Better is to predict many external outcomes and average across them to find the best set of estimates. This is best on the assumption that maximizing reliability and construct validity will maximizing predictive validity. I suggest you download the SPI and use the indicators as your outcomes. Average across them (remember to use absolute values, or reverse by the S factor loading), and see which set does the best. https://www.socialprogress.org/
Or instead of averaging, make a model that predicts outcomes from some cases where you are really quite sure of the correct answer. Ideally Lasso regression IMO.
There's also the issue of selective migration. Personally, I doubt the genotypic IQ of natives in those countries is much lower than those of NW Europeans.
Any thoughts as to why Mexico seems to punch above its weight class compared to other Latin American countries despite having relatively higher Amerindian admixture? For example, Mexico would edge out Argentina even if the standard errors were applied in the latter's favor. This despite Argentina having a higher HDI, so it seems to me minor environmental effects cannot explain the gap. Perhaps pure Mesoamericans have a small but consistent cognitive advantage over indigenous Americans in South America?
In fact according to this Kenya scores higher than India a non African country. I am curious on why you think that is and why india has lower iq than other Asian countries
About half of "israel's" jewish population is mizrahi jews and a growing segment of the ashkenazi jews are the highly inbred ultra-orthodox jews. The secular, well-educated ashkenazi with somewhat western values are increasingly a rare breed.
Additionally, Mizrahim have among the highest attrition rates. Only 66% took the exam, compared with 71% of Muslims, 80% of Ashkenazim, and 85% of Christians. That's bound to deflate the Christian Arab IQ and inflate the Mizrahi IQ, assuming that data is correct all else being equal.
Income on Wikipedia, you can google for cheating scandal stuffs.
I agree that the Christian IQ there is too low. It’s a small population so hard to sample. 98 for all non-Ashkenazi makes sense, including Iraqi and Euro Sephardi who should be triple digits. I’d guess Arab Christians are also in the high 90s, but they’re 1% of Israel so the effect on NIQ is negligible.
What did you norm the IQ to, that is what did you set IQ=100 as? Usually it is set at the average IQ of the UK, but that doesn't seem to be the case here.
National IQ is not that meaningful for big countries like US, China, India, Indonesia, Brazil, etc. The variation of IQ within a country can be as large as that of an European country and an African country. Take Brazil. The southern Brazilian states have large White European populations while the north of the country is full of mixed/Africans. Their IQ difference is 1 SD just like between Europe and Africa. Han Chinese mean IQ is 101. But if you compare Hans and Tibetans/Uyghurs, there's also a 1 SD difference.
1. North/South Korea - basically all nutrition impact
2. Africa, Persia - poor education systems + nutrition
3. Surprised you didn't include Antarctica.
4. Israel scoring low indicates another hidden elephant in almost all of these colours - distribution. Some countries have elites that would score extremely high (Iran, Israel, Turkey, USA, South Africa) but due to the political system there are sizeable minority population with no nutrition and no education.
5. I'm still doubtful of Chinese data, but their nutrition is very good due to extreme overfishing of the whole Pacific (and other Oceans...)
6. I would argue with equal nutrition and education Sub-saharan Africa will have the highest IQ outliers in the world.
In conclusion, it's disappointing i haven't seen a good data on IQ, still. You can easily find huge databases from online IQ tests. Mensa has a short one, very robust, globally taken. These tests often have additional questions so data scientist could triangulate proper setting for each country to get an actually reliable result to correct for nurture. Hope i'm wrong and there is such statistics available.
Yeah, at age 17 in the Minessota Transracial Adoption Study the average of white adoptees raised in a middle class white household was 105.6, and for blacks it was 89.4.(The g Factor: The Science of Mental Ability)
Also i forgat to add one thing - human growth hormones - are what causes almost all of the genetic and most non-genetic, non-education related variation in intelligence among fellow humans. Very simple.
Over the past 80 years there has been a massive selection against political dissidents in North Korea. This may have influenced IQ, unsure to what extent though.
