Abstract
The regression-discontinuity analysis was used to study the effects of ability grouping on the academic achievement and self-esteem of a student cohort, grades 4-6 in a large Texas school district that used a cutoff score for group placement. Iowa Tests of Basic Skills (ITBS) scores were reported in normal curve equivalents; therefore, gains referred to amount of learning expected. Scores of high-ability students with pretest scores covaried showed a significant increase after a year of ability grouping, but the effect diminished in subsequent years. Regular student scores overall showed a nonsignificant drop. High-ability students' scores on the Coopersmith Self-Esteem Inventory tended to rise while regular student scores tended to drop. Differences between the two groups were generally accounted for by pretest ITBS score differences. Students who had scored within one standard error of measurement above or below the criterion and who were prone to misclassification showed the same results. In general, students closest to the cutoff score were more affected by placement. High-ability students initially performed above expectations, but the effect was temporary.