The role of marriage has undergone profound change in recent decades. Divorce and cohabitation have become commonplace, and age at first marriage has increased for both men and women. These shifts in marriage patterns have been accompanied by changes in assortative mating-that is, the types of partners individuals choose when they do form unions. This study will contribute to the literature on marriage and assortative mating in four ways. First, it will estimate assortative mating based on multiple outcomes. Previous research has examined assortative mating on education, testing the extent to which mating is homogamous (partners select partners with similar education), and examining trends in such patterns. This study will examine assortative mating on income and other outcomes. Second, it will develop a methodology for estimating the determinants of multiple union types, generalizing the binary outcome for marriage to an ordered polychotomous outcome for cohabiting, married, or neither. Third, it will allow for non-linearity and non-monotonicity in estimating the relationship between individual characteristics and union status on the one hand, and degree of assortative mating on the other. Fourth, it will compare assortative mating patterns for cohabiting and married couples and for early and later cohorts. Lam's model predicts that cohabiting couples will be more similar than married couples, and that assortative mating has increased over time. Fifth, it will estimate a model of partner choice that allows for selection into marital or union status. Because married (and cohabiting) couples are not a random sample of the population, estimates of one partner's characteristics on those of the other is subject to sample selection bias. This can be addressed within the selection framework proposed.
© 2024 Center for Statistics and the Social Sciences