Public Policy Research



Vouchers Generate Academic Gains

Education: November/December 2000
Written By: Jay P. Greene
Published In: Intellectual Ammunition
Publication Date: November 1, 2000
Publisher: The Heartland Institute

Rigorous testing, class-size reduction, or school choice: Which reform produces the largest academic gain?

Recently released evaluations of private scholarship programs in four cities (Charlotte, Dayton, New York, and Washington) show that school choice significantly increases student achievement, particularly for African-American students.

Gains from Different Education Reforms
for African-American Students
(in Standard Deviations)
Rigorous Testing (RAND) 0.14
Class-size Reduction (STAR) 0.21
Vouchers 0.33

The evaluations in Charlotte, which I conducted, and the other three cities, conducted by a Harvard University team, took advantage of the fact that these private scholarships were awarded by lottery. Our research compared winners of the scholarship lottery, who were able to attend a private school, to the lottery losers, who remained in the public schools but were otherwise similar to the scholarship recipients.

We found that in Charlotte, after only one year, students who received a scholarship to attend a private school scored 6 to 7 percentile points (0.33 standard deviations) higher on standardized tests than did their counterparts who remained in the public schools. The Harvard team found a similar gain in Dayton, New York, and Washington after two years.

In all four cities the gains appear to have occurred almost entirely among African-American students, confirming the impression that those students may be particularly poorly served in public schools.

The consistency of the results across all four cities is impressive, as is the size of the gain. The gain achieved by the private scholarship programs is larger than the benefit observed in the Tennessee STAR study, which measured the effect of reducing class sizes by seven students. It also costs a lot less to provide school choice than to reduce class sizes by almost one-third.

What explains the fairly large test score gains from school choice?

We know the success is not explained by differences in the backgrounds of students and their families. The lottery helped ensure that we are comparing apples to apples.

We also know the scholarship programs' results were not achieved by private schools with more resources and better facilities than the public schools. In Charlotte, the average private school accepting scholarship students charged around $3,100, about one-half the amount spent per-pupil in the public schools.

Nor did the private schools succeed because they screened for advantaged students or expelled undesirable ones. In our sample, virtually no private schools screened for students, and no students were expelled or counseled out of their private school.

Lastly, we know that smaller class sizes did not account for the private schools' success. While the private schools did tend to have smaller classes, there was no correlation in our sample between student test scores and class size.

The advantage of the private schools may be choice itself. Perhaps by offering parents greater options, the scholarships simply allowed parents to find the right schools for their kids.

Significantly greater safety and discipline in the private schools may also have been important factors in the test score gain. Roughly one-quarter of public school students reported they felt unsafe at school, compared to just one-tenth of the private school students.

Whatever the reasons for the gains, a series of well-designed studies of school choice programs make clear that choice has significant academic benefits, particularly for African-American students.

Guest columnist Jay P. Greene is a senior fellow at the Manhattan Institute for Policy Research.

For more information ...

The Campaign Against Vouchers. Howard Fuller documents and refutes "known falsehoods" circulated by opponents of vouchers. (Institute for the Transformation of Learning, March 2000, 17pp.)

Request PolicyBot document #2184714.