Using Polygenic Scores for Educational Attainment to Probe Genetic Effects on Schooling through Adolescence

• In Event: 1-180 - Genetic Effects on Health and Education Across the Life Course

Integrative Statement

Background. Genome-wide association studies (GWAS) analyze millions of genetic variants in hundreds of thousands of individuals to identify genetic associations with specific outcomes of interest (measured as polygenic scores; Chabris, Lee, Cesarini, Benjamin, & Laibson, 2015; Belsky et al., 2016). Several recent studies have both created and reported polygenic scores that have been linked to educational attainment, ranging from early academic skill acquisition through adult economic and social outcomes (Belsky et al., 2016; Rietveld et al., 2013; Lee et al., 2018; Trejo et al., 2018). Understanding how genetic variation influences life outcomes is important to researchers parsing the variability of social science phenomena, and may one day be used to more precisely control for the causal effects of the environment and to invest in maximally effective interventions (Harden, 2018). But, the mechanism(s) by which this link emerges require further exploration, particularly in diverse U.S. samples. To that end, the present study extended this field of research using the Fragile Families and Child Wellbeing Study, which obtained data from an economically and racially diverse group of children and their families followed from birth through age 15. Ours is among the first studies that examine polygenic scores for educational attainment among such a sample and is, therefore, important for validating and refining this polygenic score for future research.

Method. We analyzed educational, cognitive, and behavioral data when children were aged 5, 9, and 15. Sources included survey reports by children, parents, and teachers, and direct assessment of both cognitive and academic skill. We also analyzed school records data for a subset of these adolescents. Continuous outcomes were assessed with OLS regression to estimate standardized beta coefficients, and dichotomous outcomes with logistic regression to estimate odds ratios. The polygenic score was included as a predictor in analyses after being standardized to have a mean of 0 and a standard deviation of 1 within each ancestry group. All analyses were conducted separately by race/ethnicity (White n = 455, Black n = 1264, Hispanic n = 664) and analyses controlled for ancestry-specific principal components computed from GWAS data.

Results. Analyses are ongoing, but preliminary results reveal two important findings. First, our results replicate those of Belsky et al. (2016). Polygenic scores for White children are significantly and positively associated with teacher-reported grades at age 5 and 9, and with child-reported grades at age 15. Standardized betas range from .18 to .26 (see Tables 1-2). Second, and most important, results differ by ancestry. Though the polygenic scores significantly predict academic and cognitive outcomes for White students, they are not significantly related to these same outcomes for Black or Hispanic students (e.g., see Table 1). Moreover, at age 15 the polygenic score was significantly linked to disciplinary behavior and grade retention only among Black or Hispanic students (Table 2). Overall, our results reveal a need for polygenic scores specific to non-White individuals, as the score most commonly utilized in the field does not appear to operate similarly across ancestry groups.

Authors

• Adrienne D Woods, Penn State University
  Presenting Author

• Arianna Morgan Gard, University of Michigan, Ann Arbor
  Non-Presenting Author

• Daniel Notterman, Princeton University
  Non-Presenting Author

• Colter Mitchell, University of Michigan
  Non-Presenting Author