9511347 Kremer There have been many retrospective statistical studies of the effect of educational inputs on school performance. However, these retrospective studies will be biased if schools with different educational inputs systematically differ in other, unmeasured ways which are correlated with school performance. For example, schools with better physical facilities may also typically have parents who are more advantaged or more committed to education, and this parental input, rather than the physical facilities themselves, may be responsible for the correlation between test scores and physical facilities. This project conducts a randomized trial which can overcome many of the statistical problems of past studies. The project carries out one of the first randomized trials on educational inputs and school performance. The trial is based on the work of International Christian Humanitarian Services, (ITS), a Kenyan GO. ITS provides selected rural primary schools with textbooks, exercise books, uniforms, classroom construction, desks, and medical services. The investigator will randomly assign a group of candidate schools to either a control group or treatment group. Schools in the treatment group will receive both educational and medical services from ITS. Schools in the control group will receive medical services in return for participating in the program and cooperating with their monitoring and evaluation efforts. The project compares the change in text scores and dropout rates between the two groups. Since different components of the assistance package will be provided in different years, and will affect different classes of students, the investigator should be able to separately assess the impact of the different inputs. The project also examines whether the impact of the additional resources differs with students gender or socio-economic status, or with teachers training and pedagogic techniques.
