Association for Public Policy Analysis & Management

Many studies have detailed the significant negative relationship between student disciplinary actions and subsequent student outcomes (e.g. Skiba, Michael, Nardo, and Peterson, 2002). Fewer studies have empirically assessed the role of teacher characteristics in determining student disciplinary actions; teachers serve as gatekeepers who report the severity and frequency of disciplinary actions. Given research that indicates that teacher race and gender can matter for minority students, this gatekeeper role may have implications for minority students’ outcomes (Dee, 2005; Clotfelter, Ladd & Vigdor, 2006; Bates & Glick, 2013). In particular, same race and/or same gender teachers may report student disciplinary actions differently. The racial composition of the teaching workforce is vastly mismatched to the current public school student population. Therefore, exploring this critical relationship between teacher demographic characteristics and student discipline outcomes is essential. This paper proposes to model predictors of student disciplinary actions, based on teacher characteristics, and their interactions with student characteristics.

This study uses statewide administrative data from North Carolina to address these research questions.  North Carolina offers roughly 20 years of data where students can be matched to individual teachers at the classroom level.  The state also collects disciplinary records for each student, including offenses committed by students and whether expulsion/suspension consequences are attached to each disciplinary incident. The state includes demographic data for both students and teachers. This is crucial in order to identify the effects we study here.

We use multiple strategies to identify the effects of teacher-student demographic match on student outcomes. First, following past researchers (Dee, 2005) we will conduct student fixed effects comparisons that uncover whether students are disproportionately likely to face disciplinary consequences in years (K-5) when their home classroom teacher is of a different race/sex group from them, compared to when they share their teachers’ characteristics. We will supplement these analyses with a second, novel strategy in this literature: identifying match effects off of twin differences.  This strategy will take advantage of the fact that until the mid-2000s, North Carolina state policy encouraged that twins be assigned to separate classes.  In some cases, this will mean that one twin is assigned to a same-race (gender) teacher while another twin is assigned to a different-race (gender) teacher.  Exploring whether these different assignments are then associated with different discipline rates provides a strong alternative estimation of match effects.

To explore these issues at the secondary level, we adapt an analytic strategy used in past literature on the effects of students’ being matched to tenured vs. adjunct college instructors (Bettinger & Long, 2010). This strategy uses deviations from the long-term trends in the race/sex composition of faculty at a given school-grade level to determine whether, for instance, male students have lower disciplinary incident rates in years when a higher share of teachers in their specific school-grade combination are male compared to the historical average.

These analyses should have important implications for debates over the importance of diversifying the teaching workforce.