SIO386: Actually, Opposites Very Much Do NOT Attract

Transcript

This transcript was generated automatically and may contain errors — the audio is the record.

0:00:16 Thomas: Hello and welcome to Serious Inquiries Only. This is Episode 386, I'm Thomas Smith, and I'm excited to bring you another interesting science thingy today. I'll be speaking with Tanya Horowitz, one of the authors of a new meta study, a very large meta study that looked at whether in a human, well, I guess it would be reproduction, to put it, you know, not romantically.

Saw this study and thought it was fascinating, and not just for like the top line takeaway, which I think is, I don't know, kind of obvious, but for the details. You know, there's a lot of interesting data in here. There are a lot of traits that were looked at, a lot of fun stuff to talk about. So results were surprising, which weren't. And also, this area of study is something that I hadn't really thought about before.

So, talking to Tanya a little bit about this specific conjunction of specialties that she has that's what we're going to talk about. Before we do, a couple of things: episode, Stormy Decisis, also known as Dr. Alan Smirbeck, that we might do a Q&A, and I posted the idea on Patreon, and we got a lot of takers. We got a whole lot of questions, so if you'd like to participate, ask Stormy a question, go to patreon.com/seriouspod, and find the post, and post a question there.

Maybe vote on the questions you want to hear. We already got a good sample there, so it's going to be fun. Can't wait to do that. Excited to get back on with Stormy.

And while you're there, you know, just pledge on patreon.com/seriouspod anyway so that you get ad free shows, most importantly, and the bonus stuff, and in the coming months we're going to be looking to add on to that bonus stuff a little bit, um, some exciting announcements that I haven't really gotten to say yet, but stay tuned. But at any rate, to support the show, like all your much cooler friends have already done, go to patreon.

Com slash seriouspod otherwise, otherwise you're just not going to be super duper cool, like, like all the cooler people who do it. I don't I'm trying to peer pressure. I've been trying a different angle. I think also, try these cigarettes. No, don't do that. That's bad. Anyway, patreon.com/seriouspod. Without further ado, let's talk to Tanya Horowitz, who is a grad student out of University of Colorado Boulder, about or not opposites attract in human dating or reproduction. Here we go. How are you doing today?

0:02:31 Tanya: I'm doing really well. How about you?

0:02:32 Thomas: I'm doing quite well. I'm excited to speak to you. I think that this is going to be a fun topic. It's something that, you can't help but be fun and relatable in many different ways, I'm sure.

0:02:42 Tanya: That's definitely what I felt when I was working on it.

0:02:46 Thomas: Yeah, yeah, definitely more and less fun studies to do. This feels like a more fun one to me.

0:02:51 Tanya: Yeah, it's also one of those fun ones that can interest the public, which is, you know, not 98% of science work, so that was cool.

0:02:57 Thomas: Yeah. Yeah well before we get going, do you want to give us a little bit of your background? Kind of how you got into this this area, your studies, that all stuff.

0:03:09 Tanya: So I'm a PhD candidate at the University of Colorado at Boulder. In a psychology program, but I'm also in a program called the Behavioral Psychiatric and Statistical Genetics Program. So, I'm sort of at a crossroads between psychology, statistical genetics, and And this project actually came out of a failed first year project a bit. It pretty unexpected. Yeah. Was a lot more phenotypic focused than what people in my program usually are working on.

0:03:44 Thomas: And define that for all the not scientists listening, if you don't mind.

0:03:47 Tanya: Right. So, sorry. We're looking at, we're not specific genetics. We're looking at people's actual trait values for things. But it's actually an important question for geneticists.

0:03:58 Thomas: Okay, that's interesting. So that might explain, and correct me if I'm wrong on any of this, I'm not a scientist or anything, but when I first saw the study kind of read through the headline, I would have assumed it was just like a psych study. But now that you say this, is that kind of why that, more of this genetic underlying to it, but it sounds like the way you got to this project was a little bit, you know, roundabout.

0:04:18 Tanya: Right, it totally can be just applicable to a bunch of different fields, psychology, sociology, economics, etc. as a geneticist, this is how I came into it. Basically, a lot of genetic designs and models assume that mating is quote-unquote random. So basically, the models themselves, not the scientists who made them, kind of assume that partners do not correlate for traits, right?

0:04:48 Thomas: We talking humans or anything, any species?

0:04:51 Tanya: We are mostly concerned with humans, though this is actually something that has been looked at in animals as well. Of course, what we found and what we suspected was that for a lot of traits, that's actually wrong. And Because it's wrong, it actually has the potential to bias a lot of our designs in genetics. So in my field, we are actually really interested in this question. Was initially going to be a genetics project, actually. But then I got really interested in the portion where I was looking for these base estimates of closely how aligned do partners tend to be for all these traits. That kind of just evolved into 22 meta-analyses about, you do partners correlate for these various things. And then based on a reviewer comment, we also tacked on our own original analysis of about 79,000 pairs in a biobank, British biobrank, to look at a other traits as well.

0:05:03 Thomas: Yeah, right.

0:05:48 Thomas: That's super cool. So I'm interested in why the the models would have assumed, you know, of random reproduction. Is it because when you talk about random, are you talking about with respect to specifically maybe genetic markers? Whereas because nobody would have thought like, well, it's completely random. Like I just might, you Know, I might end up having a child with somebody from the opposite end of the earth. Like, I doubt anybody thought that. But is it because we're not talking about like hobbies or stuff that's just like character traits or whatever, stuff like that, it have been talking with respect to actual traits that is more genetically represented or predicted.

0:06:05 Tanya: Right.

