Association for Public Policy Analysis & Management

While there is quite a bit of evidence of charter school impacts on test score achievement, much less is known about impacts on other important outcomes. This is an important omission to address given that societal goals about schooling are not limited to raising academic achievement, but more broadly enhancing college and career readiness and civic participation and reducing criminal behavior, among other important outcomes.

An experimental study of six overenrolled No Excuses charter high schools in Boston found that students who attended the charter high schools had higher SAT scores and were more likely to attend four-year post-secondary institutions than students in traditional public high schools (Angrist, Cohodes, Dynarski, Pathak, and Walters, 2013). While this study has strong internal validity, it is limited by poor external validity. Two large scale observational studies conducted using data from Chicago and Florida estimate charter school impacts in large samples of charter schools (Booker et al, 2011 and Sass et al 2016). Results from Florida show that charter school students are more likely to receive a standard high school diploma, attend a two- or four-year college within six years, persist in post-secondary education at least two consecutive years, and earn more in the labor market. In short, the evidence base on the effect of charter schools on students’ long-term outcomes is thin.

We add to this emerging literature by evaluating the impact of charter school attendance on secondary, post-secondary, and non-schooling outcomes in North Carolina, a state with a large and growing charter school sector. Specifically, we examine the effect of attending a charter high school on a wide range of educational outcomes— including course-taking patterns, attendance, dropout rates, graduation rates, post-secondary attainment and success—as well as criminal behavior and voting.

The data from this project will utilize and merge together population level student- and school-level data files from North Carolina’s K-12 system; post-secondary enrollment and completion data from the University of North Carolina General Administration, the North Carolina Community College System Office, and the National Student Clearinghouse; young adult voting behavior from the North Carolina State Board of Elections; and criminal records from the North Carolina Department of Corrections. Once merged together, this project will create one of the largest and most comprehensive statewide data sources ever assembled to analyze charter school effects.

Following findings from within-study comparisons (Bifulco, 2012; Clair, Cook, & Hallberg, 2014; Steiner, Cook, & Shadish, 2011), the study will employ propensity score matching models that control for baseline test scores, behavioral indicators, socio-demographic background, and sending school middle school. These models have been shown to closely replicate test score impacts, although the validity of impacts on other outcomes is less clear. We will also probe the sensitivity of our results to techniques introduced by Sass et al (2016).