Beyond additive genetic effects: Explaining family resemblance in school performance across millions of pairs of Norwegian relatives

Abstract

We investigate the hypothesis that family resemblance on school performance can be fully explained by additive genetic effects and assortative mating. Our sample consists of all schoolchildren who took Norwegian national standardized tests between 2007 and 2019 (N = 936,708). These tests measure aptitude in math and reading comprehension, and are taken the years children turn 10, 13, and 14 y old. We identify millions of pairs of relatives within our sample (82 different kinds, in total), including not only conventional biological relatives such as siblings and cousins, but also relatives-in-law, relatives through adoption, twins, and relatives connected through twins. When fitting models which assume that family resemblance arises solely from additive genetic effects and assortative mating, we find that they describe much of our data well, but that they systematically underestimate the similarity of close relatives (particularly monozygotic twins), maternal relatives, relatives-in-law, and relatives through adoption. We discuss potential explanations for these deviations, including shared-environmental effects, nonadditive genetic effects, and gene–environment interplay.