State Comparison of High-Achieving Students in High-Poverty Schools

• In Event: Investigating Academic Resilience Using Data From the 2017 NAEP

Abstract

Research shows that achievement scores vary widely across states, even when controlling for student demographic factors (Chingos, 2015). Given this variation, and the role that states can play in implementing policies to help low-income students succeed (Plucker et al., 2018), this presentation focuses on resilient students at the state level. It uses school-level National School Lunch Program (NSLP) eligibility as a proxy for socioeconomic status (SES), considering that students in high-poverty schools, on average, do not perform as well as those in low-poverty schools (NCES, 2017). The purpose of this presentation is to identify states whose achievement gaps between students in low- and high-poverty schools are significantly lower than the average national achievement gap and to examine the possible sources of the variation in those states. The analytic approach is as follows:

• NAEP grade 8 2017 mathematics data was separated into two sample groups: (1) students who attended schools where 0-50% of the students were NSLP-eligible (“low-poverty schools”) and (2) students who attended schools where 51-100% of the students were NSLP-eligible (“high-poverty schools”).
o Approximately 1.4 million students in the nation attend high-poverty schools, and the percentage of these students varies across states (Table 1).
• The mean scores for students in low- and high-poverty schools were obtained for each state and the District of Columbia, and for the nation.
• To explore patterns and variations across states, a difference in difference (DID) method was utilized to compare each jurisdiction’s achievement gap of students in low- versus high-poverty schools to the national achievement gap.

As shown in Table 2, states varied in their average scores for the two student groups. States that are doing well for their students in high-poverty schools were defined as those states whose achievement gaps were one quarter of a standard deviation unit smaller than nation’s. The top eight states listed in Table 2 met these criteria. It is worth noting that the states with smaller achievement gaps were not necessarily the states with the highest overall mean scores or with the lowest percentage of students in high-poverty schools. For these states, further analysis includes: (1) analyzing policy implementations and (2) using an imputed dataset to conduct multilevel regressions to characterize how much of the variation in student scores can be attributed to student- versus school-level factors. Level one variables will include student demographics, (e.g., individual-level NSLP, parent education), and level two variables will include school-level factors (e.g., school-level NSLP, course offerings).

This study has implications for closing SES achievement gaps, as states that are doing well for their students in high-poverty schools can disseminate their policies and best practices to others. It also has important implications in using state mean NAEP scores as benchmarks of success, as many of the states with top NAEP scores still have persistent SES gaps.

Authors

• Stephanie Straus, Georgetown University

• Alison Filbey, American Institutes for Research