Design, Sites, and Data Sources
Each of the eight elementary schools selected for the study had a high proportion of at-risk students, had implemented the main components of the Accelerated Schools reform by the early 1990s, had not instituted other major reforms over the eight-year study period, and were able to supply the data needed for the analysis. These schools were located in Santa Rosa, California; Waimanola, Hawaii; Aurora, Illinois; St. Louis, Missouri; St. James, Missouri; New York City, New York; and Charleston and Spartanburg, South Carolina.
MDRC’s Accelerated Schools study pioneered the use of an innovative methodology for measuring program impacts when random assignment is not feasible. At the core of the study was an interrupted time series analysis of the reading and math test scores of successive cohorts of third-graders in eight Accelerated Schools. In the analysis, the scores of third-graders who attended the schools during the three years before the reform was launched were used to predict what these scores would have looked like over the next five years had the reform not been put in place. Effects on achievement were obtained by comparing the predicted test scores with the scores that were registered. MDRC has since refined this design and applied it in other evaluations of school-based reforms.