0:06:26 Tanya: So the issue here all goes back to the classical twin design which looks at twins. If you've ever heard of a twin study or a twin design, you're twins to each other and also people in their families. It was actually just a limitation of the model itself. So with more sophisticated models and larger family models, we can actually estimate assortative partner resemblance, which we call assortitive mating.

0:06:26 Thomas: Okay. Yeah. Yeah. And I first misread as assertive mating, and I think that's a whole different, that's not what she studied.

0:07:00 Tanya: Think that's A whole different thing that one could study though. It's actually just comes down to how many different parameters you can estimate with the information you have. So it was not because the scientists thought, oh, well, well, yeah, just think that people are mating randomly. They didn't actually believe that. It was just an assumption and a limitation of the model because you can only estimate so many parameters and get so many estimates with a certain amount of information.

0:07:00 Thomas: Yeah.

0:07:28 Thomas: There's only so many twins. If it was up to the scientists, we'd all be twins. It'd be like, yeah, everybody just have an identical twin. Well, not every identical. Some, I guess, have to be fraternal or whatever it is. So, you know, I feel Like that's if you had your way, that's what we do.

0:07:42 Tanya: There was really like larger family models too that allow us to look at more relations and estimate more parameters. And then when we became able to sequence the human genome a of decades ago, we had this way to look molecular things directly, right? Recently what people have found is that even those designs can be biased if partners are not randomly mating, which again, they for lot of traits at least. Earlier studies were more about trying to estimate things like heritability and at genetic and environmental components for any given trait and kind of trying to partition them into portions. What basically, we realized is that a lot of these estimates are probably wrong in part because of this. Yeah.

0:08:31 Thomas: That's fascinating. So like a really cool area, I guess, to back up a little bit to your personal story there. It seems like a really interesting area to be kind of both those topics, obviously, psychology, genetics, all that that stuff. But the, the sort of the marriage of the two seems particularly interesting. Other sort of things are looked at in your field or, do or you plan to study now or in the future? What I'm other kinds of topics or areas of study?

0:08:54 Tanya: I've also done research related to genetics. I've done research related to addiction. I'm a little all over the place now. I am really excited about a follow-up to the current study we're doing that looking at in the same British Biobank that we used for this initial study. We're actually looking at same-sex couples and of comparing those estimates, how similar are same-sex couples for these same traits? And do they differ in their degree of similarity compared to opposite-sex couple?

0:09:01 Thomas: Yeah.

0:09:24 Thomas: That will be important to note, obviously, when we talk about the study in particular, this, this is Is hetero opposite sex couples limited in this study, right? Or,

0:09:32 Tanya: The most part, yeah. And the two reasons that we decided to mostly limit it to opposite sex couples was because, for one, because there are a lot more opposite sex couples than same sex couples. Um, the latter would get drowned out anyway, because their effects, you know, would be overshadowed by the same sex couple. So it probably wouldn't really matter statistically, but, um, we've also seen studies that have suggested there are different patterns of similarity and same sex versus opposite sex couples. And so in order to like look at that more, it seems like it made more sense to look at opposite sex and same-sex couples separately.

0:10:10 Thomas: Yeah, that's going to be super fascinating. We'll have to have you back if that gets, you know, published.

0:10:15 Tanya: I'm excited about that one.

0:10:17 Thomas: But let me ask you this. There something that like just for you personally made you think this is the field I want to get into? Was was a lot of times I've had people on where, for example, I had somebody studying Alzheimer's and it was like, yeah, you know, I just, I worked in a, I believe a convalescent home for a little bit. I think it was, and I saw the effects of it and I want to get into it. I'm just wondering if there's anything like that for you in this specific area of study. Oh

0:10:38 Tanya: I mean psychology was my first love from when I was in my early teens. I read a lot about it. I, I summer courses in it. Um, I realized I wanted to do research in it. And, had a stint in a molecular biology lab that was mostly focused on post-mortem brains of people with bipolar, uh, disorder and schizophrenia. So, um, that's sort of what brought me into the genetics world. And I realized in this program, too, that I really like working with big data.

0:11:10 Thomas: Yeah.

0:11:10 Tanya: There is a lot of that in my field.

0:11:13 Thomas: Isn't genetics pretty much all that?

0:11:15 Tanya: I mean, the earlier twin family studies often are using smaller samples, but especially when we're thinking about the genome or just looking at these larger biobanks, like the one we looked at that has almost half a million people in it. We are dealing a lot with big data and statistics. And I liked quantitative stuff and mathematical stuff. It was kind of a marriage between these two things I really grew up liking, which stuff, and psychology, and social science, and sort of studying how people are in the world.

0:11:49 Thomas: Yeah. Yeah, for sure. Know, I it's I'm like a lot of these things seem like like, um, ripe areas for study or I'm sure they already have been, but you know, a lot of, a lot of lay people aren't familiar, know. But, Is addiction genetic? Is, you know, are some of these disorders genetic? Are those some questions that you're interested in in the future or is that a different area?

0:12:07 Tanya: I mean, it's something I've worked on. I think it's a very tricky thing because often We start with the, the lay question is, is it genetic or is it not? And what we found is, at least according to our models, almost every trait you can think of has non-zero heritability, which means there is some at least genetic correlation or association with about every trait. Now you get into a much more complicated question of causality, what are the reasons for this, and, uh, behavioral scientists believe that the outcome of most traits are the result of a very complicated confluence of biological and environmental things.

0:12:12 Thomas: Mm -hmm. Yeah. Mm

0:12:52 Tanya: And, you know, even saying this thing is 40 or 60% genetic or whatever is, is a lot more complicated than it sounds.

0:12:52 Thomas: -Hmm.

