December 13, 2013

Twin studies debunked: Twins are "Similar but not identical"

In the UK leftosphere, The Independent strikes back against The Guardian betraying the sacred verities when the Grauniad gave a good write-up to Robert Plomin's giant twin study of heritability in school test scores. From The Independent:
Similar but not identical: study reveals more about twins than about education 
The headlines this week about a new study of genetics told only part of the story 
STEVE CONNOR  
Genes play a bigger role in educational achievement than teachers, schools or home environment, and the reason we know this – apparently – is because we can compare the performance of thousands of pairs of twins.

At least, this was the main conclusion of a “representative” sample of 11,117 identical and non-identical 16-year-old twins, who were used as the basis for the largest research effort in this country into the role that genes and environment play in a range of traits – from the chances of contracting a lethal disease to aspects of personality. 
By comparing identical twins, who share identical genes, with non-identical twins, who share half their DNA, scientists are able to tease out the differences that result from genetics from those that come from the environment. This, at least, is the idea. But not all experts agree over the importance of twin studies, and indeed some molecular geneticists are extremely hostile to them. Marcus Pembrey, emeritus professor of paediatric genetics at University College London, for instance, believes they are next to useless when it comes to telling us anything significant about the role of genes. 
“In all the years of twin studies I can only think of two occasions when they have produced a meaningful result and I’d be nervous about saying that monozygotic [identical] twins are truly representative of the population. I abandoned my twin studies in 1972,” he said. 
Monozygotic twins are one of nature’s idiosyncrasies. A few days after an egg is fertilised by a sperm the developing embryo splits in two, each sharing the same set of parental genes. 
Non-identical or dizygotic twins occur when two eggs are fertilised by two sperm and the resulting pair of embryos develops within the same womb, sharing the foetal environment but only 50 per cent of their parental genes, just like ordinary siblings. 
... Although identical twins share the same DNA they are frequently different in many ways. One twin can be larger than the other from birth, indicating an unequal environment in the womb, and it is now firmly established that identical twins can be born with very different health prospects. 
John Burn, professor of clinical genetics at Newcastle University, told a London conference on twins this month about the case of a pregnant woman who was an alcoholic. She gave birth to identical twins. One had foetal-alcohol syndrome while the other did not. So even though they shared the same genes and the same foetal environment, the twins were different. “We can’t explain it,” Professor Burn said. 
There is also the case of enantiomorphic or “mirror-image” identical twins. Although physically similar, they show certain features that are mirrors of one another – their hair parts on opposite sides of the head or they suck different thumbs when babies, for example. 
At the same London conference, organised by the charity Progress Educational Trust, a member of the audience said she and her identical twin sister needed to sit or stand on a particular side of the other in order to feel comfortable. This had been the case for as long as she could remember. “Is this because our mother always put us in our cot on the same side?” she asked. 
Professor Burn was unsure, but suggested it might be because she and her sister were enantiomorphic. 
It is increasingly clear that identical twins are not in fact identical. This is even more so when epigenetic factors are considered. These control the way genes are expressed and even though the DNA sequences are the same, the way their genes work are almost certainly different. 
And yet, the principal assumption behind twin studies is that identical twins share the same genes and, largely, the same environment. This is crucial to working out heritability, which is a measure of how much variation in a particular trait is down to genes.

Interesting points, but, correct me if I'm wrong, isn't the logic of this argument against twin studies 180 degrees backward? Don't twin studies work by positing that the difference in degree of phenotypical variation between identical twins and fraternal twins is due to genetics, so by emphasizing all the random non-genetic differences between how similar identical twins turn out, you're saying that twin studies are actually less sensitive to finding the full magnitude of heritability?

