The taint of the Minnesota twin studies lingers on

Except when it doesn’t bother with the DZ part.

For me, it all began with the Mike Douglas show in the 1970s. If you’re not old enough to remember, that was an afternoon talk show, pre-Oprah, pre-Ellen, etc., sort of a primordial congenial-host-with-a-panel-of-guests kind of thing, and my mother often watched it, just as I was coming home from school. That’s where I learned about the Minnesota twin studies, as a nerdy teen in middle school/high school. I found that kind of stuff fascinating, anything about science would draw me to the TV. Or science-fiction/horror movies, although my mother tended not to have those on.

Douglas is the one on the left.

So I’m watching this, expecting to learn a little biology, but the Minnesota people (it might even have been Bouchard, I don’t recall) were doing a dog-and-pony show, very light on the science and rich in pandering to people’s biases about human nature, and my developing skeptical antennae were twitching. I remember them talking about how one pair of twins separated at birth had both grown up to be firemen. My future job was written in my genes, really? Another pair were married to women with the same first name. Now just hold on there, you’re telling me that somewhere in my genome was a hard-coded response to potential mates based on the sound of their name? Worse, another pair had given their dogs the same name (not the same name as their wives, the two dogs had the same name). There was no evidence that maybe there was a single locus for wife’s name/dog’s name.

I was taken aback. This sounded like complete bullshit. Are you telling me I can’t trust Mike Douglas?

I started researching the topic, back in the days when there was no google, and you had to physically go down to the library and read books to figure out what was going on. I quickly found lots of material, besides the sheer unbelievability of the nonsense they were spewing on daytime TV, that questioned the whole idea. It wasn’t hard. Now that I do have Google, here’s an an article that points out the methodological problems of twin studies.

  • Twins aren’t actually separated at birth. In these studies, 33% were separated after a year or more spent growing up together;
  • 75% of the pairs of twins still had contact with each other while growing up;
  • More than half (56%) were raised by a close family member;
  • In 23% of cases, the twins ended up being raised together again at some point or lived next door to each other.

Besides the fact that they were obviously cherry-picking coincidences for talk-show audiences, their premises were deeply flawed. These were bad studies with exaggerated conclusions drawn from flimsy data.

That link will also take you to this paper, A Reevaluation of the 1990 “Minnesota Study of Twins Reared Apart” IQ Study. It’s a damning analysis of one crucial result from the twin studies.

In 1990, Thomas J. Bouchard Jr. and colleagues published the widely cited 1990 “Minnesota Study of Twins Reared Apart” (MISTRA) Science IQ study. To arrive at the conclusion that “IQ is strongly affected by genetic factors,” Bouchard and colleagues omitted their control group reared-apart dizygotic twin (DZA) IQ score correlations. Near-full-sample correlations published after the study’s 2000 end point show that the reared-apart monozygotic twin (MZA) and DZA group IQ correlations did not differ at a statistically significant level, suggesting that the study failed the first step in determining that IQ scores are influenced by heredity. After bypassing the model-fitting technique they used in most non-IQ MISTRA studies, the researchers assumed that the MZA group IQ score correlation alone “directly estimates heritability.” This method was based on unsupported assumptions by the researchers, and they largely overlooked the confounding influence of cohort effects. Bouchard and colleagues then decided to count most environmental influences they did recognize as genetic influences. I conclude that the MISTRA IQ study failed to discover genetic influences on IQ scores and cognitive ability across the studied population, and that the study should be evaluated in the context of psychology’s replication problem.

Whoa. So the control group for their study of IQ in separated monozygotic twins, who are mostly genetically identical, was supposed to be measurements of IQ in dizygotic twins, who share on average 50% of their genes. That’s basically an essential comparison, as far as I’m concerned.

The Bouchard paper didn’t bother to do that comparison!

My mind is blown. They had the data, they must have done the statistical analysis, but they didn’t publish it in this well-known paper. They instead just compared identical twins to each other without bothering to look at how similar or different ordinary brothers and sisters were to each other, and declared that the numbers they got were sufficient to declare IQ to be significantly heritable!

The paper above dived in and re-did the analysis with the MZ twins and the DZ twins, and found that IQ was no less heritable in DZ twins than MZ twins. This does not fit what was expected if IQ was determined by genetics alone. Environment must play a significant role.