0:13:02 Thomas: Yeah, yeah. Definitely simplifies to just say genetic or not, but you know that's how, that's That's how we rubes think. No, I'm just kidding. But yeah, no, but there's definitely, I am sure, interesting parts about where it where it's like, well, this thing happens to be very highly heritable and this thing, you know, maybe not. It's kind of interesting to to study the differences there. And I am also sure when it comes to treatment or something like that, it's got to be a very instructive to kind any distinction there that you can, right?

0:13:02 Tanya: Um, And I think that, back, uh, when the genome was first sequenced, the idea was, oh, we'll, uh, sequence the human genome and we'll find the depression gene. I mean, that was probably an oversimplification. And then what we've actually found is the is so much more complex than I think so many people imagined. There are many, many, many different places the genome that correlate with most behavioral traits and non-behavioral traits for height. There are, I think, over 12,000 independent places on the genome that are associated with how tall you'll be.

0:13:39 Thomas: Yeah, yeah, yeah. Yeah. Jeez.

0:14:08 Tanya: And the effect sizes for especially most when they relate to something that is not biological or directly biological, you know, has something to do with behavior or psychiatry, these effect sizes are really tiny. It's really complicated. And And how it's going to tie into treatment, I think, is also a really complicated question. But I think that's the hope for most people working in, you know, the overlap between psychiatry and genetics.

0:14:25 Thomas: Yeah.

0:14:38 Thomas: Yeah. Well, I could keep asking you questions about that all day because it is super fascinating. But why don't we go ahead and tackle the study that you did do? And finish the of kind of how it started as something else?

0:14:46 Tanya: Yes, let's do that. Is more to it. I think I was originally going to look at was I was going to compare the trait estimates to the methodology is kind of a lot to right now. But basically, people who are geneticists are also interested in basically how are these these trait correlations for the phenotype or the thing that we observe, how do those relate to relationships between parents, right?

0:15:03 Thomas: Hmm. Yeah.

0:15:20 Tanya: An interesting question because as we briefly talk about in this study, there are a lot of reasons. There are a lot of possible reasons we see these correlations. And some of them are more, quote unquote, environmental or social. Not all going to have the same degree of bias that I was talking about before, or true downstream effects.

0:15:45 Thomas: Hmm.

0:15:45 Tanya: They will be reflected in different ways, depending on how the genetic, genotype kind of compares to the phenotype in that sense. So, initially what I was going to look at. I use the genome to further explore the mechanisms by which these correlations come to be, basically. So, didn't end up doing the second part. I just did the first part really intensely, ultimately.

0:16:11 Thomas: Well, I will say that as much as I do ask you to, you know, phrase it in a way that non-scientists can understand, I will Say, actually, this is a pretty science show. Like there's a ton of scientists in different fields who listen, professors and all that stuff. So I will SAY, we are into method talk. If you want, if we want to get into the methods, don't be, don't be shy, but yeah. So,

0:16:28 Tanya: Okay. Well, a methodology that basically correlates the odd chromosomes with their even chromosomes. Basically that can tell you with respect to a trait of interest, potentially about the genetics of the people that came before them. Kind of what we're looking at. So another we talk about the possible ways that partner resemblance can affect people downstream, stream, it's not just a source of bias. It also has the potential to have true downstream effects. Example, if tall people all mated with tall people and short people all mated với short people, what we would expect to see in the next generation is that we see basically a higher proportion of really tall people, and a higher proportion of really short people as well.

0:17:15 Thomas: Right. It'd get harder and harder to get into the NBA as a 5 '11 dude, I think is what it would be.

0:17:24 Tanya: Think that's actually a great way of putting it. This can also have similar downstream effects because it comes to the disease and disorder space. But I also want to point out this is kind of theoretical. We know statistically that this can have this effect, it's both because of the genes your parents give you and possibly for genetic and other biological phenomena they might be more similar for. That's sort of another one of the big implications in and genetics you people are more likely to be similar for a trait, that could kind of affect the distribution and the next generation for that trait for a lot of different reasons.

0:17:27 Thomas: Yeah yeah.

0:18:10 Thomas: Yeah, that's interesting to think about for all kinds of traits, and that's one of the reasons I was excited to to speak to you is because of the, of all the variety of traits that you looked at. And thinking about in terms of, you know, humans mating is really, really interesting and funny sometimes. Like, I wonder if that would also be true of non-genetic stuff. I think of the most obvious ones in this day and age, like political views.

You know, I like, there's whole different dating services for, you know, people on the right and people on the left essentially, and it's one of those things where I wonder if that would hold up at all to the same degree. Like, would we expect, you know, not that this is something you're looking at, but just a question that arises from this discussion is, like, people will get more extremely tall or more extremely short, possibly, would that also apply to stuff like politics?

Will we get even more polarized because of these like patterns? I don't know, or is that something that hopefully other outside influences could change something like that, rather than it being more predetermined? But I feel like my knowledge of the studies is that somehow political views are a little bit predetermine, but that's just off the cuff.

0:19:12 Tanya: Mean, when it comes to politics, that's definitely a theory and a way that you could think about it, right, with respect to the political polarization, because you're right, it's that same kind of, at least statistically, it's the same idea. The making polls more extreme. But when it comes to any trait, including politics, I also think about this when I think about wealth, because why some economists study this, partly because of the subject of wealth inequality. On the other hand, I think of politics and as sort of these really unstable variables, right?

0:19:34 Thomas: Yeah. Yeah, hopefully.

