Evaluation of the DC Opportunity Scholarship Program: Impacts After Three Years
Patrick Wolf, Babette Gutmann, Michael Puma, Brian Kisida, Lou Rizzo, Nada Eissa
Institute of Education Sciences
Amid all the sound and fury surrounding the D.C. voucher program, this study is a significant feather in the cap of the program's supporters. Why? Because despite the study's rigorous methods (a gold-standard randomized controlled trial, which usually finds "no effect"), students offered a voucher were performing at statistically higher levels in reading after three years (equivalent to a 3.1 month gain) than students not offered a scholarship. The reading finding is even more striking since the treatment group was highly mobile--a factor that likely contributed to the null findings in years one and two. (Over the three years, in fact, 51 percent of the treatment group switched schools 2-3 times.) Unsurprisingly, both groups performed similarly in math and the program did not have a significant impact in reading or math for those students who applied from the worst performing public schools. While this latter finding is unfortunate and has been cited by Secretary Duncan as reason to shut down the program, we should remember that students coming from the very worst schools are the hardest to remediate and require the most time to do so; this study only measures three years. In addition to the headline finding, the study discovered that the voucher program had improved reading achievement for 5 of 10 (statistically significant) subgroups: students not previously attending a school in need of improvement, those with higher levels of performance at time of entry into the program, those entering grades K-8 when they applied (i.e., everyone but high school students), female students, and first year students (though these last two are less reliable than the other three). Shame on the Department for pulling its Friday-afternoon-release stunt for such important data. We're hoping Congress perused the Saturday papers. Read it here.
blog comments powered by Disqus