Identifying Actionable Classroom and Program Features for Scaling High-Quality Prekindergarten

Abstract

In this dissertation, I provide actionable evidence on three related pressing questions for the early education field: what instructional features of classrooms predict children’s academic gains in the prekindergarten year, whether the complexity of instruction from prekindergarten through first grade is aligned and contributes to children’s within-year academic gains; and how to optimize the selection of centers participating in universal prekindergarten programs diminishing the risks of unintended patterns that could affect the quality of expansion programs. I use descriptive, psychometric, predictive, and geospatial methods to answer my research questions. In my first study, I focused on time use in prekindergarten classrooms. I compared the measurement properties of two instructional quality observational instruments: the Narrative Record (Farran et al., 2015) and the Individualizing Student Instruction (Connor et al., 2009) systems. Findings show that the NR and the ISI produce partially different descriptions of time use, and their resultant classroom instructional profiles are inconsistent. Although these profiles do not predict children’s language and working memory gains, they do predict math gains with opposite directions depending on the used measure. I also discuss demographic differences by classroom type and illustrate a modeling approach that accounts for such differences in predictive models. My results can aid practitioners in monitoring equitable instruction across classrooms. In my second study, I used multi-level linear regression models to identify the degree of instructional alignment in relation to children’s exposure to content complexity and examine its contribution to children’s within-grade gains in language (Dunn & Dunn, 2007) and math (Clements et al., 2008; Woodcock et al., 2001, 2005). Findings show within-grade variation in the complexity of content instruction across classrooms. Moreover, I show that children in prekindergarten classrooms are exposed to highly complex content, and such complexity is not sustained but rather decreases in subsequent grade levels in reference to national and state learning standards. Parametric measures show consistent patterns, demonstrating that children are exposed to content of roughly the same complexity from prekindergarten to first grade. Although such variation does not predict children’s within-grade language and math gains, these results highlight the importance of assessing instructional alignment and identifying its potential contribution to children’s fade-out or convergence of skills after prekindergarten. In my third study, I identified statistical and geo-spatial differences between centers that self-select to participate in the Boston Universal Prekindergarten (UPK) program and other centers in the Boston area, using administrative data from their licensing, quality rating and improvement system, and accreditation status. Results show that UPK appliers are located in and serve similar communities to non-appliers but are more likely to receive subsidies and participate in Quality Rating and Improvement Systems (QRIS). Differential participation in QRIS between appliers and non-appliers increases when models are restricted to CBOs receiving subsidies. These findings highlight the importance of monitoring quality at the population level using strategies independent of monetary incentives to secure equitable access to high-quality settings for low-income families. Together, these three studies contribute to the discussion in the field about how to operate and scale high-quality public prekindergarten programs. First, by monitoring disparities in access to varied and complex instructional experiences across classrooms and schools. Second, by implementing strategies that include population-level information to maximize equity and quality in selecting partners for universal prekindergarten programs.