I may have this backwards, so don't take my word for it.
Heritability is at the centre of the TwinsUK study run by King’s College London, which for 18 years has built up and followed a cohort of identical and non-identical twins. The latest effort on the GCSE performance of 16-year-olds found, for example, that the heritability of compulsory core subjects was 58 per cent, of English 52 per cent, of mathematics 55 per cent and of science 58 per cent. 
This is why the scientists concluded that genes played a bigger role in a child’s GCSE performance than any other environmental factor. “We suggest a model of education that recognises the important role of genetics,” the researchers said in their study, published in the journal Plos One.
A serious problem with heritability as a metric for measuring anything, however, is that it varies depending on what you are measuring, on which population it is based, and on the time of assessment. Significantly, few politicians seem to understand this limitation. 
The heritability of general intelligence, for instance, rises with age. In infancy, about 20 per cent of a child’s intelligence is attributed to genes, whereas in adults it can be as high as 70 or even 80 per cent, according to Robert Plomin, professor of behavioural genetics at King’s College, who led the twins study into educational achievement.

As a commenter pointed out this, new study should be considered a triumph for nurturists, since it found Shared Environment accounted for 36% of variations in these high-stakes tests at age 16. That's unusually high for a twin study, which typically find that Unshared Environment accounts for most of the non-heritable percentage.

32 comments:

Anonymous said...

No, your're right. Those criticisms against twin studies would only weaken the genetic explanation of the variance and not strengthen it. would only increase the random variation. They are unknowingly arguing for the strength of genetic associations! Interesting of the shared environmental variance number. Most studies regarding cognitive capabilities has larger unshared environment (probably random noise) component.

Alcalde Jaime Miguel Curleo said...

"A serious problem with heritability as a metric for measuring anything, however, is that it varies depending on what you are measuring"

Shhhh, everybody -- social scientists at work!

2Degrees said...

One thing I have noticed about liberals is how rapidly their value system can change. Once upon a time, not so very long ago, Mother Earth was threatened by overpopulation. Now it is a hate fact.

At the moment, suggesting that genes and IQ are related is the ultimate hatefact.

However, my years in academia taught me two things about these people.

1). They are utterly convinced of their own moral and intellectual superiority. Stupid is a term they use to describe anyone who disagrees with them, but then deny that their any such thing as an innately stupid person.

2). They are increasingly nepotistic and inbred. Recently Obama, Cameron and the Danish PM took a selfie at Mandela's funeral. She is married to the son of the former head of the British Labour Party, Neil Kinnock. For years her held a very well-paid sinecure in Davos. If I could be bothered I have no doubt I could track down many similar connections.

Suddenly we find the Guardian writing about IQ and the Mayor of London making speeches about the importance of IQ in schools.

Is this new superclass about to tell itself that it has the right to control all the goodies because it is genetically superior?

Stranger things have happened.

neil craig said...

The criticisms in the newspaper article is that identical twins aren't wholly identical. Logically that means that twin studies will underestimate the importance of genetic inheritance not overestimate as opponents claim to believe. It is not an argument against heritability let alone against doing twin studies merely one for adding a few points to the effect of heritability.

ring ring ring said...

Is this new superclass about to tell itself that it has the right to control all the goodies because it is genetically superior

Two Harvard guys already wrote a book with a conclusion positing that outcome in 1994. It wasn't favorably received at the time.

jody said...

i'm starting to think this stuff is random. that is, i'm coming close to rejecting the environment hypothesis.

tons of these twins grow up in almost exactly the same environment. how do we account for such a large variation in performance among some sets of twins, when they have exactly the same genes, and a very similar environment from birth to age 18?

it must be gene expression. gene expression must be the factor here which best explains the variability between twins raised in very similar environments. both players start the game with the same deck of cards, but the cards are issued slightly differently as the game progresses.

with respect to the brain in particular, this is something i studied during my undergrad - that the genome does not have enough space or information to contain a map of the exact position and connections of each neuron in the brain, and must instead build it from the ground up by having the neuron growth cones expand outward. this has to occur in a very, very similar way in every person - but it's still unique to each. in the same way fingerprints are unique to each. every carpenter starts with the same blueprints, but the houses are slightly different when finished.

the liberal take on the issue is simply incorrect - environment has been controlled for here in many, many of the cases. it cannot explain the variability in performance by adulthood. that some twins show differences in math ability by age 18 in no way opens the door, for instance, to the prospect that any random person can win a fields medal if we just kidnap them at age 4 from their parents and force them into math camp for 20 years.

instead what it predicts is my clone scenario - make 100 clones, and 5 of them will be clearly smarter and stronger than the others, while 10 of them will be clearly weaker and dimmer. gene expression which nobody can control, will become the biggest factor in a scenario where you control the genes and environment.

