Using a Natural Experiment to Examine NC Pre-K Main Effects: An Updated Approach

• In Event: Disentangling Heterogeneity in the Impact of North Carolina’s Public Pre-Kindergarten Program

Abstract

Converging evidence suggests that access to public pre-k supports student learning in the short-run. However, studies examining longer-term effects of pre-k exposure have reported a more mixed set of findings, highlighting the need for more rigorous research on the long-term effects of public pre-k programs (Phillips et al., 2017). Our analyses update and extend previous work from Dodge et al. (2017) that leveraged differences in the level of funding provided to counties for the NC Pre-K program as a natural experiment to test impacts of program exposure on student outcomes. This approach leverages population-level data in NC on over 1 million public school students, and can be likened to a “difference-in-differences” design, whereby year-to-year differences in pre-k funding provided at the county level generate exogenous variation in program exposure (i.e., within-county variation in funding is exogenous to key student factors that predict selection into pre-k). We updated previous work by including six additional cohorts of students and additional covariates, and we also incorporated enrollment information to directly test the impact of enrollment in the program on fifth grade outcomes. Our outcomes included a composite measure of achievement (averaged across both math and reading scores)and indicators for special education placement and grade retention between grades 3 and 5, respectively.

We found that an additional $1000 of county-level NC Pre-K funding was associated with a statistically significant increase in fifth grade achievement of approximately 0.03 SD. To contextualize this estimate, this would imply that an investment of $2030, the average per 4-year-old investment during in the 2010 program year, would lead to a 0.06 SD increase in student achievement at fifth grade. However, we found no association between NC Pre-K funding and special education and retention status in fifth grade.

To better understand how this funding exposure affected the learning of children who enrolled in pre-k, we incorporated new enrollment data into an instrumental variables approach. With these analyses, we used county-level funding as an instrument for pre-k enrollment using two-stage least squares estimation. Our first stage analyses from this approach indicated that a $1000 increase in county-level NC Pre-K funding increased the likelihood of enrollment in NC Pre-K for the entire population of children by 15.3 percentage points (F = 1,462.28). For context, statewide enrollment ranged from 1% of the 4-year-old population at the start of the program to 28% by the 2010 program year. Our second stage analyses then found that enrollment in NC Pre-K resulting from funding expansions led to a statistically significant achievement boost of 0.20 SD at the end of fifth grade on the composite outcome (i.e., a local average treatment effect). Importantly, the achievement effects held for both reading (coefficient = 0.22, p<.01) and math test scores (coefficient = 0.16, p<.05).

These analyses suggest that funding for pre-k did lead to important increases in pre-k enrollment, as well as student achievement that were detectable in grade 5. However, we saw no indication that funding for pre-k led to reductions in grade retention or special education placement.
 

Authors

• Tyler Watts, Teachers College, Columbia University
  Presenting Author

• Jade V Marcus Jenkins, University of California - Irvine

• Kenneth A Dodge, Duke University

• Robert C. Carr, University of North Carolina at Chapel Hill

• Maria Sauval, University of California - Irvine