School Socioeconomic Composition and Young Children's Academic Development in the United States

• In Event: 72.031 - International Perspectives on Socioeconomic Segregation Between and Within Schools

Abstract

Residential income segregation in the U.S. has grown considerably over the past several decades (Reardon & Bischoff, 2011), and there is evidence suggesting increases in income-based segregation between schools as well (Owens, 2014), driven largely by stratification across school districts (Corcoran & Evans, 2010). In addition to the social and political ramifications of school socio-economic segregation, researchers have noted that school socio-economic composition itself likely influences student academic development (Van Ewijk & Sleegers, 2010). Explanations for this potential link generally fall under two broad categories: 1) disparities in institutional resources, including adult human capital and material resources, and; 2) differences in peer norms, behaviors, and expectations across low- and high-SES schools. Using a large, nationally representative dataset from the U.S., we explore the extent to which school-level SES influences young children’s academic development in literacy and mathematics, above and beyond students’ own socio-demographic backgrounds. We also examine whether socio-economic context influences the degree to which schools are internally equitable.

Our study employs data from the Early Childhood Longitudinal Study, organized by the U.S. Department of Education’s National Center for Education Statistics (Tourangeau, et al., 2015). Our sample includes 15,741 children who were attending kindergarten in one of 866 schools during the 2010-2011 school year. We then follow these children as they progress through first and second grade. All models employ hierarchical linear modeling (HLM 7.0; Raudenbush, et al., 2015), with children nested in schools. Many authors have used three-level, linear piece-wise growth models with the ECLS-K data (see Downey, et al., 2004; Ready, 2010). However, these models are unable to account for associations between initial status and growth (see Ready, 2013), which are notably negative with the ECLS-K assessments across all test scalings. In response, we constructed models that employed “locally standardized” gain scores that compared cognitive development among children within similar initial achievement strata or “bins” (see Betebenner, 2008; LoGerfo, et al., 2005; Ready, 2013). These measures use children’s assessment scores in the fall of each grade as the baseline achievement indicator, with children’s spring scores as the post-test outcome. One benefit of this approach is that it removes the potential “pollution” associated with summer learning loss.

At the child-level, socio-economic status is weakly, but positively associated with both literacy and mathematics learning across all grades. The focus of the paper is the finding that even after adjusting for a host of child-level characteristics, school-level SES is positively associated with literacy development across all three grades, and with mathematic s development in kindergarten. Indeed, with kindergarten and first-grade literacy development, the school-level SES effects are actually somewhat larger than the child-level SES estimates. We find similar estimates associated with school racial/ethnic composition. As indicated by “slopes-as-outcomes” models, although school-average SES influences mean achievement, it does not influence the equitable distribution of that achievement (e.g., the child-level within-school SES estimate does not vary as a function of school socio-economic composition).

Authors

• Hanna Dumont, German Institute for International Educational Research

• Douglas Ready, Teachers College, Columbia University