Never fear, hereditarians! Bouchard et al. had a prepared excuse to cover that eventuality. You see, environment is genetic!

We have seen that after bypassing their model-fitting procedures and their DZA control group data, Bouchard and colleagues based their conclusions about IQ heritability on the claim that the MZA correlation alone “directly estimates heritability.” However, they reached their conclusions only because they decided to count most environmental influences as genetic influences.

Bouchard and colleagues wrote in their 1990 Science article that one of the three “implications” of their genetic “findings,” and of behavioral genetic findings in general, was that MZA behavioral resemblance caused by the impact of environmental influences “is counted as a genetic influence,” because MZA pairs’ “identical genomes” cause them to create more similar environments for themselves (pp. 227–228). They continued,

MZA twins are so similar in psychological traits because their identical genomes make it probable that their effective environments are similar.… It is a plausible conjecture that a key mechanism by which the genes affect the mind is indirect, and that genetic differences have an important role in determining the effective psychological environment of the developing child. (Bouchard et al., 1990a, pp. 227–228)

The above statement is not an “implication” of the researchers’ findings; rather, it is an assumption upon which they based their findings.

Wow. That is remarkable. So everything is genetic! Therefore they didn’t have to worry about other variables or the confounding effect of differences in environment (or in the case of their flawed, not-actually-raised-apart twin studies, similarities in the environment) because, no matter what result they got, it was genetically determined.

I said at the outset that I was having problems accepting these twin studies when I was 15 years old. I was not a super genius 15 year old. As a teenager, I hadn’t even taken any classes in basic genetics, and even if I had, my experience of middle school science teaching tells me it would have been mediocre. I don’t think the published work of intelligent, established scientists should be rejected because it makes a teenager queasy.

But, man, the stuff they were trotting out on the talk show circuit was embarrassing. Somebody among them must have had similar reservations, and you’d think they would have tried to inject some meat into those public events. Why didn’t they? What was going on at the University of Minnesota?

It seems that Bouchard had some deeply seated bigotries, and he got the answers he wanted.

In a pre-MISTRA 1976 chapter entitled “Genetic Factors in Intelligence,” Bouchard argued in IQ-hereditarian fashion that “human intelligence,” as supposedly measured by IQ tests, “is largely under genetic control,” that social “class differences in intelligence have an appreciable genetic component,” and that due to reproduction patterns, the possibility of a decline in national intelligence “should be subject to continual scrutiny” (Bouchard, 1976, p. 193). Two decades later, Bouchard (1995, p. 417) and Lykken (1995, pp. 216–217) endorsed Richard Herrnstein and Charles Murray’s conclusion in The Bell Curve that “both genes and the environment have something to do with racial differences” in IQ scores (Herrnstein & Murray, 1994, p. 311).

Holy crap! I don’t think Bouchard said those things to Mike Douglas.

It’s clear that we’re still seeing the legacy of Francis Galton and eugenics, that too many scientists have a bias favoring genetic determinism. We’ve been struggling to eradicate the weedy toxins of eugenics, but we have to dig deeper and clear out the ongoing poison of these badly done twin studies. We’re probably going to have to bear that burden for at least another century.

Then we have to start on all the badly done GWAS studies…


  1. says

    My mother also watched Mike Douglas. I remember him as a mediocre band singer who occasionally had decent guests but was himself bland and “safe” for afternoon audiences.
    My clearest memory was the time a guest explained to Mike where his center of gravity was–six inches below his navel. Mike seemed horrified to learn that–asking the guest “can my center of gravity really be in that unflattering place?”
    His exact words, indelibly etched into my memory. I always referred to him as Mike Dickless after that, and didn’t look to his show for any science education. Fortunately, the rest of the afternoon’s programming featured Gilligan’s Island, The Flintstones, and Star Trek, so I was in good hands.

  2. robro says

    We’re probably going to have to bear that burden for at least another century.

    Do you mean just in the scientific community? Because I suspect we’ll have to bear the brunt of this kind of misinformation in the larger society forever. Hell, there are still people promoting eugenics. If some people want to believe or promote the belief that other people are inferior, they’ll use any argument they can muster including flawed scientific studies.

  3. Ed Peters says

    That paper sounds like scientific malfeasance in support of racism. Was it peer reviewed by non-bigots? Was it retracted?