0:19:50 Tanya: Conceives of their own political views and what one believes in the issues one has to think about is so variable. And from one environment to the other. That being said, if you are raised by two people who have very similar political views, right, that surrounded by people with similar political views that's going to likely have a huge influence on you in some way. That doesn't mean you couldn't end up going in the other direction because we know that can happen. And, you know, if your parents were randomly matched with respect to political views, I think that that would just have a different kind of effect on people for a lot of reasons. Yeah, it is so complicated. And we can think about this with psychiatric or other behavioral traits as well, because what your probability of either being diagnosed with something or having a certain trait could change depending on your environment, what time you're living in, all this kind of stuff, which sort of further complicates the question, right?

0:20:21 Thomas: Right. Right? Yeah. Sure. Yeah? We actually recently did an episode kind of on that, on diagnosis, and one of the points that my raised was that it's hard to measure anything like, you depression from one cohort or generation to the next, because it's like, how you know what one generation would have referred to as, this you is too much suffering in some way. Like, we don't know how different generations would have answered some of these questions about their wellbeing. You

0:21:23 Tanya: Totally. And I've seen that really done to the extreme, like people trying to apply the Gini coefficient to hunter gatherers. And it's kind of like, how does it, how do you do that? I'm not an economist, but it sounds like it could have some holes, you know?

0:21:40 Thomas: With hunter-gatherers that I think is perhaps trying to draw more conclusions than we can in this day and age, but who knows? You want to talk about the methodology of this one, of the one that you ended up doing and not the one that was kind of the false start?

0:21:46 Tanya: Yes, I would love to. So basically, we did 22 meta-analyses. And the meta-analysis, for those who aren't familiar, it's basically when you're looking at the same thing across many different studies or samples, and you're kind of calculating somewhat of an average effect or a weighted average effect across all those studies. So we looked at 480 correlations from 199 studies across 22 phenotypes or traits. We wanted to see, okay, so typically how similar do spousal or co-parenting or cohabitating pairs tend to be for these traits. Um, then the raw data analysis looked at, uh, almost 79,000 pairs in a British sample. Um. And we, for that one, we looked at 133 different traits.

0:22:46 Thomas: Imagine just the data kind of determined what traits you could look at there and that's the difference.

0:22:51 Tanya: The meta-analysis specifically, you know, we want to have at least a few studies for each trait, right, or else it wasn't really a meta-analysis. So that was more restricted to more commonly studied traits for the most part, and ones that we have kind of focused more on in the field. But the cool thing about the original raw data analysis we did with the 133 traits is you can do that a lot quicker than meta-analysis. So, instead of looking at these 22 sort of pre-selected traits that already had sort of preconceived hypotheses about, we could ask ourselves, are people really more likely to be similar for these traits? Or is it just for the obvious ones, quote unquote, like height and educational attainment? Because we're looking at all kinds of things in the British sample. We're looking at red blood cell count, breastfeed, whether you're breastfed as a baby.

0:22:59 Thomas: Right. Hmm. Yeah. Is always my first question on a date. Like what's your red blood cell count? You know, roughly, is it like 10. I actually don't know how, I don't know how it's measured. So I'm not very good at asking this question. Seven? Mine's, mine's seven. I don't know.

0:24:01 Tanya: I It's like the kind of of stuff that you most people probably don't know about their lifelong partners or unless you have to start doing medical testing and stuff. Right, it was really interesting to see, because I think it kind of adds this extra layer to this narrative of opposites do or don't attract. Right. Because we found overwhelmingly for over 80 percent of traits at a statistically significant level, partners are on average more likely to be somewhat similar for traits than different. And it is very, very rare that you see people are actually more likely to be dissimilar. Now, there were a handful of traits for which that was basically zero, meaning you were about as likely to be with someone who was the same as someone who is different for a trait.

And we found three traits just in the British sample with relatively small negative correlations, three out of 133. not a lot. And that kind of, we knew, at least the literature had said that people are more likely to be similar in partnerships for most traits. So that wasn't really a surprise, but it was kind of interesting to, well, as I was saying before, tack on all these sort of random traits that you wouldn't necessarily know about someone because it adds an extra layer to this opposites attract, opposites don't attract. Because it kind of demonstrates that there are other factors that determine not just what you see or what's salient, but factors that stratify you, that put you near people who are similar to you for that trait. Or maybe

0:24:08 Thomas: Yeah. Yeah, yeah, yeah. Yeah.

0:25:51 Tanya: There is a preference, but it's for another trait that that correlates with that trait.

0:25:55 Thomas: Right, right, right. That we should go through more of the traits, do you want to talk about kind of some of the major ones or the interesting ones?

0:25:56 Tanya: Yeah. Well, ones that we tended to find larger correlations for across both analyses were those related to educational attainment, many of the substance use traits we looked at, and in our meta -analysis, we didn't have this in the British sample, but we looked political and religiosity, and those were at the very top. Also things like some of the sexual characteristics, like how many sexual partners have you had? Have you ever had a partner of the same sex? Those showed pretty reasonably high correlations.

0:26:20 Thomas: Yeah, Yeah.

0:26:35 Tanya: And The biological traits, things like height, BMI, weight, waist to hip ratio, those showed pretty consistently positive correlations, but they weren't quite as high, um, as some of the psychological traits as well. And then the ones that really showed the lowest correlations were for things like neuroticism and extroversion, um, certain personality facets. Some of the medical traits interestingly enough generally did show positive correlations, but they were smaller. And then our biggest negative correlation, so more likely to be with someone who's different from you, was in the sample for chronotype, which was whether you're a morning person or an evening person.

0:26:35 Thomas: That's That's interesting.

0:27:20 Thomas: Wow.

0:27:20 Tanya: And I know you've had some episodes on that. So those are really interesting to listen to.

