Achievement Gaps Between Poor and Non-Poor Students in U.S. School Districts

• In Event: Socioeconomic Status and Standardized Tests: Achievement Gaps, Alternate Assessments, and Opting Out

Abstract

There is substantial evidence that students’ family socioeconomic status and neighborhood socioeconomic conditions impact their educational opportunities and outcomes. At the individual level, students from poorer families tend to perform worse on tests compared with peers from wealthier families leading to test score gaps within schools and districts (Reardon, 2011; Sirin, 2005). At the aggregate, students in poorer communities tend to perform worse than students in wealthier communities (Reardon, 2016). In our work, we analyze how test score gaps between poor and non-poor students within U.S. school districts are a function of the local socioeconomic context. This is the first study to systematically look at disparities in test scores between poor and non-poor students within school districts across the U.S.; and, will inform our understanding of the factors that shape differences in educational opportunities available to them.

We use test score data from the EDFacts Data Initiative provided to our team by the National Center for Education Statistics to estimate achievement gaps between poor and non-poor students. States select their own definition of economic disadvantage; however, in a majority of states it is defined as whether or not a student receives free or reduced lunch. From the EDFacts data, we estimate standardized mean test scores for both economically disadvantaged (poor) and non-economically disadvantaged (non-poor) students in approximately 9,000 school districts across the U.S.

We find that there are almost no U.S. school districts where poor students are outperforming non-poor students, but that there is substantial variation in both poor and non-poor students’ achievement across school districts. Figure 3 shows the Empirical Bayes test score mean estimates (pooled across grades, years and subjects) for non-poor students plotted against those for poor students in each district. The observations are weighted by the total number of poor students in the district, which shows that although there are a small number of school districts where poor students outperform non-poor students they serve a very small fraction of the poor student population within the U.S.

In order to analyze the relationship between the difference in the means with district-level characteristics, we use precision-weighted hierarchical linear models with state fixed effects. The state fixed effects reduce any bias that results from the different definitions of economic disadvantage used across states. Our preliminary results show that communities with higher median income, more income segregation, income inequality, and school funding have larger test score gaps between non-poor and poor students (Table 4). The result that there are larger achievement gaps in more affluent and higher-spending school districts is rather counterintuitive – we might expect that higher average levels of local and school resources would lead to more equal opportunities among students despite their differences in family context. In contrast, larger gaps in school districts with more economic segregation and inequality are not unexpected – poor students and affluent students likely have more different experiences within these types of school districts. Notably, however, the socioeconomic factors explain only 27% of the within-state variation in test score gaps, leaving much unexplained.

Authors

• Sean F. Reardon, Stanford University

• Erin Michelle Fahle, Stanford University

• Heewon Jang

• Elise Dizon-Ross, Stanford University