  4. says

    Not the point of the OP, but the Mike Douglas Show was where I first saw many good bands play on TV, including Frank Zappa, Genesis (unfortunately, right after PG left), and Renaissance. The producers must have known that they had a large number of high school students in the audience, given their time slot.

  5. anthrosciguy says

    Mike Douglas (to add to the derail) had lots of different kinds of musical guests because he, like Dinah Shore, liked all kinds of musicians. This was unusual for talk show hosts in those days. He also had guests hosts who would host for a week alongside him, both talking and then joining in asking questions of other guests. This led to more interesting questions at times. My favorite guest hosts of his were the comedian Davis Steinberg, Little Richard, and John and Yoko.

    As for the Twins study, they basically ran it out as far as they could without putting things to paper so they could offer these news-attracting cherry-picked tidbits. The eventual 1990 paper, and the fact they buried their control group, shows why they did that. The paper only gets in the way of the publicity.

  6. says

    Those separated twins who married women with the same first name probably did so because that name was popular at one point. In 2020 Olivia was the most popular girl’s name for girls born in Alberta, so I would be entirely unsurprised if separated twins both married women named Olivia a couple of decades from now.

  7. ealloc says

    I’ve been meaning to watch the 2018 doc, “Three Identical Strangers”, describing a whole other level of scientific malpractice. The scientists in that case intentionally split up twin/triplet babies for adoption, and assigned them to families of different economic levels to see if the wealthy children would turn out differently from the poor children.

    Thanks for finding this important update to the Minnesota study – more fuel for my lunch conversations which coincidentally touched twin studies a few days ago!

  8. kome says

    It’s a shame that white supremacist eugenicists like Terman and Burt took Binet’s concept of IQ (which was essentially “academic achievement in children”) and turned it into what it is. Especially damning is the idea itself of IQ/intelligence being genetically determined rests on nearly 30 years of completely faked data by Burt that is only kinda sorta acknowledged by genetic determinists today, who are trying their god-damnedest to preserve the biological essentialism despite such a profoundly messed up history. People like Robert Plomin and Kathryn Harden and so on, who are responsible for most of the GWAS bullshit you mentioned, have done so much damage by preserving the eugenicists’ ideas of intelligence. I’m starting to wonder if, following their own deaths (or, who knows, maybe after someone hacks their emails or something), we’ll discover, much like we did with E.O. Wilson following the release of his private correspondences, that they’re secretly very buddy buddy with virulent white supremacists while trying to publicly sanitize the pseudoscience they push.

  9. says

    Most of psychology, being blended with made up pseudoscience, racism, and proof by vigorous assertion, is crap. They are trying to get it onto a scientific basis but it’s still dodgy and it will take decades. Unfortunately, psychologists have chosen not to repudiate the crap – they continue to propagate it because it makes them sound knowledgeable. Obviously, Jordan Peterson is a good example of this. Any scientist who examines IQ testing will realize that it is mostly entirely garbage pseudo-science* yet psychologists talk about it anyway. They talk about Jungian theories, and they talk about studies that have repeatedly failed to replicate. They even talk about Zimbardo’s horribly unethical and pointless experiments (he was trying to prove there was a genetic authoritarian component in Germans!) the whole field needs to be wiped clean and restructured based on evidence-based practices. But they won’t do that because there wouldn’t be a whole lot to talk about, if all the bullshit was gone.

    (IQ tests are useful for measuring how an individual is performing compared to a baseline of themself, to check for dementia, etc.)

  10. Pierce R. Butler says

    … monozygotic twins, who are mostly genetically identical…

    “Mostly”??!? What happens during the zygote-splitting process that changes DNA?

  11. says

    There are all kinds of little mutations that occur during somatic cell division. We also don’t know about variation in mitochondria.

  12. whheydt says

    Re: Marcus Ranum @ #9…
    So far as I can tell, the thing IQ tests measures is…the ability to take IQ tests.

  13. chrislawson says

    Monozygotic twins are not 100% identical.

    From Resnik 2019: “Monozygotic twins are generally assumed to be genetically identical. However, it is now well established that phenotypic and genotypic differences between monozygotic twins are common. The original zygotic genome may be modified via a number of mechanisms including unequal blastomere allocation and postzygotic genetic, epigenetic, or environmental events. Missegregation of genetic material in the course of the monozygotic twinning process may result in two distinct cell populations secondary to discordant postzygotic nondisjunction or crossing over, imprinting differences, activation or expression of selected genes, X-inactivation, and differences in telomere size. Depending on the timing of the genetic event relative to zygotic cleavage, the genetic difference may be seen in multiple somatic tissues or may be mosaic.”

