Lottery Evidence on the Impact of Preschool in the United States: A Review and Meta-Analysis

Jesse Bruhn and Emily Emick

November 2023

This paper reviews research on the design and effectiveness of preschools in the United States. Three randomized controlled trials that enrolled roughly 350 total children in demonstration studies during the 1960s and 1970s shape much of the current discussion of preschool. The authors examine what these studies and more recent ones based on random assignment research designs reveal about preschools today.

The authors find that the broad conclusions of the demonstration program literature are remarkably robust. Preschool generates large initial improvements in academic outcomes that fade out over time. The beneficial effects of preschool then re-emerge later in life on academic outcomes, like high school graduation, and non-academic outcomes, like criminal justice contact. Modern studies using larger samples also find positive effects on school discipline, SAT taking, and college attendance.

Variations in magnitude between the demonstration programs and modern studies can be explained by (1) differences in the quality and intensity of the interventions; and (2) the expansion of preschool, which changes the nature of the counterfactual for students not in preschool. Both factors highlight the importance of quantifying the features that create high-quality preschool experiences. Scaling effective practices have the potential to benefit the large number of current preschoolers and new children enrolled as a result of expansion.

Despite the concordance of findings between the demonstration and modern-era programs, the overall volume of evidence on preschool based on random assignment remains thin. Over the last 60 years, only seven experiments have randomized access to preschool. Since 1990, a total of 14 studies have randomly varied preschool characteristics. By leveraging lotteries used to allocate seats in oversubscribed schools, researchers can evaluate the impact of preschool in a broader variety of settings as well as identify the drivers of quality.