I see North Korea number isn't to deviatted from other estimates which It really proves the case of hereditarianism I mean literally there are way richer and with better quality of life countries with less IQ
have you considered the possibility that at least some of the data sets are just plain inadequate? That the phrase "most-accurate-possible" is a subjective estimation, and also implicitly a climb-down observation? That some of your observations are merely intuitive gainsaying? (ex. from the reply thread below: "I doubt the genotypic IQ of natives in those countries is much lower than those of NW Europeans")
I'm really dubious about pseudo-Big Data. Realize, I get the temptation: "Lookit this unprecedented amount of massive numbers generated to inform metrical assessments and comparisons!" But basic principles are disregarded. A problem that is hardly confined to psychology. But the elasticity of definitions and connotations related to interpretive assessments of human psychology make it particularly vulnerable to error.
In particular, time and again I notice a lack of insight about possible confounders (even ones that appear obvious to me*); the embrace of the fallacy that "correlation may not precisely yield causation, it's near enough"; hand-wave excuses to explain the reliance on a paucity of data "because it's all we have (but it sure looks like a lot on the screen!)"; and the "reverse statistical engineering" of generating selected metrics to support a preordained hypothesis. Or sometimes even inventing them: for instance, at least a few of the "national IQs" originally assigned by Richard Lynn were based conjectures related to the data interpretations of test results drawn from neighboring countries, because the countries in question had no data of their own to supply.
Speaking of test results, how does anyone imagine that they can accurately weight the results of different test methodologies used in different nations, or have any objective assurance of a random sampling for all of the countries supposedly assessed in order to assign their "national IQ?"
Finally, does anyone involved in this "worldwide national IQ assignment" project evince any appreciation of how vast and ambitious it is, in terms of the scope of its goal? I notice a remarkably casual and cavalier attitude toward the endeavor that doesn't strike me as especially professional, given the uneven and sometimes threadbare condition of the information database.
(*whereas you don't know what you don't know, do you?)
The Singapore IQ doesn't doesn't really sit right according to its demographics (assuming a hardcore hereditarian viewpoint). With 10% Indians and 15% Malays, it would be more believable if Singapore was closer to being high 90s rather being in same ballpark as East Asia proper. 25% non-Chinese isn't everything, but it certainly isn't nothing. Opponents of immigration to East Asia should take this seriously and spend some time on how to make sense of this. As it is, one could easily make the case that an East Asian country could be 25% non-East Asian, and still perform as well as pure East Asian countries.
A cope could be to say the Indians are literally all Brahmins (no idea if this is the case). But if it's been multiple generations like this, and the American H1B posters are correct that there's actually no difference between Indians and all of them are genetically bound to be retarded, regression to the mean should have caused the Indians diminish significantly in intelligence over time. Again, that's assuming they're all even Brahmin, which would be the strongest case for why they don't seem to lower Singapore's IQ enough to bring them lower than racially homogenous East Asia.
They are largely Tamils and other South Indians. I'm not sure of caste, but Wikipedia lists the composition of early migrants as follows:
1. Educated elite (Over 0.5%)
Generally university-educated professionals, such as lawyers and doctors, as well as senior civil servants. This group was wealthy and led a somewhat 'European' lifestyle.
2. Mercantile class (Under 10%)
Petty traders, moneylenders and other traditional business proprietors. Although often wealthy, they occupied a second-tier social rank as they were less well educated than the elite.
3. Middle class (5–7%)
English-educated white collar service-workers, including middle and junior ranks in the civil and military services. Occupations included teachers, nurses, policemen, postal workers, clerks, etc.
4. Working class (Over 80%)
Uneducated and unskilled manual labourers, or 'coolies'. They often worked at the port docks, in construction or private warehouses or factories.
About the last group:
"Coolies: a large number of Indian migrants were rural Tamils and Telugus who were landless peasants including many from shudra castes as well as Adi Dravidas or '(untouchables)'. They were transient unskilled workers who performed manual labour at the docks and construction sites. They were stereotypically seen as (and valued for being) docile."
Unless later migrations significantly changed these demographics, a good lot of Singaporean Indians must be lower caste or outcastes.
Completely wrong. Iran has been confirmed as a top 5 nation (2nd, 3rd and 4th globally based on 2022, 2024 and 2025 studies and meta-studies) in average IQ in several independent recent studies, alongside China and South Korea, yet you jokers keep using data from the least representative and completely outdated study (Lynn & Vanhanen) to bolster your bias and your preconceived ideological narratives.
