23andMe Couples Correlated

For Valentine’s Day, we analyzed data from 15,298 real-world couples who had children together and found that people paired with others who were more like themselves than they were different.02122014_ValentinesDayPost_4

Opposites, it turns out, may not attract after all.

Partners don’t just tend to be about the same age, according to our data, they also tend to share the same education level, a similar Body Mass Index and even a tendency to be apologetic.

Athletes tended to pair with other athletes, downhill skiers with downhill skiers, hikers with hikers and weightlifters with weightlifters. We also found that people who were punctual paired with others who were on time, and that vegetarians tended get together with other non-carnivores.

These are real correlations, but it is important to note that a correlation does not imply causation.

Since correlations can be misleading in other ways, one of 23andMe’s computational biologists, Aaron Kleinman, looked a little deeper. Aaron computed the effect size of an individual’s phenotype on their spouse’s phenotype, controlling for age and ethnicity. Even when the effect size is strong, it is still important to note that that doesn’t indicate causation either. We don’t know why couples share some of these phenotypes.

And this doesn’t give us any special insight into why people fall in love. It could be that they share interests in common, but those shared interests could also come after they became a pair. For example, we found strong odds ratios for people who were former smokers being coupled together. Maybe overcoming their shared history of kicking the habit brought them together. Maybe not. We also found individuals who frequented drive-through food establishments paired with others who did the same,  and we found that individuals who could speak a second language paired with partners who also spoke a second language. Some of these correlations seem to make intuitive sense to us. For example, if you’re with someone who goes to drive-through food places, odds are you’ll likely go too.

It is also important to note what we did not find. There is no way to determine if two individuals are a couple just by looking at their DNA. Still, we found a lot of correlations.

So what about opposites?

We did indeed find that in a few ways opposites appeared to attract. In our data, it appears that a night person is more likely to get together with a morning person. Those who attract mosquitoes end up with those who don’t, and luckily people with a good sense of direction appear to pair with those who don’t have one.

But another of our researchers, Emma Pierson, found more data that appears to show that these cases of opposites attracting were the exception to the rule. Emma found that couples who had similar BMIs reported that they were happier. She didn’t find any effect for height, however, so no need to break up with your basketball player hubby.

Happy Valentine’s Day.






  • Jeff Messer

    So why the focus on couples with kids? :) Not enough data on couples with no kids perhaps?

    • Scott23H

      We don’t have people take tests as couples. This is also aggregated data. What we did is find couples by finding a man and woman with a biological child in common. Researchers call these “trios.” It can be helpful for other kinds of research. So the answer to your question is that this was how we identified couples, using the biological child in common.

  • Nadav

    You know, we are in the 21st century. You could at least put one or two same sex couples…

    • Scott23H

      Nadav, I meant to respond earlier. We agree with your sentiment, but in this case these are couples who we’ve identified as couples because they have one biological child in common. There is no other way to use genetics to identify two people as a couple. That’s why there are no same sex couples in this subset of data.

    • kre8ter

      This aint about you tryingvto fit in buddy! Keep your political views aside. Get real moron!

      • Abby

        How is the mention of same-sex couples inherently a political issue? Seems like someone else should get real.

      • Gary Calderone

        “aint”, “tryingvto”, who’s the real moron here?

  • jacjoe

    Why no analysis of DNA or ancestral markers, now that would have been a great addition, to see if opposites really do attract, from the same “tribe” or lineage. Some people swear they can identify others with the same facial features and can recognize their ancestry’s links, and may choose a mate from their “pool” without realizing, what are your thoughts?

  • Michele

    So this post is saying there is a correlation between people mating and have similar interests. Make sense those who like dogs might mate with someone else who has a dog or like dogs. Makes choosing a partner for child rearing which takes a significant amount of time easier. How about actual physical attributes that we see in others that can be seen as beneficial to our genes to pass on? I understand the word phenotype was used however how was the actual data interpreted? For instance, I am a female with dark hair and medium (olive) tone skin and pretty much most of the opposite sex who have been interested in me have been blonde and blue eyes. No tall dark and handsome males. Also, I have always had an aversion to male redheads and recently noticed I have natural red highlights. Would be great to get some of this data from this study.

