A deviant case analysis pilot study analyzed California local education agency data to determine the usefulness of regression analysis in predicting change in achievement from 1984 to 1989 and identified outliers or districts that show greater achievement changes than would be expected given changed demographic conditions. This report on the Successful Indicators Study discusses some previously identified statistical and methodological problems associated with the use of regression, presents the findings of the pilot study, and recommends alternative methods for selecting case study sites. Focus is on developing indicators of conditions and programs within a metropolitan school district that predict success. Data from the 1980 Census, the 1989-90 California Basic Education Data System, and the California Assessment Program tests for the school years 1984-85 through 1989-90 are used. Reading and mathematics scores for grades 3, 6, and 8 from 1984-85 to 1989-90 were used. The findings indicate that the regression procedure has not helped identify local education agencies that are doing well and that have experienced large changes in the demographic conditions under which they operate. Use of a combination of qualitative and quantitative methods is recommended to identify successful local education agencies. Included are six tables and two graphs. (RLC)