Assortative mating trends in the era of online dating: Partner choice in Sweden 1990 to 2017
Author(s): Jesper Lindmarker
Thursday 15 | 10:00-10:20
Room: TP53
Session: Social stratification in a digital world
The interplay between geography and assortative mating plays a crucial role in shaping who meet and how couples meet, and in recent decades this dynamic has undergone significant transformation, to some degree driven by the advent of online dating. First, couples are now less likely to meet through social networks or through shared social foci (Rosenfeld, Thomas, and Hausen 2019). Second, the prominence of online dating facilitate more diverse interactions, as the influence of homophilous networks is diminished in the digital dating arena (Thomas 2020). Third, with dating apps people are less restricted by distance in who they can meet, even if online dating markets tend to be defined by geographical areas (Bruch and Newman 2019). A question that remain unanswered is how the new dating environment influences assortative mating patterns.
This paper presents a study of the trends in assortative mating in Sweden between 1990 and 2017. First, as an indirect test of how online dating has changed how we meet, I test changes in the importance of having attended the same school or workplaces for the likelihood of partner choice. Second, I test the hypothesis that geographical proximity (propinquity) has decreased in relative importance. Third, I examine the changes in ethnic, educational and age homogamy over time. Importantly, the modelling approach is able to test these trends while accounting for changes in relative group sizes, composition and changes in the patterns of segregation.
We use micro-level, conditional logistic regression models to estimate the probability that the observed unions were formed relative to counterfactual unions drawn from the available pool of singles (Haandrikman and van Wissen 2012; Gullickson 2021). This flexible approach allows us to estimate parameters associated with various couple characteristics while also accounting for the marginal distributions of ethnic groups and sorting along other socioeconomic and demographic dimensions.
We perform this analysis using data derived from the Swedish population registers, maintained and provided to us by Sweden’s Statistika Centralbyrån (SCB). These extensive longitudinal data allow us to identify people’s ethnic backgrounds, socio-economic status, the timing of their entry into cohabitation or marriage, and, the residential locations, workplaces and schools across the 1life course, coded down to a 100 meter spatial resolution. Beyond the level of detail in our data, Sweden is also a compelling place to study assortative mating due to both its reputation as a ‘family forerunner’ (Ohlsson-Wijk, Turunen, and Andersson 2020) and its rapidly growing ethnic diversity during our period of study, 1990-2017.