Disaggregation of standardized test data is essential in determining the differential effects of the school's climate and program upon groups of students formed by social class. In a recent secondary school study comparing the outcomes of two approaches to school assessment (the Connecticut School Effectiveness Survey vs. The National Study of School Evaluation), standardized Comprehensive Test of Basic Skills test data were disaggregated not only by social class, but also by class year, sex, program of study, and ability levels. In addition, data were analyzed for trends across three years of testing for each student. Findings included some surprising trends within the whole of the student population when disaggregated by the above variables, while holding the effects of curriculum (class year) constant. Results suggest that the "snapshot" approach in disaggregating test data by social class may mask the condition that the school may be improving or deterioriating on one or more of the standard indicators of pupil performance. Since patterns of pupil performance are often correlated with teacher attitudes about school climate, program and students, the more detailed the analysis of test data the more useful it may become in driving school improvement. (Author)