0:27:22 Thomas: Yeah, yeah. I yeah, there's so much fun stuff there. For me, I think this whole opposites attract, again, I had that same sort of not obviously super scientifically informed opinion, but I was under kind of that same impression that had been basically debunked, like, nah, opposites don't really attract. And it feels like it's the kind of thing where maybe those outliers just catch your attention more you oh it's who's that uh that old the the the raging Cajun that old political operative what the hell is that guy's name I this is an ADHD thing where I know exactly who I'm talking about, but I just cannot think of his name.

0:27:44 Tanya: Yeah. Yeah, it's great. I have ADHD too, so it can relate.

0:27:58 Thomas: There you go. Oh, boy. We're talking about that correlation with. Anyway, this guy whose name will come to me, you know, sometime later on in the episode and I'll blurt it out randomly at a totally inappropriate time. He was famously married to a Republican. I think still is. And so I think it's those examples where you're like, yeah, opposites attract. Tracking. And it's like, yeah, but what about all the other traits and times that that's not at all true? You know, it's one of those only counting your hits hits and not counting your misses.

0:28:00 Tanya: Yeah.

0:28:29 Tanya: Especially for those with the lower correlations, finding a couple, you know, like let's look at height, which has a lower correlation, uh, still in positive the positive direction, but we probably, not that hard to think of a couple, you know, that has a short person and a tall person, right? I

0:28:44 Thomas: Yeah. Yeah. It'd be something you'd notice, you know, it's not something like, like you're not, I'm, I don't notice people's red blood cell counts. I just don't, maybe I should be, maybe I should be paying more attention to that. I don't know, but I, but

0:28:47 Tanya: Yeah. Exactly. Well. Maybe you should be. Maybe we all should be, um,

0:28:57 Thomas: I'm Sure, we all can think, like, our brains can just, and this is more of just a general, like, cognitive kind of, you

0:29:03 Tanya: Like a social psychology thing.

0:29:05 Thomas: Yeah. And it's interesting, as a, as somebody who's kind of in the skeptic community, some of these things are interesting, you know, when it comes to, like, people thinking psychics are real, you're only counting your hits and not your misses and stuff like this, where we could all think of, okay, what's a couple I know that's, one person's really short and one person is really tall, just, you would just remember that, it's something that would be sticky in your mind, whereas you're probably not remembering all, all the other traits that, they're, you, very similar on that, you just don't really think about.

0:29:05 Tanya: Yeah.

0:29:28 Tanya: And I'm trying to remember what that bias is called, I'll probably also pop up to me at a random time in the interview, but, um, there is a name for that.

0:29:34 Thomas: Yeah. Like outlier or something. I don't know. Yeah.

0:29:38 Tanya: Um, even in, like, for height, you know, the correlation isn't very strong, what that tells you oh how

0:29:44 Thomas: James Carville. That's his name. James Carvel. He was married to somebody. Now, just for fun, because it's funny, I'll look up who he's actually married to. It's also somebody who I think is notable. Yeah, what if she, she, oh yeah. Consultant with, that was always the thing. She's a consultant with the Republican party. He's a, he's the raging Cajun, James Carville. So there you go. All right. Now I can put that one to bed. Sorry.

0:30:08 Tanya: I'm so glad that, that you were able to think, I know how satisfying that is when you forget something and then it comes to, um, yeah, anyway, what it means is that it's really just an average. It's just, if I had to guess, if I had a random couple, if I had to guess, are they more similar in height, uh, more different, I would guess they were more similar. However, the weaker the correlation is, the closer to zero, the more that is just the typical level and the more exceptions there are, essentially. So for something like extroversion, which we found a very positive but very low correlation for, that means there are a lot, a lot of couples where there's an extrovert and an introvert together, right? There are a lot. It's just that on average, it is a little more typical for extroverts to be with extroverse and introverts to be with introverts, but it's close to being random.

0:30:12 Thomas: Hmm.

0:31:03 Thomas: So that's that's funny. That's one of the ones that I would have guessed like, oh, maybe that has a chance of being an opposites attract one still. You mentioned there's only like three of those, you know, however many you studied. So there's not not many, but that could have been one because, and that's probably just whatever that bias is, the familiarity bias, whatever it is, the availability heuristic or something, because I'm an introvert and my wife's an extrovert. And only Other reason, it's not just because of that. It's also because of how, like, useful it is. I, I guess if you're both the same, it's fine too. Well, I don't know, if you're both introverts, it can be a challenge. Let me tell you, like, it's very nice having introvert-extrovert matchup because there are things that I like to do and there are things that she likes to do. And we kind of, kind of split those, you know, duties appropriately. I feel like double extroverts, though, they're probably just taking over the world. You know, that, that feels too powerful.

0:31:14 Tanya: Yes.

0:31:25 Tanya: I feel like extroversion, introversion is one of those big ones people are thinking about when they say opposites attract. They're not really thinking about like, no one is like, oh, oh yeah, it's totally totally the norm for a high school dropout and a doctor to marry each other. Like we I think no one was surprised by that, even people who don't study it, right? But I was trying to like trace back the origin of this idea of opposites attract. And I'm sure it goes back further than this, but there was this paper I found from the 50s about the so -called theory of complementary in mate selection. They were looking for people with the same personality traits, but they were also looking at complementary needs. So for example, really liking to be a nurturing person and being a person who likes to be nurtured, right?

0:32:00 Thomas: Yeah. Yeah, right. Yeah mm -hmm. Right.

0:32:48 Tanya: We didn't look at any of those sorts of personality traits. We looked at what are called the big five. They a very small sample, and I don't know how much this one has held up in more recent research, but these were traits that were sort of by their nature meant to be opposite, right?

0:33:05 Thomas: Forget what the big five went to the dark triad or whatever. I was like, oh, is it like psychopaths and murder victims? Do they love being together? Yeah, right.

