Poster #40 - Assessing Heterogeneity with Moderation: Students' Bilingual Status and Science Achievement in Elementary School

• In Event: Poster Session 09

Abstract

Emergent bilinguals (EBs) are a growing and highly diverse student population. EBs display lower science achievement in U.S. elementary, middle, and high schools (NAEP, 2019). These science achievement gaps begin by kindergarten and persist across elementary school (Morgan et al., 2016). The purpose of our study is to examine 1) the heterogeneity among EBs in their science achievement during elementary grades; and 2) potential moderators among such heterogeneity.

We analyzed the nationally representative Early Childhood Longitudinal Study, Kindergarten Class of 2010-11 (ECLS-K:2011). Our main outcome was individually assessed science achievement. Our main predictor is children’s bilingual status, which was determined by combining parents reported primary home language and directly assessed English proficiency in kindergarten. This has resulted in five categories: 1) English monolinguals; 2) Primary non-English monolinguals; 3) English dominant bilinguals; 4) Bilinguals with English proficiency; 5) Bilinguals with limited English proficiency.

Few studies have controlled sufficient socio-demographic factors in their analyses (Castro, 2014). We adopted an Antecedent-Opportunity-Propensity framework (Byrnes, 2020): Antecedent variables included students’ demographic characteristics. Opportunity variables included aspects of learning contexts at school and home. Propensity variables included students’ academic, executive and behavioral functions. In each grade, three models have been fitted into ordinary least squares regression. Model 1 included the main predictor. Model 2 added covariates. Model 3 added interactions between bilingual status and executive functions and reading achievement.

Table 1 profiles weighted students' science achievement and covariates across five bilingual groups. The heterogeneity in socio-demographic characteristics of bilingual students is well documented (Kim et al., 2018) and illustrated in Table 1. For example, bilingual students consistently score lower than their English monolingual peers in science achievement; however, students with limited English proficiency always score the lowest. In addition, this group showed the lowest literacy level, came from a low-social-status family, as well as attended schools with low income, and a high degree of racial/linguistic minorities. Thus, it is imperative to distinguish this heterogeneous student body and control these socio-demographic characteristics simultaneously.

Table 2 presented regression results. We observed negative associations between bilingualism and science achievement across grades in Model 1. Once covariates were added in Model 2, the results changed. In the fourth and fifth grades, for instance, non-English monolinguals and bilinguals with English proficiency score significantly higher than their English monolingual peers. The Model 3 further demonstrated that working memory and reading achievement in kindergarten moderated the relationship between bilingual status and science achievement across grades. Bilinguals with limited English proficiency would significantly increase their fifth-grade science achievement by 2.61 points for every 1-unit increase in working memory during kindergarten.

In summary, we found heterogeneity within bilingual status: children who grow up in family without any primary English explosion or with sufficient English proficiency would outperform their English monolingual peers in science achievement once we controlled social-demographic characteristics. Meanwhile, even though all students benefit from higher kindergarten working memory and reading achievement, it additionally benefits students who are primary non-English Monolinguals, and Bilinguals with limited English proficiency.

Authors

• Eric Hengyu Hu, Pennsylvania State University, University Park
  Presenting Author

• Paul L. Morgan, Pennsylvania State University, University Park