RSS
No one paper can determine a debate, but each contributes to a pattern, and eventually to a shifting of opinion as to where the probable truth lies. Until 2011 the studies of the genetics of intelligence were based on twin studies, which are fine; and adoption studies, which give some indications if the samples are of reasonable size and followed for long enough; and admixture studies, which give a rough indication of the likely size of the genetic effect.
The hypothesis that there is a genetic component to all human differences does not stand or fall by a single observation, but to a pattern of results.
At a very simple level, it can be observed that parents have children who share their inherent characteristics, most notably that parents transmit their ancestral group characteristics. That transmission includes many aspects which are more than skin deep, which include differences in brain which lead to differences in behaviour.
At a more complicated level, it can be argued that the genetic transmission of characteristics and behaviours is strongly influenced by the environment, and that the effects of genetics will be less powerful if environments are bad. If so, this could complicate the investigation of genetic factors in racial differences in intelligence. Do heritability estimates vary by race?
This was reviewed earlier this year by Pesta and colleagues:
Racial and ethnic group differences in the heritability of intelligence: A systematic review and meta-analysis
Bryan J.Pesta, Emil O.W.Kirkegaard, Jante Nijenhuis, Jordan Lasker, John G.R.Fuerst
https://doi.org/10.1016/j.intell.2019.101408
https://www.sciencedirect.com/science/article/pii/S0160289619301904
They say:
Via meta-analysis, we examined whether the heritability of intelligence varies across racial or ethnic groups. Specifically, we tested a hypothesis predicting an interaction whereby those racial and ethnic groups living in relatively disadvantaged environments display lower heritability and higher environmentality. The reasoning behind this prediction is that people (or groups of people) raised in poor environments may not be able to realize their full genetic potentials. Our sample (k = 16) comprised 84,897 Whites, 37,160 Blacks, and 17,678 Hispanics residing in the United States. We found that White, Black, and Hispanic heritabilities were consistently moderate to high, and that these heritabilities did not differ across groups. At least in the United States, Race/Ethnicity × Heritability interactions likely do not exist.
These are large samples. Studies of this type compare the relative contributions of heritability (a narrow measure of the additive effect of genes) and everything else: “environmentality”, which has two components: the common, shared circumstances of family life which make families alike, and the unique, personal accidents of fate, and the partly self-created experiences which make family members different; plus plain old error of measurement.
A is for additive genetic inheritance, C is for common family factors, and E is for exceptional events and error. This is the ACE model. It is relatively simple, and if there are any interactions between the three components, these can be identified and measured.
Results for the general population show that the proportion of variance in IQ explained by genes increases with age (Plomin et al., 2014). Specifically, in early childhood, genetic effects explain less than 50% of IQ variance, and the effect of the shared environment is relatively strong. As children age, though, genetic effects become increasingly prominent, and the environmental variance due to factors common to siblings decreases. In adults, the heritability of intelligence is 60–80%, while the effect of common environment is small, if not zero (Plomin et al., 2014). The unshared environment explains the rest.
This is the astounding recent finding which would have confused just about all researchers in the 1960s, including myself, who expected that the common effects of family life would be extremely strong, as the sociologists claimed.
Are these findings true for poor people, whose environments are poor at nurturing talent?
One putative moderator is the quality of one’s environment. Poorer (richer) environments supposedly correspond to lower (higher) heritability, to a presumably measurable degree. Said differently, “natural potentials for adaptive functioning are more fully expressed in the context of more nourishing environmental experiences” (Tucker-Drob & Bates, 2016, p. 1). This prediction is known as the Scarr-Rowe hypothesis (Scarr-Salapatek, 1971; Turkheimer, Harden, D’onofrio, & Gottesman, 2011).
How does this apply to race differences in intelligence? Most people who have paid even passing attention to a century of data now accept that there are racial differences in intelligence, but there is less agreement as to how much of that difference can be attributed to genetics. If bad environments reduce the heritability of intelligence for poorer races, this would buttress the hypothesis that their poor performance is largely due to poor circumstances. Improve those circumstances, and performance should improve quickly and substantially.
