Association for Public Policy Analysis & Management

This study aims to identify school characteristics or policies that may place students at greater, or more disproportionate, risk of suspension. Using a unique administrative dataset from a large, urban school district, this project assesses school “effects” on suspension risk for high school students between 2009 and 2013. I address several methodological hurdles often present in school-level analyses of discipline by using student-level administrative data. First, I account for preceding characteristics of students in high school addressing concerns about school variation in suspension rates due to variation in student populations and behaviors. Most importantly, I focus this analysis within a single, large district that allows the identification of school effects above and beyond general associations between school characteristics and suspension rates. Most studies of school suspension rates explore characteristics but are unable to identify unique school effects due to sample size. More, given the sample size, I am able to identify the heterogeneity in school effects as associated with student characteristics.

The first analysis uses a multi-level, variance components analysis highlighting the between-high-school variance in suspension rates for 4 cohorts of students entering 9^th grade between 2009 and 2012. This analysis also focuses on the influence of the racial context of a high school on a student’s risk of suspension. By identifying similarly defined infractions associated with different levels of punishment I am able to estimate a student’s likelihood of suspension for similar behaviors, net student characteristics. I then examine the between and within school variation of these likelihoods identifying school characteristics, particularly racial context, that may be associated to this variance. Initial findings, using multi-level regression models with interaction effects, suggest that certain concentrations of race-groups in schools are associated with student suspension risks. Yet these relationships do not change the prevalent finding that Black boys are the most likely to be suspended, net school and student characteristics. These findings suggest that while there are compositional factors of schools that influence how suspensions are administered, even within the same district discipline code, unobserved factors remain influential in the disproportionate rates of suspension for students.