Panel Paper:
Racial Disparities in School Disciplinary Outcomes
*Names in bold indicate Presenter
This research study proposes to use statewide, student-level data on Indiana public school students from the 2008-09 through 2013-14 academic years to examine racial disparities in exclusionary, punitive disciplinary actions in school. In particular, we propose to examine the extent to which disproportional disciplinary outcomes may be attributed to differences in student behavior, as compared to differences in student characteristics (e.g., poverty status) that are correlated with both race and special needs status, or differences in the types of teachers, classes, schools, and school district policies to which students of color may be differentially exposed. We will examine the extent to which child versus school characteristics explain variation in children’s disciplinary outcomes, lending important insights into how school environments may influence the disciplinary gap and exclude students of color from educationally productive experiences that promote later life outcomes.
Our study will use statewide, student-level panel data on all K-12 Indiana public school students in the 2008-09 through 2013-14 academic years. We propose to use panel data techniques that include school and/or district fixed effects to examine the extent to which variation in disciplinary outcomes is explained by within- versus across-school variation. Our dependent variables of interest will consist of a series of dummy variables indicating whether each student in each year receives a suspension/expulsion for any of the potential 17 categories of behavior included in our dataset. We will employ the Blinder-Oaxaca decomposition technique, which will allow us to disentangle the extent to which disciplinary disparities may be attributed to observed student and school characteristics, rather than to unexplained variation. In essence, the Blinder-Oaxaca decomposition method allows us to estimate the predicted level of suspensions and expulsions for students of color if they had the same distribution of characteristics as white students (i.e., if the poverty and special education distributions of students of color within and across schools were similar to those of white students). The difference between actual student suspensions/expulsions and predicted suspensions/expulsions from the model, then, may be attributed to unobserved factors including discrimination. Given that our unique panel dataset enables us to examine a richer set of student data—including rich data on student special education categorizations—than has been examined previously in this literature, we argue that the student characteristics in our data explain a larger portion of the variation in student outcomes that may be attributed to observed student and school characteristics.