For example, Lee et al. (2018) found that they could predict intelligence (or about 13% of it) in Europeans just on the basis of genetics. When they used the European prediction on Africans, they predicted 1.6% of it. Not much, you may say, but Lee et al. explain that all European based predictions lose accuracy when used on Africans, because the snips of genetic code may not be at exactly the same position in the sequence, a problem which goes by the daft name of linkage disequilibrium. (The individual variations which make you unique stand out from the usual pattern for your racial group, so your sequences are a little bit out of equilibrium from the rest of your tribe). The right musical notes, but not necessarily in the right order, as an English comedian said when his dreadful piano playing was criticized by conductor Andre Previn.
The Lee et al. explanation makes sense to me, but it could also be argued that it is the lousy environments which account for the lack of relationship between genetic markers and intelligent outcomes.
In a process which took several years, the authors went through all the studies which had investigated racial differences in intelligence, and had sufficient data to make ACE calculations.
What did the authors find?
There are many comparisons and tables of results, but the most contentious are the black-white comparisons, since those are assumed to have the biggest differences due to historical and cultural factors, and those show no differences in heritability. Despite cognitive differences (White-Black mean d = .83 and White-Hispanic mean d = .60) there were no heritability differences in either case. This damages the argument that a substantial cause of low ability is poverty and poor environments. It makes it more likely that the differences are due to inherited genes.
In conclusion, our meta-analysis reveals that the heritability of cognitive ability is generally moderate to high for Whites, Blacks, and Hispanics in the United States. The other groups featured here (e.g., Asians) had sample sizes that were too small to allow making strong conclusions. We also found that differences in heritability across these three groups were mostly trivial. Nonetheless, we cannot rule out the existence of modest differences in population parameters in our analyses. We can, however, conclude that the correlations between phenotype and genotype are essentially the same for Whites, Blacks, and Hispanics residing in the USA.
This is an important finding, because some researchers have argued that “environmentality” is a major factor which diminishes the real-world effect of genetics. In crude terms, some proponents of this argument seem to think that high heritabilities restrict how much change can be achieved by social improvements. They might be right, but social improvements benefit everybody, even though they raise all boats, without necessarily annulling the differences between those boats. Some ride higher out of the water than others. The authors have shown that the heritability estimates hold up equally for white and blacks, and for whites and hispanics.
As always, we have to put in a caveat that more studies might show something different. A good way forward would be to collect much more data on African intellectual and scholastic attainments, and to link that to African DNA. How long will it take to gather that for 1.1 million Africans? Then we can compare European to African predictions with African to European predictions, which should increase our understanding considerably. Until that happens, perhaps we can improve the European to African prediction by doing more work on the gene linkages.
English
Arabic
Chinese (Simplified)
Dutch
French
German
Hebrew
Hindi
Italian
Japanese
Korean
Persian
Polish
Portuguese
Russian
Spanish
Swedish
Turkish
Vietnamese
Well that’s interesting – and surprising (to me, at least).
But how about a different interpretation? It turns out that IQ-promoting, and IQ-impairing, environments for different racial groups in the USA differ only triflingly. In other words “relatively disadvantaged environments” are scarcely disadvantaged at all. It’s all much of a muchness, for IQ at least.
Is there any way of telling how long this state of affairs has lasted?
P. S. What is K, doc? Does it differ from k?
The applicability of Falconer’s formula: H^2=2(r_MZ – r_DZ) is questionable. Strict additivity must be assumed. This assumption is wrong. To go around it the doctrinaire Zhdanovs of the IQism like Plomin make hand waving argument of environment being genetically mediated or in other words that genes select environment.
The summary of heritability values in the Table 3
White 0.58
Black 0.60
Hispanic 0.73
Asians 0.30
raises intersecting questions. The ethnic group that usually achieves the highest IQ test scores has the lowest heritability while the ethnic groups that usually have the lowest IQ scores (Blacks and Hispanics) have the highest heritability. Does it make sense? Yes, if you go against everything that the doctrinaire IQists want us make to believe. Heritability is not fixed just like IQ is not fixed. Both are chiefly controlled by environment. You work harder and study more then you get higher IQ score but not all DZ twins (see Falconer’s formula) study equally hard while among Blacks and Hispanics most DZ twins study equally not hard, so their heritability is high and IQ scores are low.