0:33:06 Tanya: Like dominant submission. That that has been studied. What we looked at the big five are extroversion, agreeableness, openness to experience, conscientiousness, and neuroticism. And I think in our meta -analysis, we found of those the highest for openness to to experience. Huge, but it was bigger. One kind of makes sense, I guess,

0:33:21 Thomas: Right, the covered other studies that, you know, kind of political science studies, or I guess the psychology too, where those highly correlate with political views. So it's interesting that that could kind of go either way. It's like, if you're open to new experiences, uh, I guess, what, you're probably not, not a Trumper. Like, it's just usually how it works. And so it feels like that could be a powerful way that that sorts, you know, or maybe it's, you know, direct, but that feels like one where it could be both direct and indirect in sorting that way.

0:33:44 Tanya: Yeah, that's Really interesting, actually. And I hadn't thought about it that way, because you're right, there have been studies that have pretty, like, repeatedly shown that there are certain traits that correlate with politics like that. I think that there's this other factor, which is that most of us tend to be in a lot of social and group situations with a large variety of extroverts and introverts, right? Like a big mix that meet a lot of people across the spectrum there. Whereas with some other things, like educational attainment, while you're probably more likely to socialize and be around people who have a somewhat similar level of education because of class-related factors and all that, right?

0:34:25 Thomas: I mean, if you're talking about we're socializing, that's got a skew extrovert just because, you're out, you left the house.

0:34:50 Tanya: Yeah. Well, I feel like so many people are like, I'm an extroverted introvert these days. So I

0:34:50 Thomas: Yeah. Yeah, I just, that'S, I don't buy it. Well, I mean everything, everything's a spectrum, But I know I have a joke that I've probably said too many times, but I remember I was in college and I happened to be like it was after a party or something. You know, the ways that this introvert had to get along was, you know, go out and drink a little bit and loosen up. And I ended up, talking you to a couple of girls after. And I remember this girl had just been one. She's one of those. She was delightful. One of those extroverts that had been, you Now, I would be negative to say dominating the conversation. I don't mean in a negative way.

But, you know, it was the whole conversation for a lot of the party in the night and the evening. And I remember I think I mentioned at some point that I was more of an introvert. And I don't remember how because this is 20 years ago or whatever. But I remember her saying, yeah, me too. It's just like, no, I'm sorry. Like she had this thing. She's like, yeah. Because I always have to really recharge after.

0:35:09 Tanya: Yes.

0:35:46 Tanya: Sometimes I like to read a book alone.

0:35:49 Thomas: No, man. Like the whole reason I brought this up was in contrast to what I've witnessed from you all night. You know, like, it's so funny.

0:35:55 Tanya: That Interesting too, because I feel like I hear people identifying as introverts all the time, and I never hear anyone self-identify as an extrovert. It's like not cool anymore.

0:36:02 Thomas: Yeah. Yeah, Yeah. Well, part of the joke is the whole reason you're doing this thing right now is because you're an extrovert trying to relate to me really hard. Like, oh, yeah, I'm an introvert, too. It's like it's that impulse to really be kind of reaching out and finding that commonality with another human more so than, you know, I probably would in a random social situation. I think that very thing is because you're an extrovert trying to, you know, make new friends, which is great. Like, man, if you had a couple that was two extroverts and two morning people, that's actually Barack Obama and Michelle, like you would just take over the world you know the

0:36:15 Tanya: Maybe. Yeah, and like girl boss, uh, like way too, too powerful of a power couple, yeah, the world is built for you

0:36:43 Thomas: They just start radiating when they get together, the, ah, it's too bright, can't see, too powerful, yeah.

0:36:50 Tanya: Interesting thing is, like, about extroversion, it has so many different facets. We always talk about the part that's like, oh, how much do you like being with other people, but it's a lot of other things. It's your level of energy and the level of groups you like. For me, in lot of ways, I can go for weeks and not really want to talk to anyone or answer any of my texts. But if I'm very passionate about something I'm talking about, I have a tendency to dominate the conversation or talk a lot.

0:37:06 Thomas: Oh, that's the ADHD thing, too, though, that's that's my problem. I'm either silent or I can't shut up if the rare times in a social situation I'm talking about something I actually care about. Yeah.

0:37:21 Tanya: Yeah, it's very Very confusing. Introverted ADHD, like, creates a lot of cognitive dissonance, and it's very confusing. But

0:37:35 Thomas: Hey, but, but it's fun when we get on the right track, you know, when we're on something that we're passionate about, it's a lot of fun. It's hard to shut us up. That, so that's the kind of introvert extrovert thing, but the, the bottom line there is more positively correlated though, not like super highly positively correlated. Was that what it was?

0:37:40 Tanya: Yeah, absolutely. Absolutely.

0:37:53 Tanya: Extroversion, yeah. But it did depend on the study. That was just the average.

0:37:54 Thomas: Okay. Right. Um. Wondering with stuff like Like, you know, breastfed as a baby and, you know, red blood cell, all those things that you mentioned earlier, I think that's a really fascinating component. You said, it could be, oh, that correlates with something else that is more selected. I'm wondering, and, and you're able to explain this, the stats part is always where it gets a little tricky, but I do wonder, you know, is that not just a statistical artifact? Because when I see something like that, when I see like, oh, red blood cell count correlates or whatever. It's like, okay, well that can't be, there's no fucking way that's true. You know? So, so is that just a statistical error? But I know you science people are so smart. You've often accounted for that. And so it's like no, it's actually, we've already somehow canceled that out. So was there anything where you're like, well, this must just be a statistical artifact of the sample or it's, like, no, this is a really weird result, but it is probably like because of some other correlation that but then, you this correlation or something.