Better is to predict many external outcomes and average across them to find the best set of estimates. This is best on the assumption that maximizing reliability and construct validity will maximizing predictive validity. I suggest you download the SPI and use the indicators as your outcomes. Average across them (remember to use absolute values, or reverse by the S factor loading), and see which set does the best. https://www.socialprogress.org/
Or instead of averaging, make a model that predicts outcomes from some cases where you are really quite sure of the correct answer. Ideally Lasso regression IMO.
National IQ have been my main obsession for the last few days. Spending countless of hours. Figuring out the correct numbers etc.
Thank you for this beautiful post seb. This feels like Christmas
About SE Europe results:
https://commons.m.wikimedia.org/wiki/File:Romani_population_average_estimate.png#mw-jump-to-license
There's also the issue of selective migration. Personally, I doubt the genotypic IQ of natives in those countries is much lower than those of NW Europeans.
Good point. And good source.
Any thoughts as to why Mexico seems to punch above its weight class compared to other Latin American countries despite having relatively higher Amerindian admixture? For example, Mexico would edge out Argentina even if the standard errors were applied in the latter's favor. This despite Argentina having a higher HDI, so it seems to me minor environmental effects cannot explain the gap. Perhaps pure Mesoamericans have a small but consistent cognitive advantage over indigenous Americans in South America?
Why Israel IQ is inferior to many white gentile countries?
African/Arab immigration.
why do you think for example Morroco has a lower iq than subsaharan African countries like kenya?
Not sure if this is the updated dataset, but I suspect that the cause is African admixture.
Yeah but isn't kenya pure african ancestry, yet Kenyans score higher than morrocans on iq tests
seems no answer sadly
In fact according to this Kenya scores higher than India a non African country. I am curious on why you think that is and why india has lower iq than other Asian countries
It seems plausible as they have a high proportion of them as I have checked
About half of "israel's" jewish population is mizrahi jews and a growing segment of the ashkenazi jews are the highly inbred ultra-orthodox jews. The secular, well-educated ashkenazi with somewhat western values are increasingly a rare breed.
Israel is only 35% Ashkenazi at most. Rest are Mizrahi 40% (IQ 95), Arabs 20% (IQ 85), Others (Ethiopians etc.) 5%.
Crem here https://www.cremieux.xyz/p/updated-estimates-of-iqs-within-israel said Israel has a g of ~97 (~110 Ashkenazi, ~100 non-Ashkenazi Jews and Russians, ~85 Arabs). PISA scores are lower I think mostly due to effort, when you extract the g-factor the score jumps quite a bit. Look here and command f Israel: https://www.oecd-ilibrary.org/docserver/cf9910a0-en.pdf?expires=1723936109&id=id&accname=guest&checksum=3F5B444018B0E83CDC886D6DC1AFC76A
Those results are suspect because Arab Christians have the highest PET pass rate, even higher than Jews.
https://www.timesofisrael.com/christian-arabs-top-countrys-matriculation-charts/
https://israel365news.com/280454/christian-arabs-likely-graduate-high-school-israel/#Yw4jp6yTtgoxsjcA.97
https://en.wikipedia.org/wiki/Christianity_in_Israel#High_school_and_matriculation_exams
Additionally, Mizrahim have among the highest attrition rates. Only 66% took the exam, compared with 71% of Muslims, 80% of Ashkenazim, and 85% of Christians. That's bound to deflate the Christian Arab IQ and inflate the Mizrahi IQ, assuming that data is correct all else being equal.
There’s a lot of cheating on the test in church schools. Their income is lower too.
Source on both of those claims?
Income on Wikipedia, you can google for cheating scandal stuffs.
I agree that the Christian IQ there is too low. It’s a small population so hard to sample. 98 for all non-Ashkenazi makes sense, including Iraqi and Euro Sephardi who should be triple digits. I’d guess Arab Christians are also in the high 90s, but they’re 1% of Israel so the effect on NIQ is negligible.
What did you norm the IQ to, that is what did you set IQ=100 as? Usually it is set at the average IQ of the UK, but that doesn't seem to be the case here.
