Achievement gaps between white students and their black and Hispanic peers remain a stubborn problem in American education.  Although they narrowed considerably in the 1970s and 1980s, achievement gaps remain large—roughly 0.65 (white-Hispanic) and 0.80 (white-black) standard deviations, equal to 3-4 years of learning in middle or high school.  

In this paper we pursue three goals.  First, we examine trends in white-black and white-Hispanic achievement gaps over the last two decades.  Using data from 4th and 8th grade State NAEP, as well as state accountability assessments given in 2nd through 8th grade, we fit a series of fixed and random effects models and find that the racial gaps have been closing slowly over the past two decades, at a range of 0.006 to 0.009 standard deviations per year.  Should these rates persist, it would take 70-100 years for white-black and white-Hispanic gaps to be erased.

Second, we examine cross-state variation in the magnitude of achievement gaps, as well as in rates of gap closure.  We find an inverse relationship between the two: states with the largest gaps experience the fastest rates of narrowing—at least 0.050 standard deviations per decade faster than rates of gap change in other states.  We test proximal explanations for these findings, including racial differences in poverty and state social and educational policies aimed at ameliorating inequality.

Third, we investigate the implications of No Child Left Behind (NCLB) for racial achievement gaps.  To begin, we estimate NCLB “dose-response” models, assessing the relationship between gaps and years of exposure to NCLB across cohort-by-grade observations.  Next, building on the work of Dee and Jacob (2011), we assess the extent to which gap trends changed with the introduction of NCLB by comparing states that already had comprehensive accountability systems to states that developed these systems as a result of the policy.  Finally, we investigate whether states in which NCLB exerted more accountability pressure showed larger changes in achievement gaps than states in which NCLB exerted less pressure.  Here we hypothesize that states with no consequential accountability policies prior to NCLB, no subgroup-specific accountability policies prior to NCLB, and more minority students in schools meeting minimum reporting thresholds are likely to experience the sharpest changes in their gap trends following the onset of NCLB.  

For these analyses, we fit a set of fixed and random effects models alongside comparative interrupted time series models.  We find mixed evidence for an effect of NCLB on white-black and white-Hispanic gaps.  The dose-response models suggest NCLB has had marginal effects on achievement gaps (narrowing them by roughly 0.010 standard deviations per year).  The interrupted time-series models, however, yield less consistent results.  States that introduced general and subgroup-specific consequential accountability only after NCLB implementation did not, in general, experience any more change in their gap trends than states that had consequential accountability systems prior to NCLB.  As a result, we rule out any substantial impact of NCLB on white-black and white-Hispanic achievement gaps.