I’m excited to feature a new collaborating author on the SD Notebook – Jeremy Anderson is a Ph.D. candidate at Penn State and a colleague on the voluntary integration research team led by Erica Frankenberg.

Racial disparity in school discipline is a national problem that is especially troubling when it comes to exclusionary school discipline practices such as suspension and expulsion. On ProPublica’s interactive map of school discipline rates, for example, it is difficult to find any district in the country where Black students are not more likely to be suspended than white students. (See if you can find one!)

A collection of new research, using sophisticated statistical analyses, aims both to explain the causes of the racial discipline gap and to understand its impact on student outcomes. This post features a recent article that focuses on causes:

• “Understanding the Racial Discipline Gap in Schools” by Maithreyi Gopalan and Ashlyn Aiko Nelson in AERA Open

The authors had access to an enormous amount of data: discipline rates for all students in Indiana public schools from pre-kindergarten to 12th grade for 6 years, starting in the 2008-09 school year. That’s over 3 million students and over 7 million individual student-year observations.

Their data set included a wide range of student and school characteristics, so they used these characteristics to try to build a statistical model that would explain the discipline gap. Their findings underscore not only the pervasive nature of school discipline disparities but how those disparities interplay with school and district segregation. Specifically, they identify the nonrandom sorting (e.g., segregation) of Black and Hispanic students into school districts across Indiana as a key mechanism underlying higher suspension and expulsion rates as compared to their white classmates. Three main findings that I found particularly powerful were that: 

1) The uneven sorting of students of color across the state corresponds to the uneven distribution of resources, teacher quality, and other factors that are associated with adverse disciplinary outcomes. So, various forms of segregation are related to each other.

2) If Black students were re-sorted into districts predominantly attended by white students in the state and received disciplinary sanctions at the average rates of those districts, the Black-white disciplinary gap would decline by 11% to 25% across grade levels. This provides some evidence that school integration could lead to a decrease in the Black-white discipline gap.

And that…

3) Disciplinary gaps between Black and white students emerge as early as pre-kindergarten, widen, and then persist across grades even after controlling for extensive student and school level characteristics

Maithreyi Gopalan (a professor at Penn State) discussed the article with us and highlights some of the findings she finds to be most salient:

What is the headline from your article and why is it important?

Using a large, comprehensive, statewide, student-level administrative dataset including all students in Indiana over a 6-year period, we find that Black and Hispanic students are suspended more often and longer than their white peers. Large Black-white disparities exist within-schools and within-districts in contrast to a few other recent studies from other states that show disparities exist only across schools and districts and not within. In contrast, Hispanic-white discipline disparities are smaller and stem mainly from cross-school and district variation.

We are also one of the first to show that the racial discipline gap widens with grade progression and to use a long panel and a counterfactual simulation/decomposition technique to account for a richer set of control variables than prior research. Finally, we also uncover interesting descriptive associations between segregation and punitive disciplinary practices. 

You found that that district differences in disciplinary practices reduce the racial discipline gap, but do not eliminate it. Instead, racial gaps still exist at the school level. I took this as evidence that unconscious racial discrimination may still be salient over and above a district’s attempt at changing policy/practice. To what extent is this an accurate/fair interpretation of your findings?

Unfortunately, the question of whether discipline disparities reflect discriminatory practices is incredibly hard to answer primarily because objective measures of student behavior are rarely available for research. Most of the behavior measures are teacher-reported and research shows how teacher-interpretations about student behavior might itself reflect some implicit biases. However, our analysis finds the persistence of racial disparities in disciplinary outcomes within schools and a counterfactual simulation analysis that provides suggestive evidence that discriminatory practices likely underlie some portion of the racial discipline gap.

You found that “about a third of the Black-White discipline gap remains unexplained” (p. 14)- Was this larger or smaller than you might have expected? What might be making up that remaining unexplained portion of the discipline gap?

