Many states, school districts, and information-sharing platforms report measures of school performance. Often called “school ratings,” they are widely consulted by parents and educators alike. Families looking for a new home are likely to see ratings posted alongside listings, while low-rated schools may be closed or placed under state supervision.
A school’s rating is often strongly correlated with the racial make-up of its student body. Higher-rated schools tend to have a greater percentage of white students. Blueprint Labs economists Joshua Angrist, Peter Hull, Parag Pathak, and Christopher Walters analyzed this correlation in commonly used ratings of schools in New York City (NYC) and Denver, formulating a new measure of school performance. In these settings, their new “race-balanced progress” rating is uncorrelated with race but just as predictive of school quality as conventional progress ratings.
The researchers begin by distinguishing a school’s quality — defined as its causal impact on student achievement — from the family background and past experience of its student body. High quality schools excel at boosting achievement for students of a given background and preparation level. Ratings that are influenced primarily by student background and preparation rather than by school quality are said to be compromised by selection bias.
This study’s findings suggest that, for middle schools in NYC and Denver, the racial make-up of a school’s student body is largely unrelated to school quality. Selection bias drives the correlation between widely used ratings and student racial composition: many schools rate higher simply because they serve students who tend to have higher test scores regardless of school quality (e.g., higher-income students). Popular school ratings based on achievement levels are particularly misleading measures of quality and highly correlated with race. At the same time, ratings that look at achievement growth or progress across grades better reflect school quality and are less correlated with race. Still, even progress ratings have room for improvement.
This study offers a simple method for adjusting academic performance ratings that removes the correlation between the rating and race. The researchers find that race-balanced progress ratings are at least if not more predictive of school quality than are conventional progress ratings.
