| Susan McElroy
Carnegie Mellon University
Why do black kids fail: The effects of poverty on racial gaps in test scores
FINAL REPORT:
This research addresses the following question: Does poverty explain the racial gap in test scores? Our data set is the National Education Longitudinal Study of 1988 (NELS:88), in which students were first interviewed in the eighth grade in 1988. We estimate a series of two-stage regressions using grades as the instrumental variable in the first stage and performance on math, reading, and science Item Response Theory (IRT) tests in the eight, tenth, and twelfth grades. We then perform a series of Blinder-Oaxaca decompositions by race and poverty status to examine the influence of race and poverty on differences in test scores. The Blinder-Oaxaca decomposition method calculates the difference in test scores between two groups (such as blacks and whites) and estimates what the dependent variable (test scores) for one group would have been had they experienced the same treatment effect as the other group.
We find significant differences in test scores by race, as documented by many previous researchers. For both poor and non-poor white and black students, a high percentage of the racial gap in test scores is unexplained by the decomposition. However, we find that the power of poverty as an explanatory factor varies by race. Our results demonstrate that the relationship between poverty and test scores is not as straightforward as one might expect at first glance. For example, poverty serves an explanation for differences in test scores among white students but not among black students. Similarly, among poor but high-performing students, the difference in test scores between white and black students is not significant while among low-performing students, the difference is significant. On the other hand, among non-poor but high-performing students, the difference in test scores between white and black students is not significant, while among low-performing students, the difference is generally not significant. Alternatively, the difference in test scores between poor white and black students who are low performers is significant while the difference in test scores between poor white and black students who are high performers is generally not significant.
In terms of policy implications, our research suggests that efforts to alleviate the poverty of black students as a means of raising test scores may not be as fruitful as they would be for white students. More importantly, if poverty is not the cause of the racial gap in test scores, then the identification of other causes of the racial gap in test scores remains a problem to be addressed by public policy.
Back to Funded Research Grants Page |