Using multilevel modeling and data from 10 Latin American countries, this paper provides new evidence on the relative effectiveness of public and private schools. There are substantial differences in the achievement of private and public schools, usually around one-half a standard deviation. A small portion of these differences is accounted for by the higher socioeconomic status of students in private schools. A substantial portion is explained by the varying peer-group characteristics in private and public schools. After accounting for peer characteristics, the average private-school effect across all 10 countries is zero, though with some variance around this mean (the effects range between-0.2 and 0.2 standard deviations). Evidence on selection bias is inconclusive, but the paper argues that these effects may constitute an upper boundary to the true effects. These results may help in predicting the impact of relatively small voucher programs. In this respect, transferring students to private schools would yield substantial increases in student achievement, remembering, however, the strong evidence that most of these gains would arise from the beneficial effect of "better" peer groups, as opposed to greater school effectiveness. Hence, the policy usefulness of the current empirical evidence should not be overstated. (Contains 68 references, 7 tables, and 2 figures.) (RT) Reproductions supplied by EDRS are the best that can be made from the original document.

These findings provide encouraging evidence that a package of multisector interventions has the potential to produce reductions in childhood stunting.

Our results contribute to the growing base of evidence regarding the validity of nonexperimental designs. We demonstrate that the CITS design can, in our example, produce internally valid estimates of program impacts when multiple years of preintervention outcome data (test scores in the present case) are available and when a set of reasonable criteria are used to select comparison organizations (schools in the present case).

No abstract