Identifying Levers to Disrupt School Expulsion: An Exploration of School and District Factors

Background: Research in the post-“zero tolerance” era problematizes the use of expulsion and other forms of exclusionary discipline (e.g., in and out of school suspensions). Although most research focuses on the most common forms of exclusionary discipline (in- and-out of school suspensions), expulsions are of interest because they are reserved for the most serious disciplinary infractions and, in some states, are accompanied by legal protections for students. Regardless of the form, evidence consistently indicates unacceptably high rates of exclusionary discipline that are highly variable across schools and, of crucial import, that disproportionately affect low-income students of color.

There are several sources of variation in school expulsion, including those emanating from within and across schools and districts, respectively (Skiba et al., 2013). Most research has considered within- and across-school variation in identification of and response to disciplinary infractions. Findings suggest that school-level characteristics (e.g., poverty and historically underserved minority rates) positively relate to exclusionary practices, as do principals’ preferences for exclusionary discipline. District-level variation in expulsion is surprisingly under-investigated, given that districts often play central roles in setting guidelines for school disciplinary practices. Moreover, multi-level and multi-pronged strategies are posited to be avenues to fair, consistent, and responsive approaches.

Drawing on a unique state-level data set, the current study focuses on schools as the central unit of analysis and attempts to clarify to what extent school and district factors influence on school expulsion rates.

Method:  

Data sources: Multiple state-level administrative archives collected in California during the 2016-2017 school year serve as data sources. The archives, linkable through unique school identifiers, include information on school-level expulsions, and school and district characteristics for 9,640 schools and 1,009 districts in the state of California.

School-level expulsion rates serve as key dependent variables, including the total expulsion rate as well as rates for key subgroups. Following prior research, key school characteristics include indicators of student body composition (proportions of students by racial background, gender, enrollment in free or reduced price lunch, English learner status, disability status, involvement in foster care, and enrollment in McKinney-Vento programming (i.e., homeless or migrant); school structural features (total enrollment, grade span, and type (e.g. charter, alternative program)); and absentee rates. Key district characteristics include district size, average expulsion rate, absentee rates, and district-level student body composition paralleling those collected at the school-level.

Analytic approach:  Multilevel models were implemented to estimate the extent of school and district variation in expulsion outcomes. These included unconditional, school-controlled, and district-controlled estimates as well as analyses of model fit.

Findings:  Depending on expulsion outcome, estimates indicate significant across-district variation (intraclass correlations ranged from 0.11-0.18). Addition of school characteristics—particularly school poverty and absentee rates—reduced the across-district variation by no more than 10%. Remaining variation was almost fully accounted for by the addition of district characteristics, notably district-level expulsion rates.

Implications: This study provides evidence that district characteristics, controlling for school characteristics, shape expulsion rates. Paralleling school-level research, widespread use of expulsions within a district appear to shape schools propensity to expel students.