I listen to a local Christian talk radio station when I'm driving, and today I heard another discussion of the alleged perils of premarital cohabitation. I say "alleged" because while I'm sympathetic to the basic argument, I'm not convinced we're measuring the effect accurately. The guests interviewed made a number of comparisons between married couples who never cohabited and married couples who did, and according to a variety of indicators non-cohabiting couples do much better: they're happier, they're more likely to stay married, etc.

Nothing was said about the identification strategies used in the studies mentioned, however. What appear to be simple differences-in-means comparisons omit a lot of relevant information, and the decision not to cohabit before marriage is likely to be related to a lot of other characteristics that might make for a long, happy marriage. The difference in the probability of divorce for cohabiting and non-cohabiting couples reported on the radio is an upper-bound estimate of the real effect of cohabitation on the probability of divorce. A quick Google Scholar search for ["treatment effect" cohabitation] doesn't yield much that I can access. Any cites identifying the treatment effect of prior cohabitation on the probability of divorce would be greatly appreciated. If this hasn't been done, it would be a great paper topic for someone interested in issues related to labor and the family.

I find this interesting for reasons beyond just curiosity or concern for the welfare of society. There is also the matter of wise stewardship. Systematically biased information about social phenomena lead to inappropriate policies, programs, and assorted endeavors. In a world with scarce resources, this means that we might not be maximizing the bang we're getting out of our charitable and educational bucks. Time and money directed toward counsel against premarital cohabitation on the belief that it leads to a substantial deterioration in the quality of the future marriage and an increase in the probability of divorce will be poorly spent if the impact of premarital cohabitation on marriage quality and duration is small. If the effect is small, then the time and money groups are spending fighting premarital cohabitation might yield greater fruit if they are used elsewhere. Here are my priors: I wouldn't be surprised if the treatment effect of cohabitation on the probability of divorce is positive, but I would be very surprised if it's as large as the people I heard on the radio are saying. I don't know, but before we sound alarm bells about the effects of cohabitation on different indicators, we should make every effort to measure those effects accurately. I'll post on this again as soon as I find out.