Jesper
Lindmarker

Living with a partner is linked to living longer. But is that the relationship itself, or do healthier people just couple up more? We compared siblings, same family, different relationship paths, to separate the two. Cohabitation does appear protective, though less so than marriage, and the effect varies across the life course.

Sibling fixed-effects survival models on Swedish register data covering 5.6 million individuals (2012–2017), separating selection from causal partnership effects on mortality.

People tend to partner within their own ethnic group. The usual explanation is preference. But when we account for who actually lives nearby, 20–40% of same-group partnering is explained by geography alone. Segregation narrows who you can meet before preference even enters the picture.

Conditional logit models on Swedish population registers (1990–2017), comparing observed unions to counterfactual partners drawn from the local singles pool, with mediation analysis separating residential proximity from ancestry preference.

Most studies compare "majority" with "minority." We look at all 2,500 possible pairings across 50 ancestry groups in Sweden. Some groups partner at 294 times the expected rate; others far below. Seven distinct clusters emerge, shaped by geography, culture, and gender. The same boundary can be open for women and closed for men.

Two-sided matching simulation with log-linear models and Walktrap community detection on full-population Swedish register data (1991–2022), mapping the complete matrix of ethnic over- and underrepresentation across 50 ancestry groups.

Interethnic couples have become more common in Sweden over 30 years. That looks like progress. But most of the increase comes from growing diversity, more people of different backgrounds simply being around. When you account for that, same-group preferences have barely changed, and for some groups they got stronger.

Conditional logit models with Oaxaca-Blinder decomposition on Swedish register data (1991–2022), separating structural partner market shifts from changes in ethnic assortativity across six ancestry groups and two immigrant generations.

Highly educated people tend to partner with each other. But is that a preference for education, or a side effect of spending years at the same university and then the same workplace? Shared institutions explain 30–40% of it, and each extra year of overlap makes a union more likely. The places that shape your career also shape who you end up with.

Conditional logit discrete choice models with KHB mediation decomposition on 1.25 million first unions from Swedish registers (1990–2022), isolating the contributions of university co-enrollment, workplace co-membership, and residential proximity.