About this srcd poster session

---

Panel information
Panel 7. Diversity, Equity & Social Justice

Abstract

---

There is growing concern that the socioeconomic achievement gap may continue to widen over time, particularly in math, as economically disadvantaged students face cumulative educational disadvantages that are difficult to overcome (Bai et al., 2021). Previous research has suggested that socioeconomic status (SES) gaps in academic achievement tend to persist despite targeted interventions aimed at reducing educational inequality (Hanushek et al., 2022). However, most national-level studies have examined trends in SES achievement gaps without disaggregating the data by grade level or accounting for state-level variations. The purpose of this study is to investigate the evolution of the math achievement gap between economically disadvantaged (ECD) and non-disadvantaged students across grade levels using nationally representative state-level data.

Methods
Data were drawn from the Stanford Education Data Archive (SEDA; Reardon et al., 2024), version 5.0, which provides state-level academic performance in math for grades 3 through 8 across all 50 U.S. states from the 2008-09 through 2018-19 school years. The independent variable was grade level, ranging from 3 to 8. A three-level hierarchical linear model (HLM) was employed to account for the nested structure of the data. Observations (Level 1) were nested within states (Level 2), and further nested within years (Level 3). This approach allowed us to account for both state-level and year-level variability in the math ECD gap. The model included random intercepts for both state and year to account for unobserved heterogeneity across these levels. Fixed effects were used to estimate the impact of grade level on the math ECD gap.

Results
HLM results indicated a significant positive association between grade level and the math ECD gap, suggesting that the gap widens as students progress through grade levels. Specifically, for each one-unit increase in grade level, the math ECD gap increased by 0.134 points (p < .001). This suggests that the math disparity between these student groups becomes more pronounced as they advance through school. The model also revealed variability in the math ECD gap across both states and years. The random intercept for states (σ2 = 0.084) indicated significant state-level differences in the math ECD gap, while the random intercept for years (σ2 = 0.045) suggested that the math ECD gap fluctuated slightly over time. The likelihood ratio test confirmed that the inclusion of random intercepts for both state and year significantly improved model fit (χ2(2) = 4878.19, p < .001), highlighting the importance of accounting for these sources of variation.

Discussion
This study is one of the first to examine socioeconomic disparities in math achievement gaps across the entire U.S. using state-level longitudinal data spanning more than a decade. Findings of the widening math ECD gap underscore the need for early interventions to prevent the gap from increasing as students progress. The significant state-level variability in the math ECD gap suggests that local policies and resources play a key role in shaping these disparities. To address these disparities, early interventions and tailored policies at both the state and federal levels are essential.

Author information

Author	Role
Shanyilin Jin, UC Berkeley	Presenting author
Hua Luo, UC Berkeley School of Education	Non-presenting author

⇦ Back to session