One of the traditional reasons for going to college is to meet interesting people. There are a lot of different sorts of relationships that come from meeting at college: academic discussion partners, business partners, roommates who share nothing but an address, lifelong friendships, rivalry, study groups, sexual partners, … . In my parents’ generation, it was common for women to go to college specifically to seek a worthy husband, but this became an unfashionable goal (at least to admit aloud) during the 1970s, and has remained unfashionable since then.
Colleges are still excellent places to meet people, particularly if you are interested in meeting intelligent people, but how good are they at finding long-term partners to live with?
I just read parts of a study Searching for a Mate: The Rise of the Internet as a Social Intermediary, which I was pointed to by a Cost of College blog post: The Internet easily overtakes college as a way to meet marriage partners.
The main result of the paper is summarized in one figure:
The graph is almost self-explanatory: about 10% of couples met in college, and this gradually rose from about 5% in 1940, peaking around 2000. Meeting through friends has been the dominant way for couples to meet, but has been gradually losing out to the Internet. The steadiest declines have been in meeting through family or primary/secondary school, and in recent years, the only growing trends have been meeting online and meeting in bars and restaurants.
When I first saw a summary of this report, I was suspicious of the method used: they had a survey where they asked couples when they had met and how they had met. This could easily bias the results, since couples that meet one way may be much more likely to stay together than couples who met some other way. The authors were aware of this risk and took several precautions to measure how much effect the bias would have. One check was having some questions that matched a 1992 survey, to see if there were differences between the 1992 answers and the answers from couples who had been together since 1992. Another check was looking at correlations between one-year breakup rates and how couples met. I’m reasonably convinced that the data they present is a good approximation to the ways couples met in each year and not overly affected by “dissolution bias”.
Of course, a lot of these modes are not mutually exclusive (people can meet through friends whom they met at college, meet online while at college, meet at bars while with friends, and so forth). This is addressed in the detailed notes on how they coded the answer to the free-response question:
It is also important to note that most couples list more than one chain of connection to their partner, and all the relevant chains are coded. So, the Respondent and the Partner may meet at church, but also have a mutual friend outside of church.
I’m a little confused, though, as I would expect the percentages to add up to much more than 100%, if most couples have more than one code, but I’m seeing sums only slightly larger than 100% (maybe 115%). I’m still a bit dubious about the coding of the data, as the authors can have easily and inadvertently colored the data by how they coded the “how we met” stories. I’ve no idea how they would have coded how my wife and I met (through the Society for Creative Anachronism, when we were both working for Cornell University). Coworkers? through friends? at college?
Assuming that the data is good, it looks to me like meeting online has saturated at 22% for heterosexual couples, but is still going up for same-sex couples. If current trends continue, within a decade there will be only 4 important mechanisms for heterosexual couples to meet: bars and restaurants (30%), through friends (25%), online (22%), and at college (10%), with all other modes combined only making 15%. Almost all same-sex couples will meet online, though gay bars look like they’ll continue to grow their businesses.
Of course, continuing the percentage curves independently is likely to lead to nonsense fairly quickly. The percentages currently add to more than 100% (because of people meeting in multiples ways), but extrapolating out past 10 years suggests percentages adding to less than 100%, which means either that the projection is nonsense (likely), or that some new mode of meeting will be taking over—which the survey has no way of predicting.
One can’t take surveys like this too seriously as ways to predict the behavior of individuals, but as broad-brush descriptors of American society as a whole, they can be fairly revealing.