Recent years have seen both the rapid expansion of state pre-k programs and the emergence of new, compelling evidence that these state pre-k programs effectively promote child development; additionally, President Obama recently announced a “preschool for all” initiative. However, much research linking aspects of early care and education quality to child development was conducted prior to the dramatic proliferation of state pre-k programs. Thus, little is known about the “active ingredients” of a successful public pre-k program. This study aims to understand the features of pre-k programs that are most strongly related to short-term academic gains. 

Using data from the Tulsa, Oklahoma pre-k program, the current study addresses this important question. The Tulsa school district is the largest in Oklahoma, and its pre-k program has received national attention because of its strong commitment to universality (70% state-wide penetration rate for 4-year olds), and its reliance on well-qualified and well-paid teachers. Prior research has found positive short-term impacts of Tulsa’s pre-K program across two cohorts of children.

Consistent with prior studies using these data, we will employ a regression discontinuity design (RDD) to estimate short-term gains (at kindergarten entry) from pre-k on measures of children’s early reading, writing, and math skills. RDD is possible because only children who turned 4 years old before September 1 of a given year could enroll in pre-k; due to this strict birthday-based pre-k assignment mechanism, any difference in outcomes between “treated” children who received pre-k in 2005-2006 and entered kindergarten in 2006-2007, and “comparison” children who were just entering pre-k in 2006-2007, are attributable to the pre-k program rather than to other characteristics of the child or family.

Using RDD, previous work on the same cohort of children examined in the current study found (pooled) positive treatment effects of pre-k participation on children’s pre-reading (d =.99), pre-writing (d =.74), and pre-math skills (d =.36). Using a sample of approximately 2,000 treatment group children and 1,840 comparison group children in 58 Tulsa schools, the current study will estimate RDD effects on these outcomes in a multilevel framework, estimating a separate treatment effect for each school. Preliminary results suggest that there is statistically significant variation in treatment effects across schools. Within the multilevel framework, we will control for child-level characteristics and model the variation across schools in treatment effects as a function of school-level characteristics. Teacher- and classroom-level variables shown in prior research to be markers of high-quality early learning environments, such as classroom instructional and emotional climate and time spent on literacy and math instruction, will be aggregated to the school level for schools where multiple classrooms are present (28 of the 58 schools had just one pre-k classroom, while only 5 schools had more than 3 pre-k classrooms). The coefficients on these school-level quality variables are our primary focus, and represent mediators of the treatment effect.

Empirical evidence illuminating the Tulsa pre-k program characteristics that promote early academic success will suggest what high-quality pre-k education could do if similar investments are made across the country.