Oh, God, Jesus in heaven Mary and Joseph. Just take a gander at all them there charts and graphs and huge indigestible gobbets of “technical” language meant to ostracize any possible critics. Isn’t it funny how, after all that snow, the conclusion this guy draws from it all is just plain bald prejudice, sans one smidgeon of objectivity.
What constitutes “good” and “bad” as far as environments are concerned?
Sure, comfortable Europeans have a better environment than dirt-poor Africans do today. But 10,000 years ago? 25,000? 50,000?
It could be argued, and often is, that the difference with Africans from all the other continents’ inhabitants is that the environment of the former was too damned easy. Low-hanging fruit, literally.
“A good way forward would be to collect much more data on African intellectual and scholastic attainments, and to link that to African DNA. How long will it take to gather that for 1.1 million Africans?”
Might be quicker to compare the Ibo to the Fulani and Yoruba, if it was politically possible.
It is not that big of a stretch to think that groups of people that had to contend with winter environments cropping up every year like clockwork might have developed higher intelligence in order to deal with the lack of “low hanging fruit,” freezing temperatures, and migratory patterns in hunted animals during these challenging times in their lives, every year.
What would you predict about those three?
Iirc the predictor that accounted for close to 13% of variance had about 1,000,000 SNPs. Now if the same predictor accounts for 1.6% of variance only on an independent sample of Africans, instead of hypothesizing structural race differences I would suggest to the authors looking into the overfitting problem first that could have occurred on the European sample and if it did it would render their 13% result invalid.
Ibo seem to be very bright (despite the outcome of the Biafran war), and also clannish. Be interesting to see the genetic differences. On the other hand the Chinese have been looking at genes for intelligence but I haven’t seen the results. Of course I guess its possible (tho unlikely IMHO) that what makes for intelligence is different between ethnic groups.
btw I forgot the Hausa, Nigeria’s main ethnic group.
https://www.jameslafond.com/article.php?id=7679
What about the intelligence of Arabs and western Indians come into this, the builders of the great Mesopotamian, Egyptian and Indus valley civilisations? What was it that lead to the eastern Mediterranean race having such genius among them? I doubt it was the harsh winters seeing the climate of the near east is generally pretty benign in the winter, sure it can get chilly on the Iranian plateau but most of it is very pleasant during the winter months.
On selection grounds we would expect genes to buffer phenotypes from the effects of environment.
If only for the reason that during bad times (bad environments) those genes more sensitive to environmental effects would be removed from the gene pool.
I’ll go and get my baton.
This is good information, but to use it effectively in an argument it has to be deployed as a counter to disparate impact. Point out how disparate impact is an assumption that does not meet the balance of probability required when the law takes from the defendant to gives to the plaintiff as civil compensation.
Scientific evidence is best used as a counter to disparate impact arguments. Like an uppercut in boxing it is very rarely if ever led with, but ideal to finish a combination. Always make clear that it is the disparate impact principle that starts by making a scientific assumption for which evidence is weak.
Of course, in effect the hereditarian has the task of a prosecutor in a criminal trial: meeting the burden of proof. But never just go in with scientific arguments, which will make you seem irresponsible. Always use them as a counter to disparate impact, and preface the scientific point with a disclaimer about how you are doing so reluctantly and only because there are unwarranted assumptions underlying disparate impact evidence.
A very surprising a result.
One possible (post hoc) explanation is to consider the FLynn effect. I haven’t heard of any FLynn interactions by race. Perhaps viewed closely enough there are differences in the human condition across races; however, relative to historical trends, the last 100 years have been a tremendously good environment for everyone? (At least on average, individuals may have crushingly bad environments)
Myriad intelligences exist, and measurement of intelligence is fraught with cultural bias too pervasive too quantify. Empirically, intelligence is measureable if the conditions are right, but invariably intelligence testing is biased as Hell and not worth any discussion whatsoever.
Modern day Eugenicists still promulgate this kind of tripe, but if one looks deep enough at the basis for the thesis one is left wanting for the metrology.
