Statewide assessment data available from two school years, two grade levels, and five sources (achievement tests; student, principal, and teacher questionnaires; and principal interviews), were aggregated to more closely investigate the relationship between student/school characteristics and student achievement. To organize this large number of distinct data sets (old and new), all questionnaire and interview data were assigned to conceptual clusters. Separate analyses were conducted for each year (1975-6 and 1976-7) and for each grade (4 and 8). A single indicator of school achievement was used--mean standardized scores on the subtests of the Iowa Tests of Basic Skills. In response to the problem of different collection times for different data sets, the primary analyses were conducted on the 1976-77 set. A series of strategies were employed to overcome the statistical problems associated with a large number of variables and limited degrees of freedom. Despite these conceptual, methodological, and statistical problems, research with existing data bases was worth the effort; however, the appropriateness of an existing data base for answering new questions depends on the similarity between the original and the new questions. (CP)