I don't know how much of the Flynn effect is a test familiarity effect, but it seems like a significant part of it must be real.
For example, let's take a look at the performance of North Koreans, we know that their IQ (something like 85-90) is 1 SD lower than South Koreans.
Do they do poorly on tests just because of a test bias? Well, we can easily find this by their performance in different categories such as delinquency, income and innovation:
First: income and employment(1)
"More than 60% of the defectors live at Seoul and Metropolitan Area. Before defection, most of them were not employed (49%) or worked as labor workers (39%). Three quarters of them had educational background of high school graduation."
Second: crime rate(1)
"During the last ten years, About 20% of North Korean defectors stayed in South Korea have committed crimes twice much higher than the average crime rate committed by South Koreans. The National Police Agency (2007) reported that 678 criminals (40.2% of the total North Korean defectors) committed crimes of murder, rape, injury and violence."
Third: innovation(2)
For this, the only fair thing I could measure their abstract thinking and creativity at a high level was the math olympiad medals, but even at that they are far worse than South Korea.
It is true that North Korea has a significant number of Math Olympiad medals relative to its population, but South Korea's world ranking in Math Olympiad is 4 and North Korea's is 26, a stark difference.
As you can see, the northerners perform much worse than their southern counterparts, both on average and at a high level, so it seems that at least 50% to 65% of the difference between these two groups is nothing but a real effect on intelligence and The effect is the same as the Flynn effect because the children of North Korean defectors do much better in South Korea than their parents.
One must be careful about the selection bias with regards to the North Korean defectors. Where do you see the outcomes of their children? I don't doubt the results, but numbers are better.
It has been a relatively short period since North and South Koreans were separated by ideology; it is doubtful that genetic differences would have occurred.
The claim that, “people around 1900 would be around 70 on some tests” — obviously, at the time, they got 100 on the tests. Why not just give the exact same test to modern populations, and check the difference? What would that show? Seems everyone has confused themselves with re-meaning things.
So is the intuition for this something like the following?
If you measured intelligence by how quickly people found the roots of a polynomial, that would depend not just on their intelligence, but also whether they formally studied that concept. If schools teach that more often today, tests will show improvement over time. Whereas if you asked those same people to simply sum large numbers, they might perform worse.
So the point here is that something like the above is happening with IQ tests? Lacking a background in this subject, it isn't clear to me what is meant by the distinction between genotype and phenotype with respect to intelligence, and the linked article didn't clearly articulate that. Unless the distinction is still theoretical at this point.
A particular skill, like adding numbers, can be trained, but variation between people also reflect their general intelligence, and its genetic component. Children spend more time in school now than they ever have in history, yet math tests don't show improvements. Unless we are teaching less addition than previously -- doubtful -- this is plainly inconsistent with a large increase in intelligence. An increase in intelligence would have to be offset by some other decline to make the numbers work out.
Genotypic intelligence is your genetic expectation of intelligence, phenotypic intelligence is your actual level of intelligence. The two aren't the same value because non-genetic factors affect your actual intelligence as well.
Again, the number one unadressed topic in human intelligence is: why is there an upper limit? Why isn't the average IQ 120, or 200, or 2,000? It has nothing to do with brain size. My parrot, with a brain about the size of a hazelnut was more intelligent than my dog, with a brain about the size of a tomato. If IQ correlated with brain size, elephants would be super-geniuses, and whales would be establishing colonies on Mars. There must be a reason why human IQ ends at about 200, but intelligence researchers are obsessed with small variances between different times and places.
2) Brain size is relevant, but brain size relative to body size is more important.
3) There's no known upper limits on intelligence, human or otherwise. Statistically speaking, sampling from a normal distribution makes it exceptionally unlikely to find a person with +8 SD intelligence, and our tests can't normally measure that high anyway, so we can't find such a person if they exist.
Yes, but it also comes with an assumption that larger animals aren't smarter, which they probably are. Regressing out body size makes this indirect assumption. That's why some studies find that one should use both absolute brain size and the residualized brain size.
Natural selection is extremely unflexible. E.g. mammals even have problem evolving other than 7 neck vertebrae or more than 5 fingers. Somehow increasing body size needs to make brain larger too, for same intelligence. A simple fit, for mammals, is to divide brain size by mass^0.75. You can search more about this by searching for "allometric brain size"
They are not super credible, however, some kind of correction of the observed decline is needed. PGS will indicate the direction and relative magnitude, but we would like to know what that corresponds to in modern IQ points. Same issue as with the races comparisons and the ancient genomes.
I don't know how much of the Flynn effect is a test familiarity effect, but it seems like a significant part of it must be real.
For example, let's take a look at the performance of North Koreans, we know that their IQ (something like 85-90) is 1 SD lower than South Koreans.
Do they do poorly on tests just because of a test bias? Well, we can easily find this by their performance in different categories such as delinquency, income and innovation:
First: income and employment(1)
"More than 60% of the defectors live at Seoul and Metropolitan Area. Before defection, most of them were not employed (49%) or worked as labor workers (39%). Three quarters of them had educational background of high school graduation."
