An index of the effectiveness of classrooms, schools, districts, or other educational units can be developed by, first, determining the differences between the actual achievements of individual children and the achievements expected on the basis of regression equation predictions and, second, averaging these differences for all children in the unit. Measures of each child's school ability are major contributions to the regression predictions, but are not always available. A study was undertaken to compare three methods for developing an index without such measures. The study data were the results of ability measures and statewide achievement tests for almost 70,000 children enrolled in grades 3 through 6 in 95 elementary schools. The complete index accounted for student ability, ethnicity, gender, age, and home language. The first alternative index simply omitted the ability variable. The second alternative substituted first-grade mathematics test scores for the ability measure. The third alternative used differences between first- and second-grade achievement as the substitute measure. The substitute correlating most highly with the complete index was the version omitting the ability measure altogether. The potential value of all three methods was indicated by their relatively high correlations. These correlations are presented in two tables. (PGD)