Next, we conducted linear regressions to assess the predictive power of EXT PGS in explaining variation in adolescent externalizing behavior scores in our sample. Among EAs, the externalizing polygenic score was associated with both T1 externalizing behavior score (ΔR2 = .03, p < .01) and T2 externalizing behavior score (ΔR2 = .05, p < .01). Among AAs, neither the polygenic score constructed using the EA weights nor the multi-ethnic polygenic score was associated with externalizing behavior score (ΔR2 = .001–.007, all ns). Because the multi-ethnic polygenic score did not improve the predictive ability in the AA sample, we carried forward the polygenic score constructed using the weights from results of the GWAS based on samples of European ancestry.