Anonymous said...

re: overpopulation ideology

In P.J. O'Rourke's "All The Trouble In The World" the population-density chapter was subtitled: "Just Enough of Me; Way Too Much of You"-- Through the years I've found this a useful heuristic.

jody said...

would be interested to know how twinning affects sexuality. could give important insight into the hormone hypothesis for homosexuality.

Allan Folz said...

Interesting points, but, correct me if I'm wrong...

Steve, you're always so nice. :) What this article deserves is a good Cochrane fisking. Then again, why waste a good fisking.

candid_observer said...

I find the following very useful comment on the Guardian site, which places the estimate of the effect of shared environment in proper perspective:

...there is another reason to believe that the GCSE study referred to in the article exaggerates environmental influences. Namely, the study assumes that the parents of the children studied mated randomly, with no regard to the educational level, intelligence, etc. of each other. This is, of course, a highly unrealistic assumption — assortative mating is the rule in all societies — but twin researchers rely on it because they typically don’t have the data needed to model assortative mating. The consequence of this for the twin method is that to the extent that assortative mating causes twins to be phenotypically more similar, it is treated as a purely environmental effect, even though parental similarity in traits like intelligence is strongly influenced by genetics. Thus, the importance of genetic influences in underestimated.

The GCSE study found that about 58 percent of exam results were explained by genetic differences, while about 36 percent were due to those environmental influences that are shared between siblings (e.g., family and school effects). For the reasons discussed above, the former figure is probably an underestimate, while the latter figure is probably an overestimate.

Anonymous said...

From the article:

A serious problem with heritability as a metric for measuring anything, however, is that it varies depending on what you are measuring, on which population it is based, and on the time of assessment.

It's funny how the critics seem to think that this criticism only concerns behavior genetics, when it of course applies with equal force to all social and behavioral science. The very same people who think that a study of 11,000 twins carefully selected to be nationally representative cannot be used to draw any general conclusions also think that small experiments involving a few dozen elite university students convincingly explains all racial differences in test scores across society.

Assistant Village Idiot said...

So...because identical twins aren't really identical, only 99% or so, any similarity between them is more likely to stem from whether their community has a Head Start program?

candid_observer said...

It is really amusing how these people at the Guardian were so discombulated by the results of the study, and so determined to refute it, that they couldn't even see that they were working against their own goal.

Maybe their sports idol is Roy "Wrong Way" Riegels.

ben tillman said...

At least, this was the main conclusion of a “representative” sample of 11,117 identical and non-identical 16-year-old twins....

An odd number? And it doesn't occur to Mr. Connor that this figure has to be wrong? Wow.

ben tillman said...

By comparing identical twins, who share identical genes, with non-identical twins, who share half their DNA....

Um, more like 99+%. They share all the DNA that their parents share plus, on average, 50% of the rest of their DNA.

Connor's (and his editor's?) intellectually sloppiness is absolutely unacceptable.

long live reggie w said...

Steve, twin studies are the province of losers. Here is an analogy - some old former pro golfer (I forget his name) is spending quality time with old friends who are unimpressed, as old friends should be, by minor details of a life well lived such as whether somebody had been a pro golfer or not. Anyway, at a random point in the weekend, two or three of these friends find themselves at a driving range where, by some kind of dream logic, only broken down old five six and seven irons are available. They all whack a few for the pure joy of it and are all more or less as successful as would be expected but off the iron of the old pro the ball flies as if it had been hit on a sunny morning in Denver with the wind at his back instead of near the Jersey shore on a dull muggy evening. Twin studies are not the province of old pros, they are the dull muggy evening of people who do not understand other people very well and who need lists of numbers to convince themselves of what they ought to have known since childhood about the various gifts the Lord has given.

Dennis Dale said...