UK Whites, if not I artificially set the UK's IQ to 99.2 (which seems to be the average when samples are normed to the mean of UK Whites).
National IQ is not that meaningful for big countries like US, China, India, Indonesia, Brazil, etc. The variation of IQ within a country can be as large as that of an European country and an African country. Take Brazil. The southern Brazilian states have large White European populations while the north of the country is full of mixed/Africans. Their IQ difference is 1 SD just like between Europe and Africa. Han Chinese mean IQ is 101. But if you compare Hans and Tibetans/Uyghurs, there's also a 1 SD difference.
Did you adjust for median age when you calculate National IQs? Aging causes cognitive decline. https://x.com/cremieuxrecueil/status/1749926101990670572
Interesting, almost zero genetics involved!
1. North/South Korea - basically all nutrition impact
2. Africa, Persia - poor education systems + nutrition
3. Surprised you didn't include Antarctica.
4. Israel scoring low indicates another hidden elephant in almost all of these colours - distribution. Some countries have elites that would score extremely high (Iran, Israel, Turkey, USA, South Africa) but due to the political system there are sizeable minority population with no nutrition and no education.
5. I'm still doubtful of Chinese data, but their nutrition is very good due to extreme overfishing of the whole Pacific (and other Oceans...)
6. I would argue with equal nutrition and education Sub-saharan Africa will have the highest IQ outliers in the world.
In conclusion, it's disappointing i haven't seen a good data on IQ, still. You can easily find huge databases from online IQ tests. Mensa has a short one, very robust, globally taken. These tests often have additional questions so data scientist could triangulate proper setting for each country to get an actually reliable result to correct for nurture. Hope i'm wrong and there is such statistics available.
Israel's low score is largely due to low effort: https://www.oecd-ilibrary.org/docserver/cf9910a0-en.pdf?expires=1723936109&id=id&accname=guest&checksum=3F5B444018B0E83CDC886D6DC1AFC76A
Cremieux factored out the g and the scores went up a lot: https://www.cremieux.xyz/p/updated-estimates-of-iqs-within-israel
I feel like you are over-estimating the affect of nutrition. In U.S., Blacks had less IQ than other races even when controlled for income.
Can you share a study where IQs were compared when controlling for income, nutrition, neighbourhood and education level?
Yeah, at age 17 in the Minessota Transracial Adoption Study the average of white adoptees raised in a middle class white household was 105.6, and for blacks it was 89.4.(The g Factor: The Science of Mental Ability)
Also i forgat to add one thing - human growth hormones - are what causes almost all of the genetic and most non-genetic, non-education related variation in intelligence among fellow humans. Very simple.
Over the past 80 years there has been a massive selection against political dissidents in North Korea. This may have influenced IQ, unsure to what extent though.
Why is Kosovo such a big outlier?
No idea. Probably selective migration + differential fertility.
Probably also partly because of SE Europe Gypsy-factor as mentioned by Stefan above.
There are actually very few gypsies
dysgenic migration, and in the case of Kosovo and Albania it continues for several centuries
I see North Korea number isn't to deviatted from other estimates which It really proves the case of hereditarianism I mean literally there are way richer and with better quality of life countries with less IQ
I imputed that one.
Buuuut they do suprisingly well in science competitions. Right tail is strong, I guess.
https://www.upi.com/Top_News/World-News/2019/07/22/North-Korea-finishes-fourth-at-International-Mathematical-Olympiad/2051563808996/
Indeed, Qatar and North Korea are good examples of a rich country with low IQ and a poor country with high IQ to prove the hereditarian position.
What are you even comparing with 'richness' when comparing a capitalist country to a true communist country?
"Richness" isn't just denominated in US dollars, it is standard of living.
have you considered the possibility that at least some of the data sets are just plain inadequate? That the phrase "most-accurate-possible" is a subjective estimation, and also implicitly a climb-down observation? That some of your observations are merely intuitive gainsaying? (ex. from the reply thread below: "I doubt the genotypic IQ of natives in those countries is much lower than those of NW Europeans")
I'm really dubious about pseudo-Big Data. Realize, I get the temptation: "Lookit this unprecedented amount of massive numbers generated to inform metrical assessments and comparisons!" But basic principles are disregarded. A problem that is hardly confined to psychology. But the elasticity of definitions and connotations related to interpretive assessments of human psychology make it particularly vulnerable to error.
