Commentary: The positive impacts of public Pre-K fade quickly, and sometimes reverse: What does this portend for future research and policy?

Lipsey, Farran, and Durkin (2018) report the results from the third-grade follow-up of children who, as four-year-olds, were in the treatment and control groups of the Tennessee Voluntary Prekindergarten Program (TVPK). Results include an intent-to-treat analysis of state test scores based on a randomized trial (RCT). TVPK produced substantive positive impacts on children during the pre-K year, but differences between treatment and control children faded to zero, and, for state test scores, turned negative as children moved through elementary school. The negative outcomes, although not without precedent, are surprising to many. The more general finding of very rapid fadeout of preschool impacts is consistent with a large number of studies. Researchers and policymakers can respond to the Lipsey et al. findings in several ways. These include dismissing the findings because of the assertion that TVPK is a low-quality program, or arguing that the research design was flawed, or claiming that the results of other strong studies of preschool interventions conflict with results reported for TVPK. Each of these lines of attack does not survive close inspection of their empirical entailments. Three paths forward are respectful of the findings: (1) continuing the priority of advancing public pre-K for four-year-olds, coupled with efforts to improve longer-term program effects by capitalizing on findings from research on heterogeneity of impacts; (2) reframing the purpose and metrics for success of pre-K away from affecting children's achievement in later grades to providing positive and engaging experiences for children when they are enrolled; and (3) pursuing new models of policy that shift the focus from school readiness to family support. Significant additional investments in research and evaluation will be required to pursue the effectiveness of new policy frameworks for early care and learning.