empirical evidence with which to evaluate the theoretical claims made both for and against vouchers, and the evidence that does exist is often hotly contested. While limited data are available that can be used to test predictions about voucher programs directly, theoretical predictions can be tested by analysis of data for existing public and private schools. Epple et al. (1998) test the aforementioned predictions about stratification made by Epple and Romano and find that all predictions regarding stratification within and across the public and private school sectors are supported by the data.14
The foregoing suggests that charter schools and vouchers, rather than interdistrict and intradistrict choice programs, are the choice options most worthy of further exploration as vehicles for improving poor-performing city schools. Charter schools are in effect a naturally occurring experiment, although one that is not being as fully and systematically evaluated as it might be. Also, the fact that charter schools have unequal access to capital funding means that they do not face a level playing field with traditional public schools.
Existing voucher programs are so small that they are not ever likely to yield the kind of answers about the effects of vouchers and the most effective voucher designs that would be necessary to allay the concerns of those who question vouchers, not on legal grounds, but on the grounds of their unproven impacts on school performance and stratification. This raises the question of whether it is time for a large-scale experimental demonstration project with school vouchers. The committee wrestled long with this issue and discusses it further in Chapter 9.