Does a Successful RCT Yield Successful Policy? Class Size Reduction in Tennessee After Project STAR

Does the use of research on small class sizes lead to successful evidence-based policy and improved student outcomes?

The investigators will study whether Tennessee’s use of research to inform changes to its class-size policies relate to changes in student outcomes. Building on an emerging body of research about the benefits of small class sizes, between 1985 and 1989 the Tennessee legislature authorized a randomized controlled trial, Project STAR, to evaluate whether small class sizes increased reading and math scores in grades K-3. The initial evaluation yielded positive effects. In response, the state launched Project Challenge in 1989 to reduce K-3 class sizes in 17 of the state’s poorest districts. This was followed by a statewide reduction of K-3 class sizes in 1993. The impact of using research evidence to drive these later policy changes was never tested. The investigator will examine whether or not Project Challenge replicated the results of Project STAR and whether differences in the conditions under which the research-informed policies were implemented relate to differences in outcomes. Projects STAR and Challenge were implemented in two different contexts, the former in counties with varying levels of diversity, and the latter in poor, rural, primarily White areas. In addition, while the implementation and outcomes of Project STAR were closely monitored and evaluated, this was not the case in Project Challenge. Funds from this award will allow the team to collect and transcribe all of the necessary data from district report cards, from 1989 onward. This includes: program eligibility criteria, information to calculate student-teacher ratios, test scores, standardized test scores from before the program was implemented, the district’s racial/ethnic composition, and location. A regression discontinuity analysis will be applied to estimate the effect of Project Challenge on test scores based on two eligibility variables: per-capita income and eligibility for free and reduced-price lunch.