Ten years ago, online dating was seen as the last refuge of the desperate; today it’s mainstream enough that the worried parents of some of my unmarried friends urge them to keep their online profiles updated. Estimates vary, but tens of millions of Americans use sites like Match.com, PlentyOfFish.com, EHarmony, OKCupid, and Chemistry.com, along with niche dating sites like JDate (for Jewish singles), Gay.com, and SugarDaddie. A single person today doesn’t have to be content with whom they might meet at work or a party, or at church or the local bar. They can go online and browse millions of profiles to find that special someone who shares their love of yakitori, corgis, and Ultimate Fighting. Many sites take things one step further, offering, through proprietary algorithms, to pick a perfect match for a person out of the cacophony of online profiles, with an accuracy that puts human matchmakers to shame. It’s “science-based” matching.
A new paper, however, calls into question much of that science. Writing in the current issue of Psychological Science in the Public Interest (PSPI), five psychologists who specialize in the study of human relationships argue that, while increasing the potential matches does increase the odds that a person will find a romantic partner, the rest of what online dating sites offer doesn’t do much at all. And some of the services the websites offer might backfire, causing users to overlook people they might be happy with. “By suggesting that compatibility can be established from a relatively small bank of trait-based information about a person—whether by a matchmaker’s algorithm or by the users’ own glance at a profile—online dating sites may be supporting an ideology of compatibility that decades of scientific research suggests is false,” the authors write. Or, as lead author Eli Finkel puts it, “We looked at the quote-unquote evidence they mustered, and said this doesn’t pass muster.”
Questions about particular matching methodologies have been raised before, sometimes by matchmaking websites themselves—EHarmony or Chemistry.com—attacking each other’s matching algorithms as they compete for users. Because the algorithms are proprietary, it’s difficult to say much about their specifics. But the authors of the new online dating paper argue that there’s a problem with the very concept of compatibility these sites are selling. Contrary to what they promise, compatibility can’t be calculated or predicted in the hypothetical, either by asking people what they want in a partner or by asking them a detailed set of questions about themselves that’s then used to calculate the best matches for them.
The problem with asking people to describe their ideal match is they’re not good at predicting what they’ll actually want. Studies of speed dating, for example, have shown that the potential partners daters reported feeling romantically attracted to didn’t actually fit the descriptions they had given beforehand of their ideal match. The process of browsing a number of online profiles can exacerbate this disconnect, as people rely on easily comparable qualities like income or height rather than less quantifiable ones that might play a larger role in interpersonal chemistry.
Matching algorithms are supposed to solve this problem, but Finkel et al argue that this is impossible. There are traits that are good predictors of how likely a person is to end up in a happy, stable long-term relationship—neuroticism in one or both parties correlates with unhappy relationships and a greater chance of breaking up, as does having parents who divorced. These, however, are not the sort of traits dating websites are sussing out when they promise to find that special someone for their users—unless you count EHarmony’s policy of rejecting people it thinks won’t be good matches for the site’s clientele.
Again, the specifics of the algorithms are secret, but the traditional idea of compatibility tends to focus on personality. But while it’s a common and intuitive idea, the PSPI authors point out that the psychological evidence that similar personalities predict successful intimate relationships is “on the whole, weak and contradictory.” In one study they cite, degree of similarity in political orientation, religiosity, and life values was not associated with levels of marital satisfaction. Another study, of newlyweds, found that enjoying the same leisure activities caused husbands to feel closer to their wives, but didn’t matter at all to the wives.
How about “complementarity,” the idea that opposites attract and complete each other in a relationship? According to Finkel and his co-authors, “empirical evidence that differences between partners benefit relationships has been even harder to come by than evidence for the benefits of similarity.”
Asked about the paper, an EHarmony spokesperson provided a statement pointing to a 2009 study commissioned by the company that found EHarmony matches result in 5 percent of the total marriages in the U.S.—an average of 542 marriages a day. “EHarmony’s matching system,” the statement continued, “is based on years of empirical and clinical research on married couples. As part of this work we have studied what aspects of personality, values, and interest, and how pairs match on them, that are most predictive of relationship satisfaction.”
Finkel is quick to emphasize that he’s not against dating sites. If the matchmaking abilities of the sites have in his opinion been wildly oversold, the simple fact that the Web has given single people the opportunity to meet many more potential mates is a boon. The trick, he argues, is to take a potential relationship from online to offline as quickly as possible. The best predictor for how well two people will “really click,” as he puts it, rather than simply click on each other’s profiles, is how much they like each other’s company. There is a budding research field on what goes into that rapport—the combination of body language and verbal tics and even smells—but for daters it’s probably easiest to just do the field research one date at a time.