The New York State Education Department conducts a Pupil Evaluation Program (PEP) in which each year all third, sixth, and ninth grade students in the state are given a series of achievement tests in reading and mathematics. The data accumulated by the department includes achievement test scores, teacher characteristics, building and curriculum data, and census data. This extensive data is organized so that analyses may be made at the building or district level. Using the data from grades three and six, a factor analytic study of the achievement data in reading and mathematics is discussed for each grade and the development of achievement criteria. Using these criteria and regression analysis, the variance among controllable and uncontrollable school characteristics is partitioned to study the independent contribution of specific variables of interest and nonlinear and interaction effects. (Author/DEP)