Intelligence is in no way objectively measured or culturally specific for anyone that exists outside of the baseline normative of WASP culture.
Eugenicists are best called out for what they really are.
Bigotry and racism are invariably disguised as objective ‘science’, but when one looks in-depth at the subject matter, and victims of it, one invariably finds what is most aptly described as pseudoscientific clap trap mixed in with what appears to be hard scientific so-called ‘data’.
Intelligence must be tested longitudinally over the life span as spot measures of intelligence are not accurate in any way.
The whole field of Intelligence Testing is like navel gazing for non-scientists where one claims interest in belly button lint instead of real scientific questions of import.
Eugenics & Eugenicists are a dime a dozen these days, eh.
RW
My own opinion is that your grandparents are more important than your parents when it comes to IQ. If you have 3-4 smart grandparents, all the grandkids are very likely to be smart. With just 1 smart grandparent, all the grandkids are likely to be dumb. The genes for brains become weighted in your favor if you have at least 3-4 smart grandparents to inherit gene blocks from.
Two smart parents could become brainly just from 1 smart parent each, but this would give the grandkids only 2 smart grandparents to inherit from. However, you need 3-4 smart grandparents to tip the odds decisively in the favor of high intelligence in the grandkids. 2 smart and 2 dumb grandparents means it’s likely some of vital blocks will be left out, and it’s becomes 50-50 chance that any of the grandkids will be smart.
Although no one seems to like to consider this possibility, it may be likely that the building blocks for intelligence are recessive genes. Dumb genes may be dominant, and that’s why it’s vital that you should never have kids with a dumb person. Whites, historically speaking, developed in Northern Europe in small populations in the shadow of the ice age when it was tough to survive, and recessive genes for brains may have been able to spread more easily in tiny groups of people.
It’s plain that the rest of the world has large amounts of dumb genes, so that’s an argument that their dumb brain genes are dominant. When smart black parents have kids, they tend to have dumb kids, unlike whites. I suspect the black grandparents had too many dumb dominant genes that filtered down to the grandkids, and any genes for high intelligence were recessives that were ‘shut off’ when inherited. This may be why black kids of successful black parents keep failing to match their parents in brains and levels of success.
There has never been a black-white mulatto that has ever created any great achievements in math or the sciences. It’s as if black genes automatically shut down high achievement. This is another argument that dumb black brain genes are dominant genes.
If I understood you correctly, indeed the imposed unsystematic approach to data gathering is painful, goes against science and rational approaches, against faster progress of knowledge. Analysis of the data blob comes in secondly. Poisoned wells, poor data, too small data-pools are first line. This is an element that always returns in public data seemingly. Economics, sociology, psychology, in short any public communication carries some intentional obfuscation of a sort.
James Flynn has abandoned the hope that the Flynn effect will annul racial differences in intelligence. Rindermann and I had argued in 2013 that there had been a narrowing of the difference in the US, and colleagues in the same special edition had argued that there was a rising trend in Africa itself, though the rate of progress was slow.
Jensen. Bias in Mental Testing. Free Press, 1980.
Also, this one:
https://www.unz.com/jthompson/africa-and-the-cold-beauty-of-maths/
Joseph Bologne, Chevalier de Saint-Georges was a champion fencer, classical composer, virtuoso violinist, and conductor of the leading symphony orchestra in Paris. Born in the French colony of Guadeloupe, he was the son of George Bologne de Saint-Georges, a wealthy married planter, and Anne dite Nanon, his wife’s African slave.Wikipedia.
I seem to remember that he was something of a mathematician too, though I’m not sure of that.
Dumas was half black, wasn’t he? See photo.
https://en.wikipedia.org/wiki/Alexandre_Dumas
Pushkin was a little bit black.
https://en.wikipedia.org/wiki/Alexander_Pushkin
As a measure of individual potential between members of a class in Eton, Archbishop McGrath Comprehensive or a rural school in Zambia IQ is a reasonable tool. As a measure of group ability between each of those groups it seems to me ridiculous. As a minimum, they have all been exposed to different levels of linguistic complexity. The culture in Zambia barely includes counting.
