Gold Standard Studies

Evaluating School Choice Programs

The gold standard methodology in the social sciences is called “random assignment.” This method allows researchers to isolate the effects of vouchers or scholarships from other student characteristics. Students who apply for a voucher enter randomized lotteries to determine who will receive the voucher and who will remain in a public school; this allows researchers to track very similar "treatment" and "control" groups, just like in medical trials. The following study citations and brief summaries are listed in chronological order, from most recent to oldest.

Study Citations & Findings

Patrick Wolf, Babette Gutmann, Michael Puma, Brian Kisida, Lou Rizzo, and Nada Eissa, “Evaluation of the D.C. Opportunity Scholarship Program: Final Report,” U.S. Department of Education, June 2010.

Washington D.C.—The students offered vouchers graduated from high school at rate 12 percentage points higher (82 percent) than students in the control group (70 percent), an impact that was statistically significant at the highest level. Students in 3 of 6 subgroups tested showed significant reading gains due to the voucher offer after 4 or more years. Overall, reading and math gains from the program were not statistically significant at the 95 percent confidence level, but overall reading gains were significant at the 90 percent confidence level. Parents remained more satisfied with their child's school and viewed it as safer if offered a voucher, even though students had similar views of school satisfaction and safety whether in the treatment or control group.

Joshua Cowen, “School Choice as a Latent Variable: Estimating ‘Complier Average Causal Effect’ of Vouchers in Charlotte,” Policy Studies Journal, November 2007.

Charlotte, NC—After one year, voucher students had reading scores 8 percentile points higher than the control group and math scores 7 points higher.

Alan Krueger and Pei Zhu, “Another Look at the New York City School Voucher Experiment,” American Behavioral Scientist, January 2004.

New York, NY—The voucher students had higher scores, but the results did not achieve statistical significance. Subsequent analysis has demonstrated that this occurred because the study used inappropriate research methods that violate the norms of the scientific community; if legitimate methods are used, the positive results for vouchers become significant

John Barnard, Constantine Frangakis, Jennifer Hill, and Donald Rubin, “Principal Stratification Approach to Broken Randomized Experiments: A Case Study of School Choice Vouchers in New York City,” Journal of the American Statistical Association, June 2003.

New York, NY—After one year, voucher students had math scores 5 percen¬tile points higher than the control group.

William Howell and Paul Peterson, The Education Gap, Brookings Institution, 2002.

Washington D.C.—After two years, black voucher students had combined reading and math scores 9 percentile points higher than the control group.

William Howell and Paul Peterson, The Education Gap, Brookings Institution, 2002.

Dayton, OH—After two years, black voucher students had combined reading and math scores 6.5 percentile points higher than the control group.

William Howell and Paul Peterson, The Education Gap, Brookings Institution, 2002.

New York, NY—After three years, black voucher students had combined reading and math scores 9 percentile points higher than the control group.

Jay Greene, “Vouchers in Charlotte,” Education Next, Summer 2001.

Charlotte, NC—After one year, voucher students had combined reading and math scores 6 percentile points higher than the control group.

Cecilia Rouse, “Private School Vouchers and Student Achievement,” Quarterly Journal of Economics, May 1998.

Milwaukee, WI—After four years, voucher students had math scores 8 NCE points higher than the control group. NCE points are similar to percentile points.

Jay Greene, Paul Peterson and Jiangtao Du, “School Choice in Milwaukee: A Randomized Experiment,” in Learning From School Choice, eds. Paul Peterson and Bryan Hassel, Brookings Institution, 1998.

Milwaukee, WI—After four years, voucher students had reading scores 6 Normal Curve Equivalent (NCE) points higher than the control group, and math scores 11 points higher. NCE points are similar to percentile points.