and mathematics instruction, are predictors of gains in child outcomes could help inform quality improvement efforts. Evidence that more time spent in center-based ECE or in instruction in specific content areas predict larger gains among preschoolers could be useful in designing public preschool programs such as Head Start or prekindergarten. Secondary data analyses of eight large studies of preschool children in center-based ECE were conducted. Analyses focused on quality thresholds and quality features examined the extent to which three types of quality measures predicted gains in children's language, literacy, mathematics, and social skills. The measures comprised (1) global quality measures that provide an overall or global rating of quality, focusing on interactions as well as on physical features of the environment, activities, and routines; (2) interaction-specific measures that focus in depth on the quality of interactions between teachers and children with respect to instructional and emotional support; and (3) domain-specific measures that focus on the quality of instruction and stimulation in specific content areas such as early language and literacy. The goal was to provide replicated analyses with data from several projects in order to address each question. Multilevel analyses that controlled for entry skills were conducted, and results were combined by using meta-analysis, nonlinear and nonparametric analyses, and propensity score analyses. With respect to thresholds, the analyses suggest that increases in the quality of instruction are related to larger gains in language and literacy outcomes, but only in higher quality classrooms. Results point to stronger associations between quality and child outcomes in higher versus lower quality classrooms for measures of the instructional quality of teacher-child interactions and of the quality of specific activities thought to promote early literacy, such as teaching phonemic skills and book reading. In addition, the items focusing on quality of interactions on the global measure also predicted acquisition of language and social skills in higher but not in lower quality classrooms. With respect to quality features, interaction-specific and especially domain-specific measures of quality remained significant predictors of child outcomes, whereas global measures of quality were never significant positive predictors, when both global and more specific measures of quality were included simultaneously in analyses. There is thus consistent evidence that more specific measures of quality are better predictors of child outcomes. With respect to dosage, several approaches were used in operationalizing both the cumulative and current dosage of children's exposure to ECE. Propensity score analyses that included baseline scores on outcomes to control for selection into larger dosages suggested that children with two as opposed to one year of Head Start had stronger vocabulary and literacy skills both immediately upon exit from Head Start and at the end of kindergarten. Fewer absences and more observed time spent on instruction were associated with stronger gains in literacy and mathematics skills. Finally, findings revealed that more time spent on instruction in classrooms with higher overall quality was particularly important to the development of mathematics skills. No other replicated evidence of quality by quantity interactions emerged.

III. TESTING FOR QUALITY THRESHOLDS AND FEATURES IN EARLY CARE AND EDUCATION.

In this chapter, we report on the analyses focusing on both quality thresholds and quality features. First, we address questions about quality thresholds, using two analytic approaches. The analyses ask whether there is evidence suggesting thresholds in the association between a specific quality measure and a specific child outcome. Second, we extend these analyses to ask whether each child outcome is more strongly related to global quality measures or to quality measures that measure teacher-child interactions or quality of instruction in a given content area. The research to date provides the basis for the articulation of two hypotheses related to quality thresholds and features: (1) the quality of ECE is a stronger predictor of residualized gains in child outcomes in classrooms with higher quality than in classrooms with lower quality and (2) more specific measures of quality are stronger predictors of residualized gains in child outcomes than are global measures. We turn now to analyses intended to address these hypotheses by using data from several data sets.