Association for Public Policy Analysis & Management

This paper estimates gender achievement gaps within approximately 9,500 school districts across the nation to explore the extent to which these gaps in academic performance vary among local contexts. We describe the spatial patterning and variation of these gaps and then, motivated by the work of Autor et al. (2015) and Chetty et al. (2016), examine the extent to which they are associated with a set of school district socioeconomic characteristics.

The test score data used in this study comes from the EdFacts state accountability tracking system, which was provided via restricted license to our team by the National Center for Education Statistics. The EdFacts data are provided as counts of students scoring at the state proficiency levels (e.g. “Below Basic,” “Basic,” etc.) disaggregated by district, gender, grade, subject, and year. The data are available for every state in the 2008-09 through 2012-13 school years, for the third through eighth grades in mathematics and English Language Arts (ELA). In total, approximately 250 million student test scores are reflected in this data set.

We estimate the district-level achievement gaps from this coarsened data, using the V-statistic (Ho, 2009; Ho & Haertel, 2006; Ho & Reardon, 2012), and adopt a two-dimensional framework for characterizing patterns in reading and math achievement. The first dimension characterizes the gaps’ alignment with traditional gender stereotypes— that boys outperform girls in math and girls outperform boys in reading. The second dimension characterizes the extent to which both gaps, instead, tend to favor one gender over the other. Among the roughly 9,500 districts in our sample, the average gender achievement gap in math is near 0, although there is considerable variation among districts. The average gap in ELA is approximately 0.25 standard deviations in favor of girls. Again, there is significant variation in the ELA gaps; however, nearly all school districts have an ELA gap that favors females.

Given the evidence of variation in the size (and direction, in the case of math) of gender achievement gaps among school districts, we conduct a descriptive analysis to investigate whether gender gaps are associated with socioeconomic characteristics of the population, as is shown in prior work. We construct measures of school district socioeconomic status and gender disparities in socioeconomic status using data from Common Core of Data and School District Demographics System. From our analyses, we find that gender gaps favor male students more in school districts in which there are larger gender disparities in socioeconomic status among adults. We also find that gender gaps follow a more “stereotypical” pattern—with male students outperforming female students in math, and females substantially outperforming males in ELA—in more socioeconomically advantaged school districts.