Educational homogamy and entry into first marriage: analyzing discrete-time event-history data with conditional logit models

Jani Erola, University of Turku
Juho Härkönen, Stockholm University

We use discrete-time conditional logit event history models to analyze entry into homogamous and heterogamous marriages. Educational homogamy is usually analyzed either using log-linear models on cross-sectional data or regular event-history models. The advantage of the former approach is the possibility for separating effects due to educational level from homogamy and heterogamy, but it has the disadvantage of having to rely on intact marriages or newlyweds (while disregarding the unmarried). The latter allow analysis of the process of entry into homogamy, but are limited in testing for hypotheses of the forces of attraction that lead to educational homogamy. Discrete-time conditional logit event history models combine the advantages of both approaches. We demonstrate the use of these models by analyzing entry into first marriage according to own, spouse’s and mother’s educational level with Finnish registry data from a cohort born 1971-80. The results suggest that homogamy according to mother’s education does not strengthen the tendency to homogamy according to own education. Rather homogamy according to background is likely to compensate the lack of it according to own education while the match according to own qualifications are considered as more important. The results show the usefulness of the applied method for the analysis of educational homogamy.