"One twin can be larger than the other from birth, indicating an unequal environment in the womb, and it is now firmly established that identical twins can be born with very different health prospects"

Could it be? Have we reached the ne plus ultra of behaviorist denial? Surely there's no point beyond the belief even the womb is an un-level playing field where resources are hogged and the weak exploited.

Anonymous said...

long live reggie w said...

Your analogy made about as much sense as the fact that coconuts cause more deaths than shark attacks implies red squirrels are very annoying.

long live reggie w said...

as Shakespeare said if you have to ask what rhythym is you ought to stick to painting. I am sorry that my disagreement with you on the value of twin studies to the non-neurologically-diverse was muggily opaque.

Janni said...

So they're saying "Aha! These studies are useless because identical twins are NOT even identical in the first place!" But they don't then wonder why the identical twins raised apart score so similarly?

So, if identical twins are not exactly equal but when raised apart they still score so similarly, the heritability estimates are actually underestimating the effect of genes? Do I have this right?

Anonymous said...

I see the Indy isnt allowing comments on that article. If only, if only...it would be carnage! Lol!

The Guardian did allow comments but of course comment moderation there is almost air-tight. Only the barest whisper of dissent is tolerated. The liberal livestock must remain unworried at all costs.

Anonymous said...

GCSEs include significant amounts of coursework, though maths was discontinued a few years earlier. It'd also be interesting to see how g loaded these GCSEs are considering that they have often been criticised as getting easier over time and the questions are derided as being far too easy.

Anonymous said...


From Connor's piece,

"The heritability of general intelligence, for instance, rises with age. In infancy, about 20 per cent of a child’s intelligence is attributed to genes, whereas in adults it can be as high as 70 or even 80 per cent, according to Robert Plomin, professor of behavioural genetics at King’s College, who led the twins study into educational achievement."

This of course means that those extra efforts to enhance a child's IQ might look promising for a while but then as young adults they fall back to the level of their relatives. Money well spent?

Prodawn said...

"Don't twin studies work by positing that the difference in degree of phenotypical variation between identical twins and fraternal twins is due to genetics..."

I think that's the opposite of it. The differences are assumed to be environment; the similarities are assumed to be genetics.

Anonymous said...

Insofar as GSCEs measure achievement and not simply 'g', the scores are more malleable than 'g' is. Nevertheless, they are reasonably g-loaded. Their malleability is reflected in the shared-environment contribution to the scores.

Measures of 'g' in adulthood show higher heritability than this study found.

Dartangnan19 said...

This might shed some light on this:

http://www.psychologytoday.com/articles/201109/the-incredible-expanding-adventures-the-x-chromosome

Anonymous said...

Having taken GCSEs, I can confirm they're very easy exams. The top 2/3rds of the bellcurve can easily get the required grades, and gaming the system by teaching/studying to the test, resitting exams, and so on, is trivial.

I would be very interested in seeing the A-Level results of this twin cohort - those exams are much meatier.

RS said...

For all that technical yeahyeah, with all the epigenetics said and done, when have you met identical twins (kids) that you can actually tell apart? I couldn't, with kids I knew as family friends.

I had an adult friend who was MZ but I never met her twin.

I think there were some in my HS, but if so I had no idea which was which, not being friends of theirs.

Anonymous said...

"indicating an unequal environment in the womb" - 8 months and 29 days alert.

Justin Irving said...

Steve, this is exactly the defense that Bryan Caplan has put up of twin studies. If identical twins aren't perfectly identical, then the estimator using fraternal twins as a control underestimates the effect of genes.

Bill said...


jody said...

would be interested to know how twinning affects sexuality. could give important insight into the hormone hypothesis for homosexuality.


You might like this and this. When one identical twin is gay, then about half the time the other twin is gay, though some studies find higher or lower values.

Anonymous said...

"The heritability of general intelligence, for instance, rises with age. In infancy, about 20 per cent of a child’s intelligence is attributed to genes, whereas in adults it can be as high as 70 or even 80 per cent"

I'd be incredibly!!!! interested in a comparison of IQ test scores at the same ages between parents and children.

Would the parents of a child whose infant IQ results are much lower or much higher than their adult IQ results also have scored much lower or much higher (respectively) as an infant than they do as adults?