Determinants of Between-School Variation in Student Achievement: Results from US Population Data

This article uses school-level aggregate achievement data to estimate the intraclass correlation coefficient (ICC)—proportion of variation in test scores that is between schools within states—and how this is related to school characteristics. This study contributes to the literature by providing an up-to-date, comprehensive set of between-school ICC estimates based on population-level data and identifying relevant school context characteristics that can explain differences in ICC estimates. We estimate 6,021 state-grade-year-subject ICCs for grades 3–8 mathematics and reading/language arts (RLA) using state assessment data from 2009 to 2019 across all 50 states and the District of Columbia. The average math ICC is 0.194 and the average RLA ICC is 0.168. Average ICC estimates vary widely across states. The majority of between-state differences in ICC estimates (70% in math and 78% in RLA) can be explained by a small number of variables representing school structures and the level of between-school racial and economic segregation. We also find average ICC estimates increased over time, with larger increases in elementary grades and most of the increases occurring in 2015 or later.