Stiefel, Leanna, & Schwartz, Amy Ellen
Disentangling the Racial Test Score Gap: Probing the Evidence in a Large Urban School District
They examine the distribution of the gap in test scores across races within New York City public schools and the factors that explain these group.
Journal Name or Institutional Affiliation:
Journal of Policy Analysis and Management
Vol. 26 No. 1 pp. 7-30
There is some evidence that school characteristics matter. Race gaps are negatively correlated with
school size-implying small schools may be helpful. In addition, the trade-off between the size and
experience of the teaching staff in urban schools may carry unintended consequences for within-
school race gaps.
Controlling for the socioeconomic and school characteristics reduces the estimated Black-White gap by about a quarter of a standard deviation; analogous reductions in the Hispanic-White gap are even larger-somewhat over a third of a standard deviation.
The difference in performance between poor students (measured by free lunch eligibility) and non-poor students approaches .5 standard deviations, making the “poverty gap” almost as large as the race gap in some instances. The significantly higher incidence of poverty among Black and Hispanic students means that some of the origin of the race gap is outside the purview of schools. Nonetheless, their results indicate that a significant race gap remains, even after differences in poverty status are controlled
The difference in performance between poor students (measured by free lunch eligibility) and non-poor students approaches .5 standard deviations, making the “poverty gap” almost as large as the race gap in some instances. The significantly higher incidence of poverty among Black and Hispanic students, means that some of the origin of the race gap is outside the purview of schools. Nonetheless, their results indicate that a significant race gap remains, even after differences in poverty status are controlled
Adding school fixed effects reduces the size of the Black-White test scope gap and the Hispanic-White test scope gap by over a tenth of a standard deviation more. The Asian-White gaps are either positive or no different from zero. Thus, differences in schools explain some of the racial test score gaps. That said, the coefficients on the Black and Hispanic variables remain negative and statistically significant even after school fixed effects are included, suggesting that significant
within-school gaps remain for Black and Hispanic students.
In most schools, Black and Hispanic students in the fifth and eighth grade achieve lower test scores then their White peers.
Controlling for classroom fixed effects, once prior test score is taken into account, does little to explain racial test score gaps. In fact, the estimates of within-school racial gaps are almost identical to those for within-classroom racial gaps.
Once they controlled for prior test scores and school assignment, racial gaps diminish significantly. Nonetheless, Black and Hispanic students continue to perform worse than their White and Asian peers, even after controlling for their performance on tests in the prior year.
Journal Article Empirical Research
Academic Achievement, Achievement Gap, Racial Composition, Urban Schools
Method of Analysis:
5th and 8th graders
Unit of Analysis:
This paper utilizes a dataset provided by the New York City Department of Education (NYCDOE) that includes individual-level socioeconomic and educational information on every fifth and eighth grade student in the New York City public school system in the 2000–01 school year, as well as an identifier for the schools and class-rooms in which they are registered, and lagged test score data.
The sample includes 70,638 fifth graders and 55,921eighth graders.