Rolf Degen summarizing... Among children from highly educated parents, genetic factors had a far greater effect on educational attainment than among those whose parents had low levels of education, accounting for 75% of the individual differences

How Social and Genetic Influences Contribute to Differences in Educational Success within the Family. Tina Baier. Doctoral thesis, Faculty of Sociology at Bielefeld University, Germany. July 9, 2019. https://pub.uni-bielefeld.de/download/2940317/2940667/Dissertation_Baier.pdf

4.4 Conclusion and Discussion
This study extended previous research on gene–environment interactions for education in two crucial ways. First, we acknowledged that not only the proximate family but also the broader institutional environment can shape genetic effects on education. Second, we extended previous research that focused originally on IQ to indicators of educational success, namely educational achievement measured in school grades and educational attainment measured in years of education. Specifically, we addressed the following research questions: Do genetic effects on educational success vary across countries, and are there differences in the social stratification of genetic effects on educational success among these countries?

We selected three advanced industrialized societies for our study: Germany, Sweden, and the United States. These countries largely differ in the setup of their educational systems and represent prototypically three different types of welfare regimes, which are often used in internationally comparative social inequality research. We hypothesized that genetic influences on educational success are overall weaker in Germany and the United States than in Sweden. Furthermore, we expected that the association between parents’ socioeconomic standing and genetic effects on educational success is stronger in Germany and in the United States than in Sweden. For Germany, our hypothesis was rooted in the early tracking system and for the United States in the less extensive welfare regime.

Our study yielded three important findings: First, we found that genetic effects on years of education are smaller than genetic effects on school grades –independent of country. Hence, genes are more important for educational achievement than for educational attainment. In addition, shared environment environmental influences on educational attainment were stronger in Germany and the United States. This supports the notion of socially stratified schooling decisions that operate over and above educational achievement (Boudon 1974; Breen and Goldthorpe 1997; Erikson and Jonsson 1996). However, we did not find effects of the shared environmental influences on educational attainment in Sweden, which diverts from previous findings based on an international meta-analysis (Branigan, Mccallum, and Freese 2013).

There are three reasons that could account for conflicting results. First, our results are based on more recent birth cohorts (i.e., we studied birth cohorts for 1975–1982, while meta-analysis examined birth cohorts for 1926–1958), and previous research shows that genetic influences on education have increased among birth cohorts born in the second half of the twentieth century (Branigan, Mccallum, and Freese 2013; Heath et al. 1985). Second, the samples used in the meta-analyses were not all population based, including the sample of Sweden where the Swedish Twin Registry was used. Third, the meta-analysis did not account for assortative mating. Without such an adjustment, genetic influences tend to be underestimated, while shared environmental influences are overestimated (Freese and Jao 2017). That shared environmental influences were absent for educational attainment in Sweden indicates that educational choices are more closely related to educational achievement, which could be explained with the less selective comprehensive schooling system.

Second, we identified cross-country differences in genetic effects on educational success. Genetic effects on educational success were least pronounced in Germany, and most pronounced in Sweden. Our hypothesis on cross-country differences was therefore supported for Germany, since genetic effects were comparatively small for both indicators of educational success. For the United States, our hypothesis was only partly supported, since genetic effects on educational attainment were comparatively small, while genetic effects on educational achievement were at least as large as in Sweden.  Together, these findings supported our expectation that more egalitarian educational systems have a positive effect on the development of genetic potential for educational success and that early tracking might be an important factor for the suppression of related genetic effects. Future research should build upon our findings and focus in a more detailed manner on the impact of the tracking system. For instance the educational system in the Nordic countries changed from a tracked to a comprehensive schooling system (see for an overview on the educational reforms in Denmark, Finland, Norway, and Sweden (Gustafsson 2018)). If tracking lowers genetic effects on education, genetic effects on educations should increase after comprehensive schools were introduced. Systematic cross-countries using a culturally homogenous set of countries (“most similar case design (Lijphart 1971)) increase the generalizability of the results.

Third, we found indications for a social stratification of genetic effects in line with the Scarr–Rowe hypothesis for educational success in Germany and the United States. We did not find any evidence for a gene–environment interaction in line with the Scarr–Rowe hypotheses in Sweden. If anything, this underlines the positive impact of more egalitarian educational systems on the development of genetic effects relevant to education.  However, differences between countries are too small and not robust enough to clearly support our hypothesis. Yet, the evidence for an interaction in line with the Scarr–Rowe hypothesis for Germany is weaker than previously found using a more fine-grained measure for years of education (Baier and Lang 2019). Thus, differences in the results for Germany between this and the previous study are likely to be driven by the harmonized measure of education which comes at the cost of preciseness. For the international comparison, however, it is crucial to investigate the same measure of education in each country; otherwise, results on genetic and environmental influences can be differently affected by the way educational attainment is measured and, thus, cannot be meaningfully interpreted across countries.

It is important to note that twins’ zygosity was unknown for our sample from Sweden.  We adjusted in line with previous research for the missing information based on the assumption that same-sex and opposite-sex dizygotic twin births are equally likely (Figlio et al. 2017). This is assumption is fairly reasonable. In addition, there is no reason to believe that the distribution same-sex and opposite-sex dizygotic twin births varies by parents’ social background which would have affected our results in regards to the Scarr– Rowe hypothesis. Nonetheless, future research is needed to gain the precise estimates of genetic influences on educational success. Since some twin pairs tend to be misclassified, our adjustment can lead to an underestimation of genetic differences between monozygotic and dizygotic twins. Therefore, our results represent lower bounds of genetic influence on educational success. Hence, the overall conclusions we draw from our cross-country comparison should not be affected by this adjustment. If anything, we underestimated the role of genes in Sweden.

For the United States, our sample sizes were comparatively small, and analyses for parents’ EGP class were based on broad categorizations (i.e., EGP classes I and II versus EGP III–VII, including the non-employed). However, the Add Health data are currently the only nationally representative dataset that includes twins. Since the quality of educational institutions varies considerably among federal states, the representativeness across states is crucial for our study purposes. Nonetheless, more research for the United States is needed to test in a more fine-grained way for the social stratification of genetic influences on educational success.

In sum, our study is the first to study cross-country differences in genetic effects on educational success. We found substantial differences in genetic effects on educational success among Germany, Sweden, and the United States. An important factor that causes these cross-country differences may be rooted in the stratification of educational systems, specifically in the strictness and timing of tracking.