(9/10) Several reasons besides direct assortment can explain partner similarities. In this paper, we cannot determine which processes are most important. However, we can distinguish between these in future research.

(8/10) Could health similarities be by-products of partner choice based on education? To some extent—partner correlations in health were reduced by 30–40% after accounting for educational attainment in adulthood or school grades at age 16.

(7/10) Do mental health and education determine partner choice? Many studies assume so-called direct assortment. If direct assortment fully explains partner choice, it would be easy to predict correlations between siblings-in-law. But they are much more similar than expected under direct assortment.

(5/10) Below, we see near universal cross-trait assortment. In the prospective data, partner correlations are as strong between different mental disorders as for the same mental disorders. Later in life, partners were more likely to have the same mental disorder (see paper).

(4/10) We see strong similarity in education, as expected, and also for school grades at 16. In dark blue, we see that partners' mental health was correlated from the start. But cross-sectional studies likely overestimate assortment. In red, we see that mental health became more similar over time.

But here's the twist: this link vanishes when we account for children's grade point average (GPA) at age 16. GPA appears to mediate the effects of family background.

🚨 New study alert! 🚨
We find that socioeconomic background—parents' income and education—strongly predicts early adult mortality.