Association for Public Policy Analysis & Management

In the face of a rapidly diversifying student population, recognition of the numerous benefits to students—particularly students of color—of a racially and ethnically diverse teacher workforce (e.g., Dee, 2004; Egalite, Kisida, & Winters, 2015) is driving substantial policy interest in achieving greater teacher workforce diversity. In particular, policy conversations have focused how to reduce turnover among teachers color, given evidence that teachers of color leave their schools at higher rates (Ingersoll & May, 2011).

The research base in understanding the factors that lead to higher turnover rates among teachers of color is large (see Achinstein et al., 2010, for a review) but dated. Few studies have investigated the racial gap in teacher turnover in the era of high-stakes teacher evaluation. Examining this gap post-implementation of multiple measures-based teacher evaluation systems is important if we suspect that differences in how white teachers and teachers of color are evaluated may exacerbate differences in turnover.

To investigate racial differences in teacher performance ratings and turnover, we access longitudinal administrative data from Tennessee spanning 2006–07 to 2014–15, which includes the first four years of the implementation of the state’s educator evaluation system. In addition to demographic and job placement information, we access teachers’ summative evaluation ratings and individual component scores, including classroom observations and value-added. We focus on black and white teachers, who make up 99% of the Tennessee teacher workforce. Our analysis documents black-white gaps in turnover rates and in each of the evaluation metrics, then estimates models of turnover and evaluation ratings as functions of teacher race, a variety of control variables, and school and year fixed effects. Finally, we test whether differences in evaluation scores mediate the racial turnover gap.

We find a large descriptive gap in turnover between black and white teachers in Tennessee. Specifically, 23% of black teachers leave their positions each year, compared to only 16% of white teachers. This difference is driven by higher rates of transfers among black teachers; black and white teachers exit the education system at equal rates.

Next, we investigate the extent to which these descriptive gaps in turnover are explained by differences in teacher qualifications (e.g., experience, education level), school context, and evaluation ratings. We find that qualification differences explain little, but that there are stark differences in the average characteristics of the schools in which black and white teachers work. For instance, black teachers disproportionately work in urban, high-poverty schools with large percentages of students of color. Student demographics strongly predict teacher turnover, but black and white teachers are differentially responsive to school context.

Even accounting for differences in qualifications and school context, black teachers receive lower average ratings than white teachers on all components of the evaluation system. Even after controlling for value-added, black male teachers receive classroom observation scores more than half a standard deviation lower than white female teachers. Moreover, these differences in average classroom observation scores explain roughly one-third of the residual black–white turnover gap, though differences in other evaluation measures show no evidence of mediation.