Second: crime rate(1)
"During the last ten years, About 20% of North Korean defectors stayed in South Korea have committed crimes twice much higher than the average crime rate committed by South Koreans. The National Police Agency (2007) reported that 678 criminals (40.2% of the total North Korean defectors) committed crimes of murder, rape, injury and violence."
Third: innovation(2)
For this, the only fair thing I could measure their abstract thinking and creativity at a high level was the math olympiad medals, but even at that they are far worse than South Korea.
It is true that North Korea has a significant number of Math Olympiad medals relative to its population, but South Korea's world ranking in Math Olympiad is 4 and North Korea's is 26, a stark difference.
As you can see, the northerners perform much worse than their southern counterparts, both on average and at a high level, so it seems that at least 50% to 65% of the difference between these two groups is nothing but a real effect on intelligence and The effect is the same as the Flynn effect because the children of North Korean defectors do much better in South Korea than their parents.
Sources:
1-
https://www.kicj.re.kr/board.es?mid=a20201000000&bid=0029&list_no=12533&act=view
2-
https://en.wikipedia.org/wiki/List_of_countries_by_medal_count_at_International_Mathematical_Olympiad
One must be careful about the selection bias with regards to the North Korean defectors. Where do you see the outcomes of their children? I don't doubt the results, but numbers are better.
It has been a relatively short period since North and South Koreans were separated by ideology; it is doubtful that genetic differences would have occurred.
The claim that, “people around 1900 would be around 70 on some tests” — obviously, at the time, they got 100 on the tests. Why not just give the exact same test to modern populations, and check the difference? What would that show? Seems everyone has confused themselves with re-meaning things.
That's what I proposed in the end.
Probably they would find Flynn effect is much smaller, and it is against modern dogma
So is the intuition for this something like the following?
If you measured intelligence by how quickly people found the roots of a polynomial, that would depend not just on their intelligence, but also whether they formally studied that concept. If schools teach that more often today, tests will show improvement over time. Whereas if you asked those same people to simply sum large numbers, they might perform worse.
So the point here is that something like the above is happening with IQ tests? Lacking a background in this subject, it isn't clear to me what is meant by the distinction between genotype and phenotype with respect to intelligence, and the linked article didn't clearly articulate that. Unless the distinction is still theoretical at this point.
A particular skill, like adding numbers, can be trained, but variation between people also reflect their general intelligence, and its genetic component. Children spend more time in school now than they ever have in history, yet math tests don't show improvements. Unless we are teaching less addition than previously -- doubtful -- this is plainly inconsistent with a large increase in intelligence. An increase in intelligence would have to be offset by some other decline to make the numbers work out.
Genotypic intelligence is your genetic expectation of intelligence, phenotypic intelligence is your actual level of intelligence. The two aren't the same value because non-genetic factors affect your actual intelligence as well.
Got it, thanks for the response 👍
Again, the number one unadressed topic in human intelligence is: why is there an upper limit? Why isn't the average IQ 120, or 200, or 2,000? It has nothing to do with brain size. My parrot, with a brain about the size of a hazelnut was more intelligent than my dog, with a brain about the size of a tomato. If IQ correlated with brain size, elephants would be super-geniuses, and whales would be establishing colonies on Mars. There must be a reason why human IQ ends at about 200, but intelligence researchers are obsessed with small variances between different times and places.
You have many misunderstandings.
1) The average IQ is an arbitrary number.
2) Brain size is relevant, but brain size relative to body size is more important.
3) There's no known upper limits on intelligence, human or otherwise. Statistically speaking, sampling from a normal distribution makes it exceptionally unlikely to find a person with +8 SD intelligence, and our tests can't normally measure that high anyway, so we can't find such a person if they exist.
Allometric equation is much more better than brain-to-size ratio https://en.wikipedia.org/wiki/Encephalization_quotient
was much written about by Harry J. Jerison in 1960-1970ths
Yes, but it also comes with an assumption that larger animals aren't smarter, which they probably are. Regressing out body size makes this indirect assumption. That's why some studies find that one should use both absolute brain size and the residualized brain size.
Natural selection is extremely unflexible. E.g. mammals even have problem evolving other than 7 neck vertebrae or more than 5 fingers. Somehow increasing body size needs to make brain larger too, for same intelligence. A simple fit, for mammals, is to divide brain size by mass^0.75. You can search more about this by searching for "allometric brain size"
>This is about the estimate one gets from correcting the polygenic score results
I'm not sure how defensible those corrections are. The uncorrected decline is consistent with IQ ~ fertility correlation in Iceland.
They are not super credible, however, some kind of correction of the observed decline is needed. PGS will indicate the direction and relative magnitude, but we would like to know what that corresponds to in modern IQ points. Same issue as with the races comparisons and the ancient genomes.