Association for Public Policy Analysis & Management

Since the early 1990s, many school districts, particularly those serving children from low- income families, enacted stricter “zero tolerance” disciplinary policies relying heavily on out-of- school suspensions. Research has shown that suspensions and expulsions are disproportionately centered on economically disadvantaged students, students with emotional and behavioral disorders, and minority students (Osher, Bear, Sprague, & Doyle, 2010). Recently, districts across the country have started to repeal zero tolerance policies in favor of more tempered disciplinary responses that keep students in the classroom. These changes offer an ideal opportunity to assess how changes in disciplinary policy affect the use of disciplinary activity.

In the 2012-13 school year, two large cities —New York and Philadelphia— made dramatic changes to their school discipline codes of conduct. These changes emphasize intervention rather than exclusionary practices, such as suspensions and expulsions, and give school principals greater discretion in responding to certain student misconduct. In this paper, we leverage longitudinal student-level data to examine how changes in disciplinary policy affect the use of suspensions within the district, the effects on disproportionality of suspensions (by race/ethnicity, gender, English language learners, economic disadvantage and prior achievement), and whether the policies were implemented differently across schools in the same district. We pose the following research questions: What are the suspension patterns in the years before and after discipline policy reform? Did the policy change affect suspensions? Does the policy change reduce or exacerbate the disproportionality of suspensions? Does exclusion become more or less concentrated among certain types of students? Do disciplinary responses vary based on differences in school context?

We compare suspension usage before and after the policy change in both cities. We consider both patterns of total suspensions as well as suspensions for infractions subject to different disciplinary responses post-policy (i.e. fewer suspension days or no suspension at all). First, we estimate school-level models to investigate variation in the change in suspensions at the school level following the policy change. Next, student-level models estimate the change in the individual probability of suspension before and after the policy change, controlling for time-varying individual characteristics, time-varying school characteristics, and student fixed effects. Finally, we compare changes in the probability of suspension for students in regular public schools to a comparison group of charter school students who were not subject to the districts’ disciplinary code of conduct. We extend the most controlled specification to include interactions with student characteristics, including race and ethnicity, language spoken, free and reduced price lunch, special education status, and prior academic achievement, to identify differences in patterns of suspensions among subgroups. Lastly, we consider variation in suspension patterns by school context using climate information from school environment surveys in both cities.

Among the most pressing policy questions in recent years is whether district-level policy changes can reduce suspension usage and mitigate racial and ethnic disproportionality in school discipline. To inform this question, this paper provides evidence from two large urban districts that recently made dramatic changes to their discipline codes of conduct.