If you read the article carefully it seems Dumas was 1/4 black.
What were his mathematical/scientific achievements? Being a champion fencer, a virtuoso violinist and a conductor is notable, but has nothing to do with mathematics or science (mathematics is, of course, a science, but if I may I’ll consider “science” here to mean the natural/applied sciences). Just like one might appreciate Pushkin, Dumas and Machado de Assis very much, but they contributed nothing towards mathematics or science. None of those three were mathematicians or scientists in any way.
Note that having a black, a mulatto, a quadroon, an octoroon, whatever you call them, be a mathematician or a scientist is still not enough — he must have contributed something worthy of to their field. There might actually be a few, though no names come to my mind, certainly not regarding major accomplishments. Most mathematicians and scientists, including the white ones, make marginal contributions at best, and accomplish very little. There are many black mathematicians and scientists roaming this Earth today, and the vast majority of them will not make any lasting contributions to either fields at all.
Don’t environmental influences, especially family influences, also have a genetic component?
“Exposed to different levels of linguistic complexity”. Where did that come from? Did it fall like dew from the heavens? Or do you actually mean that some peoples have established more linguistic complexity than others, because they have created more words as mental tools?
“The culture in Zambia barely includes counting”. For a moment, let us assume that it true. You have begun your statement by saying that group comparisons using standard assessments are ridiculous, and now, without recognizing any contradiction, you assert that Zambians barely count. There: you have used numeracy as a distinguishing feature between one culture and another.
You’re right. Does this mean that the only utterly authentic half-black genius the world has ever known is Barack Obama?
There are some 1,000 genes on the X chromosome. There are only about 50 on the Y chromosome. Obama’s mother was white. Ignoring the rumors about Frank Marshall Davis, that suggests Obama is less than 50% black.
So, I guess I can only quibble about the half-black part.
Absolutely not. Simply because the Yemeni gene pool due to its monstrous inbreeding correlates perfectly with the lower performance that they record, and given that the IQ considering the measurement error is 100% genetic and the genetic structure of the Yemeni has exactly the degree of deleterious recessive alleles that one would expect from those who have these benefits it necessarily follows that the correlation is totally causal, ergo the inheritance is 100% genetic in Yemen, therefore an extremely representative context of the most absolute deprivation. Everything is about 100% genetic, including opinions on rights, religiosity and intelligence itself. This is when you carefully consider the measurement error, mainly due to the test-retest differences. This also through time: the Flynn Effect is totally genetic, due to the lower homozygosity due to urbanization and the reduction of family consanguinity, today homophobia compared to 1900 is extremely less common in western societies for genetic changes, including the sexism, average income and racial discrimination. Japan around 1900 was extremely poorer than now solely because they were genetically different. More than half of Arab Muslims, if raised in an indigenous western middle family, would voluntarily convert to Islam. If you took a 1900 Japanese, as well as an average African today and raised him in a typical western family, he would end up gaining approximately what was normal in his context, therefore very little compared to the current average.
I believe that what ‘evidence’ you cite is the epitome of ‘anecdotal’.
R.C.
You sound like an earnest nincompoop. Are you?
Linguist complexity at Eton comes from access to a high level of cultural input enabled by hereditary wealth.
Zambian rural culture never needed to do much counting. Those who get access to book learning have no trouble counting.
Is your implication that Zambians never needed to count, but can do so if required; and that England had cultural input because of wealth? Is your further implication that if Zambia had hereditary wealth it would count better?
This.
The world is awash in data that’s rotten even before it gets badly-analysed by HelloWorld-level quants.
One of the key things I was told in my early grad-student days (in the mid-1990s) was “Don’t trust the data“.
If government publishes a number and says it is the ‘unemployment rate‘, don’t believe it bears any relationship whatsoever to any sensible measure of labour under-utilisation.
If government publishes a number and says it is the ‘consumer price index‘, don’t believe it bears any relationship whatsoever to any sensible measure of increases in the cost of living.
That exhortation to never trust government data was partly in reaction to the old saw about bureaucratic laziness and incompetence (as retold by Josiah Stamp[1]), but was mostly in reaction to the overt and deliberate political manipulation of the reporting of key economic aggregates by “Mr Magoo” Greenspan and his cronies.