0:38:29 Tanya: Right. So I think the general trend of the positive correlations, probably were not. However, in our meta -analysis, we did find a lot of variation and point estimates across samples, which means it wasn't exactly the same in every sample. So there are probably artifacts that influence it. When it comes to causality, what we were determining was causality. We were looking at correlations.

0:39:19 Tanya: There are a lot of methodologies that people have worked on, including in my field, to sort of get to get at that causality, like why, why,

0:39:19 Thomas: Right, right, right. Well, let me pause you. Sorry, I don't, yeah, I want to just make sure I'm understanding. I actually wasn't expressly thinking in terms of causality, although do you know something's even a meaningful correlation and not just, oh, the sample we happen to have had, you knows, more people. Because not going to do a sample and have it end up randomly exactly the statistical amount that people are in the world, right?

0:39:33 Tanya: I Yeah, So We did do a multiple testing correction for both analyses, and we used a very conservative one, the Bonferroni correction, probably the most well -known.

0:39:55 Thomas: Love that one. I don't know it, sorry. Have a t-shirt about that correction. It's a good one. No, I'm just kidding.

0:39:56 Tanya: Yes. Well, okay. That's great.

0:40:02 Thomas: Totally joking.

0:40:03 Tanya: Mean, I'm sure there are a lot of t-shirts about the Bonferroni Correction. I totally believe that.

0:40:06 Thomas: If we do start merch here, which I've just been too ADHD to do, I do want to have a lot of those random ass ones. That'd be really funny. Like the Bonferroni correction is my favorite one.

0:40:15 Tanya: Yeah. Yeah. My Here has a mug with a statistics joke on it that I was drinking out of yesterday, yesterday anyway, um, so it

0:40:25 Thomas: Oh, you're not going to tell us? That's such a tease.

0:40:27 Tanya: It was, it was like, "if at first you don't succeed, make three more mistakes so your failure will be statistically significant." Um, yeah, so this idea, which brings us into statistical significance, which I suppose is what we're talking about. Basically, your p-value tells you if a good chance that your finding was an accident, if really zero, but that happened to be an accident that you found this effect. And what we found was that we were still getting, even after after correcting for multiple testing, over 80 % of our traits in both of our analyses were still in the significantly statistically positive direction. And because our samples were so big for most of these traits, some of these correlations were not huge. The threshold just changes depending on things like how big your sample is, right? If you have a sample of of 79,000 pairs,

0:40:42 Thomas: There you go.

0:40:51 Thomas: Right. Right.

0:41:27 Tanya: You might have a correlation that's like 0.05 and, significant.

0:41:32 Thomas: Yeah, I guess thinking about that intuitively, like if your sample was 8 billion, it's like whatever the effect was is real or something. Yeah.

0:41:38 Tanya: Then you have like no sampling error,

0:41:41 Thomas: Yeah. Cool. Do you have any theory about the red blood cell count thing? Like, is there something that correlates with or something that would, that would make that make sense? Yeah.

0:41:41 Tanya: Right? Know a lot about these traits. I'm wondering if it has something to do with where exactly you're living and environment, environment, something like breastfeeding. That one seems a lot more straightforward, that I would guess that has have something to do with class and what class you were raised in.

0:42:05 Thomas: Hmm. Oh, good point.

0:42:08 Tanya: Right. Saw like something like point 30 something correlation, moderate correlation for breastfeeding which was really interesting Because, again, not something you tend to ask on a first date, but, you had similarities in the environments you were raised in, you are a possibly probably more likely to meet each other. And also, to the extent that mutual attraction plays a part in your long-term relationship, you those other similarities are also a factor.

0:42:08 Thomas: Yeah. Yeah, OK.

0:42:43 Thomas: Here's the super interesting one, possibly in my mind about the breastfeeding thing, because I'm pretty certain I'm I'm a dad. I do a parenting podcast as well. And I think generational differences. That's something that's flipped around a few times. So like, I think for a while it was like, oh, breastfeeding. I think there was this movement feeling like, oh no, science is better than that. Formula is better. Cause like, you know, formula has got all this stuff. And so that created a thing where it's like, well, therefore only more well-off people were able to do formula and breastfed less. And then like that's actually flipped at least once, maybe twice, where it's like, oh no, actually now the thing is like, no, breastfeeding is great. And so it was like now maybe that would flip around but because another one of your results that I think I read or you said was age, you know.

0:42:45 Tanya: Yes.

0:43:01 Tanya: Interesting.

0:43:24 Tanya: Yes, that was the highest correlation. Surprise, surprise. Surprise.

0:43:32 Thomas: Yeah, people tend to be very similarly aged, so it's interesting that that probably even if it flipped generation to generation or whatever, 10 years or whatever it would be, that would probably still hold true because of that tightness in age. Is that right?

0:43:45 Tanya: There are so many factors. If we went back even further than like wet nurses and probably, did control for age and like a follow-up thing. We didn't find anything actually too surprising when we did that. But are are so many things that go into that. Again, limitation of this British sample is that they're British compared to the world population.

0:43:47 Thomas: Course. Yeah. Right.

0:44:04 Thomas: Tell me about, oh no, sorry.

0:44:08 Tanya: They're mostly of European ancestry. They tend to be wealthier and have different substance use patterns. So every sample has, you that probably makes, has an effect on these, on correlations, right? So in the British sample, we, were mostly older folks in a pretty narrow generation, uh, who are British, mostly white, somewhat wealthier than average, somewhat different substance use patterns. Whereas in our meta -analysis, there was still a bias for disproportionately US and European countries, but we did have samples ultimately from all six human continents point in the study. And And this is another really interesting thing, right? The culture and marital customs and how those things change,

0:44:16 Thomas: Hey, Yeah!

