In general, married couples are homogamous. That is, they are more likely than not to match on a whole host of characteristics: age, income, education level, race, religion, immigration history, attractiveness level, and more.
But, does homogamy really translate into compatibility? Or, do we just think it does?
OKCupid set about to find out. This is the second of two posts about recent revelations that they, like Facebook, have been doing experiments on users. The last one was a depressing look at the role of attractiveness on the site. This one is about the impact of match ratings. Yep, they lied to see what would happen.
OkCupid users answer a series of questions and the site then offers a “match rating” between any two users. The idea is that people with a higher match rating are more homogamous — by some measure not identical to those that sociologists typically use, to be clear — and, therefore, more likely to get along.
The first thing they did was artificially alter the match rating for couples whose true match was only 30%. Users could read the profile, look at the pictures, reviews answers to questions, and see a match rating. In other words, they had a lot of information and one summary statistic that might be true or false.
People were slightly more likely to send a message and continue a conversation if they thought they were a 60% match or better. This is interesting since all these couples were poorly matched and it shouldn’t have been too difficult to discover that this was so.
Rudder’s interpretation of the data is that you can make two people like each other by just telling them that they should.
Or maybe, he considered, their algorithm was just terrible. So, they took couples who matched at the 30, 60, and 90% rating and displayed a random match rating that was wrong two-thirds of the time. Then, they waited to see how many couples got to exchanging four messages (their measure of a “conversation”).
The lower right corner suggests that the ideal situation is to be a good match and know it. Likewise, if you’re a bad match and you know it things probably won’t get very far. But the difference between actually being a good match and just thinking you are isn’t as big as we might think it would be. At least, not in the space of four messages.
So, does homogamy really translate into compatibility? Or, do we just think it does? Maybe a little of both.
Cross-posted at Pacific Standard.
Lisa Wade, PhD is an Associate Professor at Tulane University. She is the author of American Hookup, a book about college sexual culture; a textbook about gender; and a forthcoming introductory text: Terrible Magnificent Sociology. You can follow her on Twitter and Instagram.
Comments 8
YaronD — September 10, 2014
[ You have a broken link to your earlier OKCupid post. The URL in the link has 18 as the date, but the post is published for the 19. ]
I have a big problem with them intentionally lying to their users (while the meaning of the match estimation is never disclosed, and quite possibly pretty meaningless to begin with, as long as they do provide a service explicitly showing this number then they should show this number rather than a value that they know and believe to not be that number).
They also don't provide nearly enough data to estimate the statistical significance of the results they show, so while the results look like they indicate something (and sadly an unsurprising something) they actually don't. At least as far as we can tell from the published information. So it would have been interesting/useful info, and certainly feels right (yay to cognitive bias) but should be ignored just as much as a regular research article providing the same (lack of) data.
Bill R — September 10, 2014
Or maybe they lied about the results of the study too. Why trust anything they'd say?
Eric Anderson — September 10, 2014
I think the results are a lot weaker than they (and by extension, you) make them out to be: As a measure of relationship compatibility or success, exchanging four messages is an incredibly low bar. That's not even going on a single date, let alone actually liking it, having a relationship, or being happy together.
Woz — September 11, 2014
The big problem is the assumption made by researchers here that people spend time reading profiles in detail. I'm willing to bet there are a good number of people who see a high match percentage and then initiate a conversation. So while all the info proving they're not really that compatible (at least as okc defines it) is right there, if they're not looking at it, they've no reason to distrust the match percentage.
Of course, it's still interesting that people would just trust what an algorithm says without even looking into it themselves, but to assume they accept it despite the evidence is to assume they evaluated the evidence.
FeministDisney — September 11, 2014
although I know that this sort of data is enticing to handle -- everyone looks at OkCupid -- I'm kind of disappointed you did because it's... nothing. Assumptions drive the conclusion of every test they formulate.
Like why even ask why people assume the number mean something when other things don't match?? It's because 99% of the time, when you use this site, the compatibility # matches pretty well. It's nothing more than past experience of the number actually meaning something, reinforcing the idea that the current, pseudo-match number is.
I have tried to have an open mind but I've discovered every time that someone below a 60% is just so dissimilar as to be from a different "world" in terms of our experiences, and of course experiences drive who we connect with. Sure, I read through profiles, but since I trust that the algorithm works the same way -- and I've always had better convos with people who are 70%+ -- reasonably, I might give a break to someone who seems like a 60, but rates like a 80, because maybe there's something I'm not noticing.
Michelle F. — September 11, 2014
Match ratings are, as I understood it, determined in large part by quiz questions that tend to center around values, preferences, and worldview, in general and specific to relationships/sex. As a previous poster mentioned, low match numbers tend to indicate a fairly reliable incompatibility, such as that between a liberal and a fundamentalist Christian, or a person with a high sex drive and desire for kink and someone who prefers less frequent and less adventurous sex .
The convenience of the match percentage is that these kinds of preferences don't always come up in early courtship and (as this article would support) and thus you could be attracted to and date someone before finally realizing there was a critical incompatibility that prevented you from being happy in the longer term.
Basically, match percentage is unscientific and really doesn't bear any adequate correlation to homogeneity or heterogeneity in the sense that sociologists tend to think of it. Sure, some values correlate to demographics, but speaking anecdotally after years of OKC dating, I met and dated people who were more diverse in race and socioeconomic status than people I would have met "IRL," and I did take match percentage into account.
Masculinities 101 Week in Review: September 12, 2014 | Masculinities 101 — September 12, 2014
[…] Also featured is a piece that considers the compatibility factors of homogamy, when couples are “more likely than not to match on a whole host of characteristics: age, income, education lev… within the context of dating sites like […]
Bi Datingsites — March 18, 2016
Nice post
really helpful, visit some other sites also
http://www.bisexual-datingsites.com/