It was larger than we expected. After controlling for an extensive set of controls at the student- (including past disciplinary history), school-, and district-level we would have imagined that unobserved heterogeneity (i.e., the “unexplained” portion of the gap) would be much smaller, which is the case for Hispanic-white gaps but not for Black-white gaps in our sample. We believe that differential treatment discrimination as well as unobserved omitted variables (including potential behavioral differences) make up the remaining unexplained portion. This is where future thoughtful qualitative and mixed methods research might be able to add nuance to the underlying mechanisms at play.

Your study uses a non-truncated dataset (i.e., it includes all students, not just those who experienced a disciplinary consequence). Can you explain what, specifically, this allowed you to do in your analysis and why this approach is important in school discipline research writ large.

Sure. Many past studies of school discipline have used datasets that include information only on students who have received some form of disciplinary infraction or disciplinary outcome such as suspensions or expulsions. Most of those analytical samples do not include any information on students who never received any disciplinary infraction or exclusionary disciplinary outcomes. That is a major limitation that our study overcomes. Studies relying on nationally-representative datasets use self-reported measures of school disciplinary outcomes in the analysis that are also not that reliable.

Second, many past studies have also used selective data from some grades and/or convenience samples of schools and districts. But our data covers all students (from pre-kindergarten to high school) from a statewide administrative dataset. It also includes precise measures of length of in- and out-of-school suspensions, count of suspensions, and others over a long time period on multiple cohorts of students that is far more extensive than prior data sources which helps us estimate the discipline gap more precisely.  

Is there anything else that you’d like to add?

Given the complex interplay of several multifaceted factors that drive racial disparities in school discipline, studies should strive to use such extensive datasets to understand the various mechanisms that might be driving those disparities. Schools, districts, and state departments of education must also be incentivized to collect such extensive datasets like the Indiana Department of Education, and share it with researchers to move research forward in this area. Without good monitoring and tracking that such data enables, we will not be able to diagnose, and therefore fix these persistent issues.  

Racial disparities in school disciplinary measures have been widely documented and the research base providing evidence of these disparities has continued to broaden. This article joins a growing amount of research that then connects the discipline gap to student outcomes. In fact, the same lead author, Maithreyi, just published another recent piece that found a statistically significant relationship between Black-white and Latinx-white discipline gaps and achievement gaps.

In addition to Maithreyi’s work, research from Francis Pearman II, F. Chris Curran, Benjamin Fisher, and Joseph Gardella has shown that racial disparities in school discipline are associated with the achievement gap between students of color and white students partially as a result of teacher biases and feelings of isolation at school. Specifically, that study found that “districts with larger Black-White discipline gaps have larger Black-White achievement gaps and vice versa.” Their study was among the first to demonstrate this relationship at a national level.

Yet another study from Kaitlin Anderson found that “less exclusionary consequences…are associated with better outcomes” as measured by standardized tests and grade retention and that, conversely, exclusionary discipline adversely affected grade retention for under-served students.  

These disparities become even more pronounced (and troubling) when it comes to students of color with disabilities as documented in a recent report by the U.S. Commission on Civil Rights. Using data gathered by Dan Losen at the UCLA Center for Civil Rights Remedies, that report found, for example, that Black students with disabilities face exclusionary discipline at a rate that is 3x higher than white students with disabilities.   

While evidence continues to grow supporting the link between racial disparities in school discipline and adverse educational impacts on students of color, federal support for school districts looking to remedy this problem has diminished. Federal guidance issued by the U.S. Department of Justice and the U.S. Department of Education in 2014 was rescinded last year by the Trump Administration. Without federal guidance, there is significant concern that districts may decrease or even abandon plans to address the problem of racial dis-proportionality in exclusionary discipline.

In “How to be an anti-racist,” Ibram X. Kendi writes that there’s a “sinister implication in achievement-gap talk: that disparities in academic achievement accurately reflect disparities in intelligence among racial groups.” Clearly, the racial discipline gap is analogous- without more understanding of its causes and effects, the sinister (and indeed racist) implication passes uncontested. In contrast, the research here provides solid evidence (based on lots of data and robust statistical analyses) to illustrate what we know is true: it’s not the kids. It’s school policies that segregate and it’s unexamined racial bias – it’s us.