Love at First Swipe: Online Dating, Assortative-Mating, and Family Outcomes in Europe Using Machine-Learning-Based Causal Inference

• In Event: 50261 - Family Section Roundtables
  In Refereed Roundtable (60 minute): Table 12: Family Making

Abstract

This study investigates the impact of online dating on marital and fertility outcomes and partnership heterogamy (education, age, and nativity), as well as the moderating role of heterogamy across nine European countries using data from the second round of the Generations and Gender Survey (GGS-II).
Using Augmented Inverse Probability Weighting (AIPW) to estimate the average treatment effect, results indicate that meeting a partner online is consistently associated with a 10-30% lower likelihood of marriage and having biological children across sampled countries. However, online dating does not significantly impact relationship satisfaction or the likelihood of considering a breakup. AIPW and multinomial models further reveal that meeting the partner through online dating increases both education hypergyny and hypogyny, and that the effect on nativity heterogamy is mostly driven by hypergyny rather than hypogyny. In contrast, its effect on age heterogamy remains limited.
To explore treatment effect heterogeneity, Causal Forests are applied, estimating individual treatment effects (ITE) across demographic groups. Results provide evidence for all three key gradients, namely an education gradient, nativity gradient, and cohort gradient. The varying characteristics of the education and nativity gradients in online dating’s effects likely reflect how strongly education and nativity structure social and romantic opportunities in different countries. The cohort gradient reveals that later-born cohorts experience weaker negative effects on marriage and fertility, reflecting shifting social norms and life course expectations.
These findings highlight how online dating reshapes relationship trajectories in stratified ways, reinforcing existing social structures while expanding opportunities for diverse partnerships. Methodologically, this study advances causal inference in partner selection research by leveraging AIPW and Causal Forests, offering a fine-grained, cross-national perspective on contemporary dating and family formation.

Authors

• Zerui Tian, University of Oxford

• Zhuang Hao, Huazhong Agricultural University