A critical look at methodologies used to evaluate charter school effectiveness

There is no consensus among researchers on charter school effectiveness in the USA, in part because of discrepancies in the research methods employed across various studies. Causal impact estimates from experimental studies demonstrate large positive impacts, but concerns about the generalizability of these results have prompted the development of alternative methodologies for estimating charter impacts. Considerable attention has been paid to non-experimental national charter school studies conducted by the Stanford Center for Research on Education Outcomes (CREDO). To reconcile conflicting findings in the charter school literature, this within-study analysis uses statewide data from Florida between 2001 and 2009, finding large differences in the estimated charter school effects for oversubscribed and undersubscribed schools. This helps explain why experimental studies often find charter impacts that are much larger than those reported by studies relying on non-experimental methods.