Search EdWorkingPapers

Kaitlyn G. O'Hagan

There is limited empirical evidence about educational interventions for students experiencing homelessness, who experience distinct disadvantages compared to their low-income peers. We explore how two school staffing interventions in New York City shaped the attendance outcomes of students experiencing homelessness using administrative records from 2013-2022 and a difference-in-differences design. We find suggestive evidence that one intervention, which placed social workers in schools, increased the average attendance rates of students in shelter by 1-3 percentage points after 3-5 years. We discuss implications for the importance of non-instructional school staff and strategies to serve homeless students.

More →


Middle school transitions are increasingly required, despite documented negative effects on general education students (GENs). We explore if and how the move to middle school differentially affects students with disabilities (SWDs), a large and low-performing group of students. Using an instrumental variables strategy and NYC data on nine cohorts of students, we find the middle school transition causes a 0.29 standard deviation decline in SWD math performance, a 0.16 standard deviation decline in ELA performance, and a one percentage point increase in grade retention. However, after accounting for potential mediators (e.g. peer cohort stability) effects are similar for SWDs and GENs, suggesting the need to ease the middle school transition for all students.

More →


There has been an explosion of research on racial disproportionality in special education. Some recent research shifts the focus from the role of student characteristics alone to inquire whether school context moderates findings (e.g., is a Black student less likely than a White student to receive special education services as the proportion of a school’s Black students increases?). We significantly extend this emerging literature using eight years of elementary student-and school-level data from NYC public schools, examining more school contextual moderators, expanding racial categories, and distinguishing between cross-sectional and over-time differences. We find many more moderators than previous research has identified and these school context factors appear to be particularly salient for the classification of Black students.

More →