Textbooks, class size, and test scores : evidence from a prospective evaluation in Kenya

The program reduced the dropout rate by 10%, increased transfers into the schools by 20%, and reduced transfers out by 8%, leading to an overall enrollment increase of approximately 40%. It cut absence rates by one-third. There was no difference in average test scores between program and comparison schools. One hypothesis is that the 40% increase in enrollment created over-crowding which offset the direct impact of the program. Support for this hypothesis is provided by the fact that the test scores rose in grades in which there was little increase in enrollment, and in schools in relatively isolated areas, which therefore could not attract many new students. Instrumental variables estimates suggest that in the absence of enrollment increases, the program would have increased test scores by one-third of a standard deviation, but that each additional student reduced test scores by 6% of a standard deviation. The estimated effects imply that textbooks are ten times as cost-effective in increasing test scores as reductions in class size. (Author abstract)