ABSTRACT
The effect of school socioeconomic composition, or school SES, on student achievement has long been an area of interest for policymakers and scholars alike. The Coleman Report itself detected a correlation between individual students’ performance on standardized tests and the socioeconomic composition of their peers at the grade and school levels. With direct implications for school integration policy, a literature rapidly developed investigating whether this relationship is causal in nature.
Correlation analyses of the relationship face a fundamental issue at odds with drawing causal inferences: would a student with primarily low-SES classmates truly perform better in a higher-SES school, or does the correlation merely capture the tendency of higher-achieving students to attend higher-SES schools? Methodological, computational, and data availability advancements have yielded improvements in the credibility of approaches to this issue over time.
The authors’ 2018 study capitalizes on these advancements. Using a multiway fixed effects approach with administrative data from three states in the American South, we estimate a near-zero effect of school SES on individual student achievement. Quasi-experimental studies in the years since have supported the conclusion that large peer composition effects found in cross-sectional studies are largely artifacts of selection.