It was clear by the late 1980s that Western economies were top-heavy and massively bureaucratised, and that total factor productivity (properly measured) was slowing to a crawl – as would be expected when the private economy is being drained by the giant leech of government.
So Greenspan and his cronies set about redefining key concepts (GDP growth; price deflators; measures of labour utilisation) in such a way as to reduce all deflators significantly, and to understate unemployment.
Reducing reported price deflators does two things: it overstates real growth (and therefore makes total factor productivity look greater), and it understates cost-of-living increases (which helps bilk pensioners through reduced cost-of-living adjustments).
Greenspan and his corrupt cronies also helped birth the atrocious “net birth-death model” used by the BLS to goose job creation numbers – which adds hundreds of thousands of phantom jobs in every month for the purposes of the data release… the phantoms are then revised away once they are no longer in the front month.
It got to the stage in the early 00s that people were starting to notice an observable trend in GDP growth numbers: by the time the “final” number for a quarter stops being revised, it is on average 40% lower than the original estimate, with the final “final” number being published up to 8 quarters after the initial, market-goosing number. The BEA responded by changing nomenclature: what used to be the ‘final’ number is now called the ‘third estimate’ (which is treated as the final number).
As a recent example: on March 27th, the BEA reported ‘final’ 2.3% annualised growth for Q4 2019. By the latest data release (on June 25th) Q4’s “final” number has been revised down to 2.1%. This gives a lower base for the calculation of the decline for Q1 2020.
The “advance” number for Q1 2020 was initially reported on April 30 at -4.8% – “better than expectations”. As of June 27th, Q1 2020 is now -5% – which was in line with expectations in April (and hence would not have been “market moving”).
That gives you an idea as to why the first estimate is – almost always – an overstatement of growth: “beating expectations” causes a market move, whether it’s the old “beat by a penny” for quarterly EPS for a single stock, or an aggregate economic number. There are hundreds of millions of dollars made (and lost) in the 500msec after a ‘beat’.
This happens all the time, but in a world where the vast majority of analysts have no incentive to give a fuck, it becomes part of the noise. (And it’s much much much worse when some UDemy-level fucktard writes a script that makes a bot react to the numbers as read, 100msec after they’re released).
This chicanery is well-understood among quants who give a fuck about the data – it’s just that nobody’s incentivised to act as if they care; if you set things up so that your bot adjusted to the falsity of the data, your firm would be on the losing side of every HFT trade. That’s just Game Theory: there is a negative payoff to standing in front of a steamroller.
[1] “The government are very keen on amassing statistics. They collect them, add them, raise them to the nth power, take the cube root and prepare wonderful diagrams. But you must never forget that every one of these figures comes in the first instance from the chowky dar, who just puts down what he damn pleases.” – Some Economic Factors in Modern Life (1929)
Zambian rural culture never needed to do much counting.
Could be. But my experience of Scottish rural culture was that many people were ace at counting. Why the difference?
Including knowing their tables up to 20.
More or less, yes. Certainly the physical environment matters. I think Jared Diamond went too far making excuses in Guns, Germ and Steel but geography/environment matters. So do institutions and I think shaping of institutions between say, England and France, owes a lot to luck. Step further back to Germany (for long a place rather than a country) and Arabia and culture matters too. The choice to inbreed may have seemed like an intelligent way to preserve wealth once.
Reckoning in money in Scotland and social obligations in Zambia would be my guess. A lesser concept of private ownership in Zambia so less need to take account of it.
I find “luck” a poor argument.
Every stockman counts his herd, surely? Even in the days of Open Field agriculture no flock or herd was owned in common. Is it different among the Shona and Matabele?
P.S. I hadn’t known that the Matabele ran an apartheid state. Britannica: “The short-lived Matabele state became stratified into a superior class (Zansi), composed of peoples of Nguni origin; an intermediate class (Enhla), comprising people of Sotho origin; and a lower class (Lozwi, or Holi), derived from the original inhabitants.” Did you ever? Mankind, eh?
Do you understand what I said or not you fucking idiots?
