Unmarried Cohabitation and Union Stability: The Cohabitation-diffusion-hypothesis Revisited

We revisit the U-shape diffusion hypothesis of Liefbroer and Dourleijn (2006) by replicating the impact of cohabitation diffusion on union dissolution in different union cohorts in ‘European’ contexts, the USA, and Canada. The impact of cohabitation diffusion on union dissolution is expected to take the form of a U-shaped relationship because of selection effects into cohabitation of ‘separation tolerant’ couples in low cohabitation contexts, or selection into marriage of ‘conventional’ couples in high cohabitation contexts. Where Liefbroer and Dourleijn studied data from three birth cohorts from 1953–1967, we use data from women in the Harmonized Histories (HH) of 21 countries and 14 (5-year) birth cohorts (1930/34–1995/99). We perform multilevel discrete time event history analyses on the pooled countries data and we check to what extent the variation in the ‘cohabitation effect’ can be explained by the percentage and percentage squared of the cohabitation rate in country-birth cohorts. Our findings indicate a U-shaped relationship between the cohabitation rate and the relative risk of union dissolution of cohabiting couples, with prior cohabiting, currently married couples having a risk of separation lower than married couples. Implications of our findings are that future research should not neglect the role of selection.