Measure for Measure: Quantifying Racial Inequality in an Age of Educational Accountability

Jennifer Jennings, New York University
Heeju Sohn, University of Pennsylvania

This paper argues that the two goals of educational accountability systems – to incentivize and to accurately measure – are, to some extent, at odds with each other and pose challenges for measuring between-group inequality in outcomes. We develop this argument by investigating whether the implementation of the No Child Left Behind Act affected student achievement at different parts of the achievement distribution and asks whether the inferences we draw depend on whether we analyze high or low-stakes tests. Analyzing a longitudinal student-level data from Houston Independent School District over the period 2001 through 2007, we show that conclusions about the effects of schools on racial inequality and the distributional effects of accountability systems depend substantially on the choice of outcome. For example, while we find negative effects of the implementation of NCLB on the bottom of the distribution on high-stakes math tests, we find positive effects for these students on the low-stakes test.