Search Results: (1 of 1 records)
|REL 2021049||Why school accountability systems disproportionately identify middle schools' SWD subgroups for TSI
The purpose of this study was to understand why middle schools in two mid-Atlantic states were apparently disproportionately identified for targeted support and improvement (TSI) based on the performance of their students with disabilities (SWD) subgroups. The study used publicly available data to examine differences across school levels – elementary, middle, and high – and between all students and SWDs. It is relevant to compare all students to SWDs because accountability systems designate TSI schools as those with subgroups of students that perform poorly relative to all students. The study focused on two aspects of school accountability systems: (1) the number of schools in each school level in which enough SWDs took state assessments for the school to be held accountable for the academic proficiency of its SWD subgroup, and (2) the average performance on accountability indicators, by school level and subgroup. In both states, SWDs in middle schools were over 20 percentage points more likely to take state assessments than were SWDs in elementary or high schools, meaning that middle schools’ SWD subgroups were substantially more likely to meet states’ minimum sample size requirements for including academic performance in accountability scores. Also, SWD subgroups across school levels consistently performed worse than all students on academic proficiency accountability indicators. Taken together, these findings suggest that middle schools’ SWD subgroups are more likely than elementary or high school SWD subgroups to be identified for TSI because their accountability scores are more likely to include academic proficiency indicator scores – which tend to be substantially lower than the all students group’s academic proficiency indicator scores. This research suggests that when designing school accountability systems, state education agencies may wish to consider how sample size affects estimates of subgroups’ performance, and in particular how sample size exclusions may mask poor performance for small subgroups.
1 - 1