This proposal describes four inter-related projects that study the effects of educational attainment and school characteristics on children's outcomes. The first project will analyze the effects of school """"""""quality"""""""", measured by class size and other observable characteristics, on high school drop out behavior and college enrollment decisions. The analysis will be based on the High School and Beyond data set and the 1988 National Educational Longitudinal Survey. The second project will investigate the effect of school quality on measured test score performance. We will develop a data base of state-level observations for the 1970-92 period that includes SAT scores, student characteristics, and school quality measures. A key aspect of the project will be the specification and testing of models for state average test scores that are consistent with appropriate models for individual achievement and test participation. The third project will use the results of a recent randomized experiment to evaluate the evaluation methods used in the educational production function literature. We have obtained data from a large-scale controlled experiment that randomly assigned elementary-school students and teachers to different size classes in Tennessee in 1985. Previous studies of this experiment find that students in smaller classes had higher average test scores. We will contrast the results of the experiment with results from non-experimental methodologies applied to a similar population of students who wrote the same standardized test, to learn whether the methods used in traditional observational studies can correctly identify the true relationship between test scores and class size. The emphasis will be on exploiting the social experiment to test identifying restrictions that are imposed in non-experimental situations. The final project will study the effect of school resources on the subsequent earnings of students. This study will use 1990 Census microdata files to update and extend earlier analyses by the authors. By pooling the large samples available in the 1980 and 1990 Censuses we can isolate the role of changing school quality in the rise in the measured """"""""return"""""""" to education over the past decade, and compare the effects of school quality for earlier and later cohorts of workers. We will also explore the impact of nonlinear specifications of the earnings, education, and school quality relationships.
