Separate but equal?

Abstract

I estimated a linear fixed effects model using panel data collected from the Georgia Department of Education to analyze the impact changes in student body demographics have on African American student performance. The dataset contained detailed information on every public middle school in the state of Georgia over the five school years from fall 2003 to spring 2008. I found that a ten percent increase in the percentage of the student body that is African American corresponds to an approximately three percent rise in the percentage of African American students failing the math section of the Criterion-Referenced Competency Test (CRCT). Increasing the percentage of African Americans at a school had a negligible impact on white student outcomes. These effects persisted after adding a variety of student and school control variables and after allowing past student performance to explain current outcomes by including a lagged dependent variable in the regression. My results, consistent with other findings in the area, suggest that recent trends toward resegregation could cause the black-white achievement gap to expand.