0:44:56 Tanya: You know? I mean, it's occurred to me that we talk about like mate choice and stuff. Like it's this thing that everyone just chooses.

0:45:04 Thomas: Oh yeah! Arranged marriages are a thing, huh? There's that too, yeah!

0:45:04 Tanya: Um, And like the most populous country in the world right now is India, um, you Know, and marinal customs change, depending on where in India you live, but tend to be very different than what we grew up with. We're But also very underrepresented.

0:45:20 Thomas: Yeah! Yeah. Yeah, but would still show, I'm sure, interesting results. Like if you studied those by themselves, there'd be a lot of different, because, you know, the caste system, there's going to be a huge, you correlations and stuff. Yeah.

0:45:23 Tanya: Yeah, Yeah, exactly. That's another really interesting topic. We some variation in different samples when came to how similar people were for things, and there there were general trends. There was almost always a pretty strong correlation for educational attainment, pretty small correlation for extroversion, et cetera. And these were not, again, perfectly representative samples. Yeah, a really interesting thing because we do have different norms and different kind of variables that stratify us depending on and where and when we grew up and differences in marital customs and all that. And I think that's kind of something, something else that was interesting that we found was there is a lot of diversity in how people were for these traits, samples, examples.

0:46:22 Thomas: Yeah. So somehow we've already gone a lot of time. I want to make sure, are there any kind of highlights that we didn't get to? Are any surprising findings? What do you think? What

0:46:31 Tanya: Guess I want people to remember that that trends we see don't come to be for one, one, uh, really complicated, but, um, still think it's kind of interesting that for almost every trait or the vast majority of traits, people do tend to to be more similar. I that, you this is something that needs to be studied more and across more samples. But I something. I think the coolest part of this is that it's interesting to everyone. And everyone in the behavioral sciences a different reason in for wanting to study it, whether for treatment implications or looking at sociological trends or economics, which was like one of my favorite parts about this study.

0:46:48 Thomas: Yeah, totally.

0:47:16 Thomas: Or it's looking at, you know, dating websites. I can't believe I didn't mention, although I'm sure listeners would have heard me say it before, but actually my wife and I met on OkCupid and that matched us because we had like an insanely high compatibility score.

0:47:29 Tanya: Yeah, your match score.

0:47:30 Thomas: And it's like, yeah, no, that actually does make for good had relationships, not to say that, you know, any, whatever relationship works, works, but like, yeah, we like yeah, I wouldn't really want to date anyone who had these extremely opposite views. So So maybe it's a good thing in some ways, but then you get a lot of these other traits where it's, it's really interesting and it, there's no reason why you would expect it, but you it's still, we just happen to filter into these categories. It seems like.

0:47:38 Tanya: Yeah.

0:47:57 Tanya: And I want to say that most couples are different in some ways, right? And none of this means that you can't be with someone who, especially when it comes to things that aren't about values, although there are exceptions to the values thing as well, it doesn't mean you can't have a successful or happy relationship or with someone who's different, or that you can't be attracted to someone who is different. These are just general trends that we observe for, like, very complicated reasons. Yeah.

0:48:02 Thomas: Sure. Absolutely.

0:48:27 Thomas: And people change too. And that's, this is not obviously a part of your study or anything, but there are so many ways that like, I've actually changed a lot since, meeting my wife. Like I would have probably been probably answered that I was pretty closed off to new experiences back then. And now it's like, oh, no, I'm I'm, you know, yeah. Day to day, I mean, my routine, but I love new experiences now. Like I've really opened up like there's that's kind of interesting, too. Not that it's, again, something you studied, but there's no reason to, like, take this as prescriptive or anything, I think, is the point you're kind of getting at.

0:48:28 Tanya: Mm. -hmm, yeah.

0:48:52 Tanya: Well, and that's the thing that's studied too, too. It's whether become more similar in relationships over time, not for a lot of things, probably, but it looks like there could be some trend there for certain substance use traits.

0:49:10 Thomas: Right. Yeah, that makes sense. Yeah, yeah, that's another thing we would have matched on. Like, yeah, if somebody is a hardcore drinker and somebody is not, it's like that's just going to be a little difficult. And especially when you're young and the ways you socialize would then be different. You know, there's all kinds of things going on.

0:49:27 Thomas: Well, this has been so much fun. Uh, Tanya Horwitz, thanks for coming on the show. Really appreciate it. Oh, do you want to throw out? I mean, do You have anything to plug? I always ask people, some people have a, you know, a social media account or something they want to throw out there.

0:49:11 Tanya: Yeah. Right. Yeah, would like to, you can follow me on Twitter at Gene Wilder type, G-E-N-E W-I-L-D-E-R-T-Y-P-E. It's a pun about genetics and also Gene Wilder. I think it's amusing.

0:49:53 Thomas: I am so glad I asked. Cause I love that. Tell me your profile photo is something that somehow incorporates that. Is there some good mashup?

0:49:56 Tanya: Yeah. Nope, very sadly, there's not.

0:50:07 Thomas: Well, get on that. You're too busy studying stuff, but yeah, you know, you should get on there. That's excellent.

0:50:12 Tanya: Okay, awesome.

0:50:13 Thomas: Gene Wilder type. Uh, thank you so much. And, uh, yeah, I hope we stay in touch, you Know, for future results and stuff.

0:50:20 Thomas: Yeah. Because that the same sex one would be very fascinating as well. So,

0:50:21 Tanya: Do. Yeah, would love to come back. Thank you.

0:50:25 Thomas: Good luck with, that with and we'll keep in touch. Thanks so much for coming on.

0:50:28 Tanya: You so much. Bye.

0:50:47 Thomas: Carville. That's his name. James

0:50:47 Tanya: That tells you oh how