Predicting Punishment: Using Propensity Score Matching to Predict Who Receives Restorative Practices

Background

The connections between school discipline, students of color, and the criminal justice system is long standing (Cloward, 1966). A 2014 report from U.S. Department of Education’s Office for Civil Rights found that black students are three times more likely to be expelled than white students, and that Black boys were shown to be suspended at higher rates than any other group (U.S. Department of Education Office of Civil Rights, 2014). Upon reviewing data from 365 elementary and middle schools, Skiba, Sure & Williams, (2011), found that Black middle schoolers were nearly four times more likely to be referred to the office than White students, and that Black and Latino students were more likely to be expelled or suspended than their White counterparts for the same behavior. One approach to preventing exclusionary discipline is restorative practices (Gonzalez, 2012; Morrisson & Vaandering, 2012).

Methods

This study is a secondary data analysis of discipline data collected by Denver Public Schools for the 2016-2017 school year. The dataset includes information for 104,648 students. The sample is a population sample with no missing data. The treatment variable for this analysis is race, which is dichotomized into white or not white. Not white is all other categories of race, including mixed or multiple races. The dependent variable is receiving restorative practices or not.  Propensity score matching was used to estimate the effect of students’ race on receiving either punitive or restorative discipline outcomes.

Results

Of the sample of 104,648 students, 8,275 (12.6%) received some type of school discipline. Almost 70% of those were punitive interventions, with only 30.92% receiving restorative approaches. Upon conducting logistic regression, the final matching model was found to be balanced. None of the differences between the treatment group and the matched control group were statistically significant. The estimated treatment effect of receiving restorative practices found that students who were identified as white were more likely to receive restorative practices than students of color.

Conclusions and Implications

Findings demonstrate that, even when practices intended to improve school discipline outcomes are implemented, individual bias and decision making influence which students receive these practices. One implication is that, in order to reduce racial disparities in school discipline, it is necessary to consider the relational environment of the school and ensure individual buy-in to closing the disparity gap. Even when policies mandae restorative practices in schools, these policies do not appear to eliminate racial disparities in discipline outcomes. Further research is needed to examine which school-level factors impact decisions regarding who receives restorative vs. punitive discipline outcomes. Implications for practice include training of school staff, including implicit bias and culturally-informed teaching methods.