Association for Public Policy Analysis & Management

Panel: Unpacking Charter School Effects

(Education)

As the research on charter schools has proliferated, perhaps the only thing that has become clear is how much variation there is in charter school impacts. The three papers in this panel employ administrative data to shed light on different aspects of charter impact heterogeneity. They illustrate how longitudinal data from state education agencies can be a valuable research tool for determining when and why charter schools are effective.

Gulosino and Liebert exploit longitudinal administrative data to look at suburban v. urban variation for charter schools in California. A consistent pattern so far in the charter literature has been that urban charter schools are more likely to be effective than suburban charter schools. This paper adds and expands on that literature by introducing a rich set of neighborhood characteristics from the US census to control for selection bias.

Berends, Ferrare, and Waddington use longitudinal data to look at heterogeneity by charter management structures, focusing on a previously understudied slice of the charter sector: for-profits. They compare for-profit, non-profit, and virtual charters relative to traditional public schools to help unpack determinants of charter success.

Finally, Dynarski, Hubbard, Jacob and Robles present the first lottery-based estimates of for-profit charter school impacts for the largest charter operator in Michigan. Admissions lotteries matched to state longitudinal data and census records allow researchers to extract causal estimates that run counter to prior literature. They show for-profit charters in Michigan work for a very different population than the traditional, No-Excuses urban charters in other lottery-based studies.