The Department of Education’s (Education) quality control processes for data it collects from public school districts on incidents of restraint and seclusion are largely ineffective or do not exist, according to GAO’s analysis of school year 2015-16 federal restraint and seclusion data—the most recent available. Specifically, Education’s data quality control processes were insufficient to detect problematic data in its Civil Rights Data Collection (CRDC)—data Education uses in its efforts to enforce federal civil rights laws (see figure). For example, one rule Education used to check the quality of data submitted only applied to very large school districts, although GAO and Education’s own analyses found erroneous reporting in districts of all sizes. Education also had no rules that flagged outliers that might warrant further exploration, such as districts reporting relatively low or high rates of restraint or seclusion. GAO tested for these outliers and found patterns in some school districts of relatively low and high rates of restraint or seclusion. Absent more effective rules to improve data quality, determining the frequency and prevalence of restraint and seclusion will remain difficult. Further, Education will continue to lack information that could help it enforce various federal civil rights laws prohibiting discrimination.
Data Quality Issues GAO Identified in Department of Education 2015-16 CRDC Restraint and Seclusion Data
Note: All analyses used public-use file, except illogical data, which used a restricted-use file.
Officials in the nine school districts GAO visited lacked a common understanding of the CRDC’s restraint and seclusion definitions. Similarly, officials GAO interviewed in all three state educational agencies (Kentucky, Washington, and Wisconsin) and all seven stakeholder groups expressed similar concerns about the clarity of these definitions. For example, officials inconsistently interpreted the word alone in the definition of seclusion and, therefore, on whether to count an incident if a teacher was in the room. Absent clearer definitions, Education will continue to lack quality information on restraint and seclusion in public schools.
Officials in school districts GAO visited identified several benefits to collecting these data, including identifying patterns in student behavior and developing interventions that can reduce the need for restraint and seclusion. Officials also said that analyzing their data helped them identify needs for additional staff training and student support services.
Every 2 years, Education requires nearly all school districts to report incidents of restraint and seclusion. Generally, restraint is restricting a student’s ability to move, and seclusion is confining them alone in a space they cannot leave.
The House Committee on Appropriations’ explanatory statement accompanying the Consolidated Appropriations Act of 2018 included a provision for GAO to evaluate the CRDC’s restraint and seclusion data. This report examines (1) the effectiveness of CRDC data quality control procedures, (2) selected districts’ interpretation of CRDC’s restraint and seclusion definitions, and (3) selected districts’ use of data. GAO analyzed CRDC’s quality control processes for school year 2015-16, and interviewed officials from seven stakeholder groups and over 50 school and district officials in three states. GAO selected states, districts, and schools to obtain a range of perspectives on using restraint and seclusion data and interpreting CRDC definitions of restraint and seclusion. Selection criteria included changes in reported incidents year to year and laws requiring districts to report incidents to states.
GAO made six recommendations, including that Education expand its CRDC business rules to cover all districts, develop additional quality controls to address misreporting, address factors underlying misreporting, and refine and clarify its definitions. Education agreed with these recommendations.
For more information, contact Jacqueline Nowicki at (617) 788-0580 or nowickij@gao.gov.