Panel Paper:
Does Special Education Improve Outcomes for Students with Learning Disabilities? Evidence from New York City
*Names in bold indicate Presenter
In New York City, students with a learning disability (LDs) represent approximately 45% of the special education population. Moreover, LD is the most common disability identified after entry to kindergarten, with roughly 70% of LDs classified after the second grade. The timing of classification allows us to observe outcomes before and after initial receipt of special education services to explore whether academic and non-academic outcomes improve with receipt of services. Past studies analyzing the impact of special education services are mostly descriptive and only three previously published papers have claimed causal estimates at the K-8 level (Hanushek, Kain, & Rivkin, 2002; Cohen 2007; Morgan, Frisco, Farkas, & Hibel, 2010). This paper adds to this literature by examining outcomes for LDs in the nation’s largest school district, NYC.
We use longitudinal, student-level administrative data from the NYC Department of Education, which includes measures of race/ethnicity, gender, foreign-born status, limited English proficiency (LEP), free/reduced price lunch (FRPL) eligibility, grade level, test scores for the New York State Mathematics and English Language Arts (ELA) examinations, annual attendance, and school ID. For students with disabilities, the data include indicators for the thirteen disability classifications defined under the Individuals with Disabilities Act (IDEA), most importantly LD. The rich data allow for the comparison of academic performance before and after receiving special education services for students in grades three through eight – grades when the New York State assessments in mathematics and ELA are offered. With the large population of LDs in NYC schools, and the variation in the timing of receiving special education services, we estimate intent to treat effects of special education services.
We analyze academic outcomes of LDs receiving special education services linking test scores (or attendance rates) with an LD indicator, progressively introducing demographic characteristics, school effects, student effects, and finally a combination of school and student effects. Regardless of the specification, we find that LDs score higher on the NYS Mathematics and ELA examinations in all years following initial receipt of services. These results, however, vary depending on a student’s grade of classification, with larger effects for students classified in earlier grades. We find no statistically significant impact on annual attendance rates. We further explore subgroup effects (gender, race/ethnicity, poverty status, and receipt of LEP services) and possible mechanisms such as the receipt of inclusive versus non-inclusive services, school mobility, and school characteristics.