Vouchers have not helped students learn
There are no comments for this page [ Add Comment ]Friday, August 31, 2007
By Eric Jacobson
Guest commentary
In November, Utah voters will decide on a voucher program that will provide tuition for students moving from public to private school. A key goal of the program is the improvement of student learning. Will it work?
Will vouchers increase academic achievement? Evaluations of voucher programs in other places may help answer this question.
Voucher users: To determine if vouchers help students, researchers compare the test performance of students who use vouchers with a comparable group that does not. If test scores of voucher users are higher than those in the comparison group, then the program can be judged to have benefited the users. It is not easy, however, to find a group which is equivalent to the voucher students and can serve as a fair comparison.
Consider voucher programs in Milwaukee and Cleveland, both serving primarily low-income students. Using one comparison group, an early evaluation in Milwaukee found that voucher users had higher math scores (Greene, et al., 1996). As was found later, however, the voucher users looked good because they were compared to especially weak students who would have been expected to do poorly. A second, larger evaluation, using a more equal comparison group, found that vouchers did not improve test scores (Witte, 2000).
Early evaluations in Cleveland also produced contrary answers -- vouchers improve scores (Greene, et al., 1998); vouchers do not improve scores (Metcalf, 2000). The different answers depended on different comparison groups. According to the official project report, after seven years voucher users did not have higher average scores, but did better on language and social studies tests (Plucker, et al., 2006).
In educational research the "gold standard" is the randomized field trial (RFT). With an RFT, students are randomly assigned to voucher and non-voucher groups. The random assignment insures that the two groups do not vary systematically and that the non-voucher group can be fairly compared with the voucher group.
Opportunities for RFTs arose in New York City, Washington, D.C., and Dayton, Ohio, in the late 1990s. Private foundations in each city offered voucher scholarships for low-income students, but only had sufficient funds to support a fraction of potential users. Random lotteries were held to determine which of the qualifying applicants would receive vouchers. The conditions for an RFT were met with lottery losers constituting the randomly determined comparison group.
After three years, averaging across all students, there were no differences between voucher and non-voucher students on national tests (Howell & Peterson, 2002). When students were categorized into ethnic/racial groups, however, an advantage was found for African-American students in math. No other ethnic group, including Hispanics and non-Hispanic Caucasians, was helped by the vouchers. Subsequent analyses suggest that even the specific gains shown in math by African-Americans may be very small (Kreuger & Peu, 2003).
Evaluators of a new voucher program in Washington, D.C., report no difference between voucher users and nonusers in the first year (Wolf, et al., 2007).
Nonusers: Some students take vouchers when offered, but most don't.
If we are concerned about the education of all students, then we must consider how voucher programs affect these nonusers. Two possibilities have been conjectured.
An exodus of voucher-using students may be threatening to public schools. To compete against this threat, schools may improve their programs. In this way the voucher program would benefit nonvoucher students who stay in the public school. On the other hand, there is some evidence to indicate that schools with more parental involvement produce higher test scores, and that students do better if their classmates are also doing well. If the voucher program causes the better students and/or the students with more involved parents to leave, then these advantages will be lost from the public schools and non-users will suffer.
In 1999, Florida instituted a school accountability system, in which low-scoring schools (F grade) were put under sanctions and their students were offered vouchers for other schools. Test scores in F schools improved.
That is, students who stayed in F schools and did not use vouchers did better (Greene & Winters, 2003). Improvement could have been due to the threat of vouchers, as some have argued, but it could also have been due to other factors. Figlio and Rouse (2005) sorted through many of the possibilities and concluded that the social and political stigma of receiving an F, and not the voucher threat, produced the improvement in low-rated schools. Low school rankings in other states, where no vouchers are offered, produce similar test score improvements (Carnoy, 2001).
Thus, there is no evidence yet that voucher threats improve public schools. Is there evidence to the contrary, that they harm students in public schools? In Cleveland, the students who used vouchers to leave public schools were better students and they came from better schools (Plucker, et al., 2006). In Milwaukee, parents of voucher-using students were indeed those who had had more involvement in the public schools (Witte, 2000). In neither place, however, was there direct evidence that remaining students had lower achievement scores.
Conclusion: Voucher programs have demonstrated almost no academic benefits in other places. Will Utah be different?
Jacobson is retired from Weber State University, where he was director of academic computing and later did evaluation research in the math department. He is currently involved with the evaluation of a teacher development project in local schools.
Text 
