What’s the secret ingredient? Searching for policies and practices that make charter schools successful

ABSTRACT

The charter-school sector in the United States has grown steadily since the first charter school opened in 1992 and nearly 7,000 charter schools currently serve nearly 3 million students. Research suggests that the average charter school performs about the same as nearby traditional public schools, but there is great variation in the effects of charter schools. Some are successful in boosting student achievement and others are not, raising the question of what distinguishes good charter schools from bad. This article summarizes the research on factors associated with successful charter schools. The research suggests that urban charter schools have more positive impacts on student achievement than other charter schools. The policies most consistently found to be associated with positive charter-school impacts include long school days or years, comprehensive behavioral policies with rewards and sanctions, and a mission that prioritizes boosting student achievement. In addition, moderately strong evidence suggests that high-dosage tutoring, frequent feedback and coaching for teachers, and policies promoting the use of data to guide teachers’ instructional practices are positively associated with charter schools’ achievement impacts.

KEYWORDS:

• charter schools
• student achievement
• school best practices
• impact heterogeneity

Notes

¹. One might question whether there is similar variation in the impacts of individual traditional public schools. Zimmer et al. (2012) examined this issue and found significantly greater variation in school-level impacts in the charter-school sector than in the traditional public school sector in most of the states they examined.

². There are exceptions. Gleason et al. (2010) and Furgeson et al. (2012) measured a treatment contrast for selected school characteristics using charter school and traditional public school differences based on data from a principal survey. Berends et al. (2010) used a different methodology and did not directly measure a treatment contrast but did account for the characteristics of both charter schools and traditional public schools based on both principal and teacher surveys. Chabrier et al. (2016) examine variation in charter-school impacts after accounting for average test scores in students’ “fallback” traditional public schools.

³. Conceptually, the characteristics of the relevant set of traditional public schools should be reflected in measures of the treatment contrast. As described above, however, it is difficult to accurately measure these characteristics and, as a result, a contextual variable such as an indicator for being located in an urban area may serve as a proxy for traditional public-school characteristics.

⁴. The studies reviewed here have explicitly tried to limit the number of factors they’ve examined given the limited degrees of freedom in analysis based on a sample of 30–40, but still have examined correlations with charter-school impacts of between 11 and 43 different factors.

⁵. Evidence was also considered strong if three studies found a significant relationship and there was compelling supporting evidence.

⁶. By contrast, Knechtel et al. (2015) found no statistically significant relationship between time in school and the impacts of KIPP schools.

⁷. Chabrier et al. (2016) found no significant association between teachers having a master’s degree and impacts in the bivariate analysis, but did find a significant positive association in the multivariate analysis.

⁸. One study did find some evidence of a relationship between CMO status and charter-school impacts. Dobbie and Fryer (2013) found that New York City charter schools affiliated with a CMO had significantly more positive impacts at the elementary school level, though differences at the middle-school level were not significant. In addition, CREDO (2017) found that impacts among CMO schools were slightly larger than those of independently operated charter schools (e.g., with effect sizes of 0.03 versus 0.01 in math) but did not test the significance of this difference.

⁹. Zimmer et al. (2011) examined patterns of impacts among first-year versus third-year charter schools in eight states. In three of the states, impacts were substantially more positive in the third year than in the first year, while in the other states there were not substantial differences in the schools’ impacts between their first and third years. The authors did not test the statistical significance of any of these differences. Buddin and Zimmer (2005) found significant differences in the impacts of new versus more established charter schools, but the direction of these differences was not consistent. For some types of charter schools, more established schools performed better than new schools but the reverse was true for other types of charter schools.

¹⁰. Some evidence suggests that successful charter schools tend to use most or all of these policies and practices in clusters, rather than just one or two of them (Angrist et al., 2013; Dobbie & Fryer, 2013; Furgeson et al., 2012). However, the research focuses on the correlations between individual factors and impacts, either in a bivariate or multivariate model, and does not examine whether there are interactions between factors in their correlation with charter-school impacts.