  14. KG says

    Marcus Ranum@9,
    Ah, now it’s just “most” of psychology rather than all of it, as you’ve previously claimed. Any estimate of the percentage? In fact, “psychology” is such a disparate field, there’s really no way of saying anything sensible about its quality as a whole (like sociology, anthropology, economics, history, archeology, even medicine… – basically any large area of what you might call human studies).

    Pierce R. Butler@10,
    Mutations in the DNA subsequent to the split. If early enough, this could affect large numbers of cells in the adult and the phenotypic differences could presumably be substantial.

  15. KG says


    Not true. When used appropriately (as Binet intended), they can (sometimes) usefully identify areas in which people, especially but not only children, may have either cognitive problems, or unexpected talents.

  16. Pierce R. Butler says

    Thanks to PZ Myers @ # 11, chrislawson @ # 13, & KG @ # 14, I now see I had another thing that I thought I understood wrong.

    Why does learning more produce greater awareness of ignorance? (Uh, that’s rhetorical – please!)

  17. bcw bcw says

    As a twin, I was fascinated by those claims and had some motivated reasons to believe them. I do remember wondering how they calculated the odds that two people in a group that size would have the same dogs’ or wives’ names to show that the result wasn’t chance. [Of course, you also have to control for after-the-fact question selection data bias.] Being the son of an honest scientist, it didn’t occur to me then that they didn’t care whether what they said was true. As I became a scientist myself, it became really obvious that no paper should be read as inherently true. In addition to the deliberate lying, there is just so much “I need this result” self-deception and sloppiness.

  18. PaulBC says

    Wait, Mike Douglas and Gene Simmons are twins? Environment really does make a big difference! (Sorry my attention span is even lower than usual right now.)

  19. chrislawson says


    The other violation of Binet’s work is that he was trying to identify children who would benefit from extra educational support, not find ways to blame differences on innate traits which, being innate, are (wrongly) perceived as being resistant to environmental improvement, and especially not as happens currently in Australia, use a performance-based funding system that actively penalises schools that do poorly on standardised testing.

    The really pathetic thing about all this? Binet, the very first person to develop the concept of IQ, despite his outdated methods and social milieu, had a vastly greater understanding of intelligence than the fools who write regressive crap like The Bell Curve or the Minnesota Twin Study. If this was physics, it would be as if a large subset of currently practising scientists still insist on using a model that is more rudimentary and predictively far worse than Rutherford’s 1911 model (and that’s leaving aside the fact that Rutherford’s model gives no support to race, gender, or poverty determinism).

  20. says

    It’s like arguing if you’re 98% asshole or just 79%. You’re an asshole and the rest is details.

    With regard to what parts of psychology are bullshit, do your own thinking.

  21. jenorafeuer says

    Based on the ‘Reevaluation’ paper’s comments on the MZA and DZA groups, it looks to me like your comment on ‘The Bouchard paper didn’t bother to do that comparison!’ is perhaps being overly charitable. Given that they already had the control group defined, and they had all the data, it would seem a more likely prospect is that Bouchard et al did do that comparison, realized that it disproved their point, and so they deliberately didn’t put that comparison into the paper and hoped no-one would notice.

    Pierce R. Butler@10:
    And the ‘mostly’ can have significant effects; not to mention that on top of actual genetic differences there are often epigenetic differences, with genes getting turned on or off at different times. There was at least one case I’ve heard of where there was a pair of identical twins of whom one had his family’s hereditary case of dwarfism activate and the other didn’t, resulting in a very significant difference in height between the two.

  22. Reginald Selkirk says

    @6: I remember the year early in this century when the graduate biology program at my university had an incoming class with about 9 Jennifers. I suppose this had something to do with a celebrity who was popular a couple decades earlier.
    Top Five Female Names for Births in 1982

  23. PaulBC says

    chrislawson@13 Based on X inactivation alone, I would think monozygotic twin girls are significantly different. But I have no idea how often this shows up in an obvious way in phenotype. Does anyone have a good reference?

