The purpose of this study was to describe, analyze, and compare the results of five commonly used approaches to measuring school-level performance for the subgroup of students with disabilities. Using the reading and mathematics scale scores of students in grades two, four, and six on the Comprehensive Test of Basic Skills (CTBS) and the California Achievement Tests (CAT/5) within a large school district in the mid-Atlantic, five approaches were applied to classify schools as high-performing or low-performing based on the subgroup students with disabilities. The approaches applied included three status approaches (cross-sectional, cross-sectional with confidence interval, and three-year rolling average) and two value-added approaches (unadjusted and adjusted for student demographics). The characteristics of schools classified as high-performing and low-performing based on the performance of students with disabilities using each of the approaches were explored. Each approach was also examined for its reliability, fairness, inclusiveness, and usefulness.

Significant differences in the performance of students with disabilities based on socioeconomic status were observed across all grade levels, but no differences in gain scores were consistently observed. No significant differences in reading and mathematics performance were consistently found across grade levels based on the disability group and LRE of students.

Overall, none of the accountability approaches employed reliably rated schools based on the performance of students with disabilities. Even within the same subject area and using the same approaches, schools labeled as high-performing for students with disabilities one year were labeled as low-performing the following year and vice versa.

Schools classified as high-performing using the cross-sectional and three-year averaging approaches demonstrated some bias against high-poverty schools and schools with large percentages of minority students. Schools classified as high-performing using the cross-sectional with confidence approach disproportionately identified schools with small numbers of students as high-performing. The value-added approaches were least biased in terms of socioeconomic status and the percentage of minority students, but were limited in their inclusiveness. The usefulness of all the approaches was limited by complicated assessment and accountability policies and the use of non-standard accommodations. All analyses were affected by the small number of students with disabilities in the subgroup