  • Peter Parker

    The correlations between hikers, non-smokers, fast food eaters, etc, are not surprising. The more interesting question is in DNA similarities by comparing genetic profiles among paired 23andme members who state that they are married or can be shown to share a child together. Theory would predict that couples are more likely to be genetically similar broadly speaking (i.e. in terms of race or ethnicity), but they should be more likely to be genetically dissimilar at the fine-scale level (e.g. MHC complex and family-level relatedness). It is known that female mice prefer the scent of male mice with MHC antigens that are different from their own; it would be interesting to test if this is also true in humans.

  • jacjoe

    A poster below my original comment that replied to me is still awaiting moderation, hence the new post, an the obsessive compulsive need to be proactive. I was obviously not advocating inbreeding, really?? You immediately went to that? Wow, just wow. Sharing physical traits may be traced back thousands of years. Even we may not know our true lineage without DNA testing. The point I was making is simple. A DNA company doing a study about opposites attracting, would seemingly logically, look into DNA to see if selection of a mate may show ancestors from a particular region of the ancient world, perhaps choosing from that same region. Lower intelligence is becoming more common due to evolution not being able to properly occur in today’s societies with the advent of modern medicine and our food and basic necessities being met artificially. Inbreeding on a mass scale is not the cause.

  • Pat Garver

    I don’t think most of the analogies hold water. People find mates usually in social groups with same interests. Hence hikers hang with hikers, runners hang with runners. Hikers are not likely to find a mate among pilots because they are least likely to actually meet pilots but very likely to meet other hikers. I think association is the key.

  • James M. Taylor

    From what I can gather I am totally Irish Anglo. Only a bit of Italian and Bulgarian. My wife is totally Slovak. We share nothing in the gene pool, but think alike. This is my second marriage but has lasted 26 years. For all you guys out there, don’t try to win the argument. When they win, you eventually win.
    JT

  • James M. Taylor

    Very scientific approach. But I don’t believe there is a total genetic link to whom we end up with. Science has shown that some of our ideals and character are formed by how we were raised and the values we were taught. No religion intended. I think we can all be good people without that.
    Had a little old lady for a professor who would sit on the floor with me and talk about genetics. I learned more from the 5 minutes we spent than I did in her entire class. Her take was that there is no genetic marker for who will end up with who. She did say it was important that we be out looking. Been with my wife for 26 years so I guess her advice worked.

  • Lita Gato

    Yeah..No same sex couples? And although Im with another female we are very different. This is sort of a silly paper.. Of course an athletic person doesnt want to be with a over weight lazy person or intellectual with a lay person.. Did they really need to research that?

    • satansidekick

      What is a lay person? Did you mean lazy? Don’t understand that comment.

  • James M. Taylor

    Seems to be a lot of discussion about the lack of same sex couple data. I guess the first thing that comes to mind is that they do not contribute to the gene pool. God love them. My next door neighbors are lesbians and they watch our cats. Never saw that as a problem.

  • satansidekick

    Been married twice and the only one who was close to similar genetics was my ex- husband and that was a nightmare. My current husband also used 23andme and we share zero. We have been together for over 15 years and are very happy

  • Guest

    Interesting report! It remind me an article published last year on the same topic, examining animals instead of humans. In short, opposites don’t attract after all among many animal species as well, from earthworm to koalas. Those animal “couples” had been reported to resemble each other in characters such as age, size, body condition, behavior, and certain genes. (the link to the article, just in case anyone’s interested http://www.jstor.org/stable/10.1086/670160)

  • Scott23H

    ResCogitans,
    There have been a lot of studies looking at Immune System Compatibility. We have looked at this but haven’t found any clear so far.

Return to top