Worth noting that the Table 3 figure for Asians is based on only 2 studies.. Whites, blacks, and Hispanics had 16/15/7 studies respectively.
The Hart et al. (2013) study pulled the Asian value down. It is interesting to contrast the heritabilities that study found for whites, blacks, Hispanics, and Asians respectively: 0.80, 0.89 (!), 0.50, and 0.24.
FWIW, Table 2 indicates that the two “Asian” studies were Asian/Other in reality.
What makes you think that? If I am interpreting Supplementary Tables 41 and 42 from their paper correctly:
https://static-content.springer.com/esm/art%3A10.1038%2Fs41588-018-0147-3/MediaObjects/41588_2018_147_MOESM3_ESM.xlsx
the GWAS predictor uses 1255 loci and predicts 12.7% for Add Health while the MTAG predictor uses 1624 loci and predicts 13.0% for Add Health.
Worth noting these are all for EDU rather than intelligence itself.
Perhaps you are confusing # SNPs with the sample sizes?
Not sure if I am misreading that, but it seems a bit off to me. This paper goes into a great deal of detail, but I suspect is more confusing than enlightening if one does not already know what LD is.
Linkage disequilibrium — understanding the evolutionary past and mapping the medical future
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5124487/
Simply stated, LD is the nonrandom association of alleles at different loci (from first sentence of paper above). The basic LD issue with using GWAS results from one population with another population is that the SNPs detected by a GWAS may not be causal themselves. Instead they may just serve as markers for nearby causal variation (because the SNPs are correlated with that variation due to LD in the GWAS population). The problem is that correlation may not exist (or just be less strong) in another population. Thus the “meaning” of the SNP may be different between populations.
Another way of looking at this is LD is something of an opposite to random association of SNPs.
It is worth noting that a common cause of LD is selective sweeps:
https://en.wikipedia.org/wiki/Selective_sweep#Detection
Thus this is most likely to be an issue in regions of the genome which were heavily selected for. Which also happen to be the regions we most care about for GWAS.
BTW, LD is one reason Manhattan plots tend to show spread out peaks. Another reason is that if a gene is important for a trait multiple variations within it are likely to have effects.
https://en.wikipedia.org/wiki/Manhattan_plot
P.S. Another issue with cross-group use of GWAS results is that causal alleles which are nearly fixed in the GWAS population but not in the other population may not be detected by the GWAS.
“What makes you think that?” – Two years ago when James Lee at al. paper was discussed here you did not object to the fact that about 1 million of SNPs were used. See my comment:
https://www.unz.com/jthompson/journey-of-1-1-million-miles/#comment-2436546
Let’s first go to their summary where they talk what was really accomplished:
This summary suggests that with just 1,271 SNPs they can explain 11-13% of education attainment variance. But is it true? They call these SNPs a treasure trove for future analyses. Let’s go to FAQs notes for more explanation:
The truth is which does not jump at you form the main body of the paper is that to obtain 11-13% prediction each “lead SNP” from the “treasure trove” of 1,271 has to be combined with additional 787 SNPs on average. ONE MILLION SPNs in the polygenic score is a lot. The polygenic score they constructed uses 10% of all SNPs there are in human genome.
Thanks. I asked because I vaguely recalled something like that, but could not find it in a quick look at the paper.
Did you look at the Supplementary Material I mentioned? Tables 41 and 42 (especially taken together given they are right next to each other) seem extremely misleading given the FAQ.
For anyone else interested, the FAQ is at https://www.thessgac.org/faqs
Annoyingly they don’t seem to offer a way to link directly to the FAQ for this paper. Be sure to navigate to the correct paper: “Gene discovery and polygenic prediction from a 1.1-million-person GWAS of educational attainment”
“…seem extremely misleading given the FAQ…” – This is actually very annoying. I just do not understand why some scientific papers are misrepresenting what actually has been done and then shine some light on the truth in the FAQ sections as if different truths were prepared for different audiences or were they forced to write the FAQ document by the reviewers.
This makes me think of what I have read today on another thread:
P.S. Thanks for the link to FAQ: https://www.thessgac.org/faqs