Two-Sided Estimation of Mate Preferences for Similarities in Age, Education, and Religion

We propose a two-sided method to simultaneously estimate men's and women's preferences for relative age, education, and religious characteristics of potential mates using cross-sectional data on married couples and single individuals, in conjunction with a behavioral model developed in game theory and discrete choice estimation methods developed for simpler, one-sided choice situations. We use fixed effects to control for characteristics that are observed by the opposite sex but are missing from our data. Estimated mean preference coefficients determine the average degree to which measured characteristics of individuals affect others' evaluations of them as marital partners, whereas the model also accounts for variation of preferences around the means and for limitations in men's and women's information about members of the opposite sex. By assuming that each individual chooses freely from the set of potential partners that he or she finds available, we estimate preferences without having to observe these sets or specify any details of the matching process. This makes our method robust to unknown features of the process. Application of the method to data from the first wave of the National Survey of Families and Households indicates roughly symmetric or complementary preferences of men and women for age, education, and religious affliation characteristics of potential mates and a much stronger preference for religious homogamy among conservative Protestants relative to mainline Protestants than was suggested by an earlier, retrospective study of religious differences in the temporal stability of marriages. Our method should be useful in many situations in which voluntary pairings have arisen through some complex process, the details of which have not been recorded. Besides marriage and cohabitation, data on employment, college attendance, and the coresidence of elderly parents with adult children often have this character, as do some biological data on nonhuman mating.