  24. says

    Bouchard and colleagues wrote in their 1990 Science article that one of the three “implications” of their genetic “findings,” and of behavioral genetic findings in general, was that MZA behavioral resemblance caused by the impact of environmental influences “is counted as a genetic influence,” because MZA pairs’ “identical genomes” cause them to create more similar environments for themselves…

    Yeah…right…because babies and little children have that much control over their environments, wherever they happen to be. The grownups they depend on? Guess that dependence is overrated. Put one twin in a bomb shelter in Kyiv during a Russian invasion, and he’ll make his environment just as cozy as the other twin growing up in a Paris suburb, right? Or will the Paris twin turn HIS family’s home into a terrifying hellhole?

    I can’t believe any paper containing such an assertion even made it TO peer review, let alone through it.

  25. cartomancer says

    As a monozygotic identical twin myself, I have to say I’ve never been all that interested in these kinds of “separated at birth” studies / conceits. What interests me is precisely the opposite – the ways in which growing up with a twin brother (or sister) can shape and affect one’s personality and outlook on life. That complicated interaction between seeking similarity and seeking difference, the sense of one’s identity as bound up with someone else’s, the social consequences of viewing being on one’s own not as a baseline of zero but an inadequacy of minus one. That sort of thing.

  26. Erp says

    And “Mary” led from the start of the Social Security data until 1961 (except for a handful of years around 1950 when it was number 2 (“Linda” was number 1). Given the time of the study all the wives would have been born in that time frame. Also note the more common names were given to a lot more children. For instance in 1950. about 4.6% of newborn girls were called “Linda” and another 3.7% were called “Mary”. The top 16 girl names each had at least 1% of the newborn girls. In 2020, the most popular girl’s name was “Olivia” and only 1% of babies were given that; no other name broke 1%. Boys’ names btw were even less diverse; the top 22 names each had at least 1% in 1950 (2020 only one name, “Liam”, had more than 1%).

  27. chrislawson says

    Reginald Selkirk@22–

    As any evopsych proponent will tell you, that is clear evidence that Jennifer was the most favorable genotype for your generation, only to be outcompeted by the Emily allele from 1996-2007.

  28. chrislawson says


    Just doing superficial lit searches on X inactivation in twins, there’s this and this and this Haven’t read them enough to comment on their quality.

    I also found this but can definitely recommend against it as it’s yet another idiotic attempt to find genetic causes for complex behavioural traits and its very first sentence is so egregiously wrong that it should have been taken out behind the barn and shot. (That sentence: “Although members of monozygotic twin pairs are identical in genome sequence, they may differ in patterns of gene expression.” When this paper was written in 2004, the first clause of that sentence was already known to be comprehensively wrong. But why would evopsych fools bother to learn any actual genetics? It only gets in the way of spinning their hypotheses. There’s enough crap just in the abstract to warrant a 10,000 word rant, so I’m not going to start. For those of you with some genetic background, just feast your eyes on this: “This effect would be seen because male siblings, always receiving their only X chromosome from their mother, have a 50% chance of receiving the mother’s paternally-received X and a 50% chance of receiving her maternally-received X” which of course could only be true if X chromosomes did not undergo recombination in XX females. This statement reveals a fundamental confusion about the difference between genes/linkages vs entire chromosomes. It should get anyone a fail in an undergrad genetics exam question, and yet here it is in a peer-reviewed paper. What a ghastly bunch of peers they must have been.)

  29. pacal says

    After finding this posting I did a bit of research on the Twin Studies; more specifically the Minnesota one. Well I had never done any detailed research on it. I thought it was overall suggestive and partly confirming the moderate to high heritability of certain traits, like personality and intelligence, although, in my opinion, providing little to no support for genetic group differnces in such traits.

    I was not aware of just how BAD such research was in fact. Also I was completely unaware of the, so-far, near utter failure to find any actual genes for such characteristics in humans and the ad-hoc explanations by those supporting the idea of moderate to high heritability of such traits, to rescue the idea. The rather annoying, virtualy complete refusal, to consider the possibility that these traits may not be moderately to highly heritable to begin with by these people was problematic to say the least.

    I was also not even slightly impressed by Bouchard’s refusal to show the raw data to critical scholars, (Like Kamin.). A red flag in my opinion.

    Right now this is my opinion. At best studies like the Minnesota Twin study are mildly suggestive as the very best gloss of these studies. Right now I can’t take these studies very seriously after looking at the literature. In fact there is, in my opinion, the strong possibility that these studies can be characterized has borderline fraudulent.