Design, Sites, and Data Sources
What Matters Most for Teachers and Young Children is a secondary data analysis using quantitative data from Making Pre-K Count, a large-scale cluster randomized controlled trial conducted by MDRC. Making Pre-K Count evaluated the effect on children’s outcomes of an evidence-based math curriculum (Building Blocks) combined with extensive teacher training and in-classroom coaching. The sample consisted of 69 full-day preschool programs funded by New York City’s Department of Education and Administration for Children’s Services — in public schools and community-based centers, including Head Start — serving a low-income population of 4-year-old children. The preschool sites were selected to reflect the geographic, racial, and ethnic diversity of New York City’s low-income population, although the sample was not designed to be statistically representative. Preschool sites were randomly assigned either to receive two years of curriculum plus professional development (program group) or to continue with “preschool as usual” (control group). A total of 34 preschools (88 classrooms) were in the program group, and 35 preschools (89 classrooms) were in the preschool-as-usual group.
The study was conducted during the school years of 2013-2014 and 2014-2015; data on children were collected in the second year. The data from Making Pre-K Count are uniquely suited to answer this study’s research questions about links among teacher practice, children’s learning, and professional development, because multiple quantitative measures of these constructs were collected, but explicit modeling of such associations was not a primary aim of the Making Pre-K Count study.
In What Matters Most for Teachers and Young Children, structural equation modeling will be used to examine the first set of research questions that follow theorized pathways from teacher professional development to teacher practice to child outcomes and how pathways may vary for different subgroups of children and classrooms. For the second set of research questions, a combination of latent profile analysis and multinomial logistic regression will be used to dive into implementation processes in those classrooms receiving a math intervention by examining “profiles” of teacher practice and which teacher and classroom characteristics predict them.