In particular, time and again I notice a lack of insight about possible confounders (even ones that appear obvious to me*); the embrace of the fallacy that "correlation may not precisely yield causation, it's near enough"; hand-wave excuses to explain the reliance on a paucity of data "because it's all we have (but it sure looks like a lot on the screen!)"; and the "reverse statistical engineering" of generating selected metrics to support a preordained hypothesis. Or sometimes even inventing them: for instance, at least a few of the "national IQs" originally assigned by Richard Lynn were based conjectures related to the data interpretations of test results drawn from neighboring countries, because the countries in question had no data of their own to supply.
Speaking of test results, how does anyone imagine that they can accurately weight the results of different test methodologies used in different nations, or have any objective assurance of a random sampling for all of the countries supposedly assessed in order to assign their "national IQ?"
Finally, does anyone involved in this "worldwide national IQ assignment" project evince any appreciation of how vast and ambitious it is, in terms of the scope of its goal? I notice a remarkably casual and cavalier attitude toward the endeavor that doesn't strike me as especially professional, given the uneven and sometimes threadbare condition of the information database.
(*whereas you don't know what you don't know, do you?)
Proof that Finnish people are Asian
How much does being raised in a Western country improve the IQ of Third Worlders?
Presumably through health and nutrition, though the evidence for it is not great at the moment.
The Singapore IQ doesn't doesn't really sit right according to its demographics (assuming a hardcore hereditarian viewpoint). With 10% Indians and 15% Malays, it would be more believable if Singapore was closer to being high 90s rather being in same ballpark as East Asia proper. 25% non-Chinese isn't everything, but it certainly isn't nothing. Opponents of immigration to East Asia should take this seriously and spend some time on how to make sense of this. As it is, one could easily make the case that an East Asian country could be 25% non-East Asian, and still perform as well as pure East Asian countries.
A cope could be to say the Indians are literally all Brahmins (no idea if this is the case). But if it's been multiple generations like this, and the American H1B posters are correct that there's actually no difference between Indians and all of them are genetically bound to be retarded, regression to the mean should have caused the Indians diminish significantly in intelligence over time. Again, that's assuming they're all even Brahmin, which would be the strongest case for why they don't seem to lower Singapore's IQ enough to bring them lower than racially homogenous East Asia.
And none of that is to speak of the Malays
They are largely Tamils and other South Indians. I'm not sure of caste, but Wikipedia lists the composition of early migrants as follows:
1. Educated elite (Over 0.5%)
Generally university-educated professionals, such as lawyers and doctors, as well as senior civil servants. This group was wealthy and led a somewhat 'European' lifestyle.
2. Mercantile class (Under 10%)
Petty traders, moneylenders and other traditional business proprietors. Although often wealthy, they occupied a second-tier social rank as they were less well educated than the elite.
3. Middle class (5–7%)
English-educated white collar service-workers, including middle and junior ranks in the civil and military services. Occupations included teachers, nurses, policemen, postal workers, clerks, etc.
4. Working class (Over 80%)
Uneducated and unskilled manual labourers, or 'coolies'. They often worked at the port docks, in construction or private warehouses or factories.
About the last group:
"Coolies: a large number of Indian migrants were rural Tamils and Telugus who were landless peasants including many from shudra castes as well as Adi Dravidas or '(untouchables)'. They were transient unskilled workers who performed manual labour at the docks and construction sites. They were stereotypically seen as (and valued for being) docile."
Unless later migrations significantly changed these demographics, a good lot of Singaporean Indians must be lower caste or outcastes.
Kek. Muh superior Aryan genes saaar...
Completely wrong. Iran has been confirmed as a top 5 nation (2nd, 3rd and 4th globally based on 2022, 2024 and 2025 studies and meta-studies) in average IQ in several independent recent studies, alongside China and South Korea, yet you jokers keep using data from the least representative and completely outdated study (Lynn & Vanhanen) to bolster your bias and your preconceived ideological narratives.