37 ms·

The catch here would be thinking in distributions, and from that separating individuals from said distributions. A current topic would be representation within

by esyir 5y ago

The catch here would be thinking in distributions, and from that separating individuals from said distributions. A current topic would be representation within a given domain. Particularly since in most representation cases (high paying jobs, celebrities, top universities, etc), we're already looking at the tails of the distribution. There, small shifts can heavily affect the quantity of users in the right tail, especially for domains with stronger requirements (see your case with the runners). Of course I won't try to discount potential cultural impact, but anyone trying to sell you a story of how it's all nature or all nurture is typically spouting bullshit.

tsimionescu 5y ago
But it's a bad idea to compare the world's fastest runners with people getting high paying jobs - that's the main point of the GP's comment. Even for something as clearly and uncontroversially genetically influenced as running ability, the group differences only really matter for the top 100 people out of 7 billion. If you look at top 100k runners, the group differences are almost entirely gone. Even more, the top 100k runners are probably within a second or two of each other - entirely irrelevant if you were to hire based on a bar of performance, and not some elitist concept of hiring the very best possible.
- inglor_cz 5y ago
  Socially, who makes it to the top hundred matters a lot. We can know only so many people by name: Nobel awardees, composers like Bach and Dvořák. Our own brain capacity for knowledge makes us concentrate on the top 100s. While the 100k top scientists do matter scientifically, they do not matter culturally.
- marcusverus 5y ago
  > Even for something as clearly and uncontroversially genetically influenced as running ability, the group differences only really matter for the top 100 people out of 7 billion. Sure, but there is no social contention arising from racial differences in the world's fastest runners. What gets us in trouble are the the racial differences in high-prestige jobs. To really understand the impact of differences in distributions, you need to look at some numbers. The excerpt below comes from Douglas Murray's new book. I am not going to attempt to convince you that racial differences in IQ exist. But he makes a strong case that if such racial differences did exist, the impact on the distribution of high-prestige jobs would be absolutely enormous: "...Why are there so few minorities in these high-prestige jobs? It's a numbers game in which the odds are against a Latin (he uses geographic indicators in place of the colloquial 'black' and 'white', and 'latin' in place of 'hispanic') achieving one of those positions are high and the odds against an African are prohibitive, even if we assume that there is no racism whatsoever among the employers for high-prestige jobs. To illustrate, I'll use the cohort of young Americans ages 25-29, the age at which the potential candidates for such jobs are coming out of law schools, medical schools, business schools, and graduate STEM departments. In 2019, there were 23.2 million Americans in that age group. About 228,000 people in that age group can be expected to have IQs of 135 or over. The racial distribution of Americans ages 25-29 in 2019 was more multiracial than among the older population. Only 54% were European while 20 percent were Latin, 15 percent were African, and 6 percent were Asian. But that reduced dominance of Europeans in total population doesn't make a lot of difference in the 135+ pool. Employers seeking these exceptionally intelligent young adults were choosing from a pool that contained about 2,800 Africans [~1%] and 9,500 Latins [~3%] compared to 50,700 Asians [*~22%*] and 160,000 Europeans [~70%]." [~3% being 'other'] So, while the marginal IQ differences between individuals do not matter not one whit, when considered at the population level, these differences can hardly be ignored.
  - tsimionescu 5y ago
    > But that reduced dominance of Europeans in total population doesn't make a lot of difference in the 135+ pool. Employers seeking these exceptionally intelligent young adults were choosing from a pool [...] This is one mistake, and the danger of using IQ as anything more than a trend indicator. Employers don't need 'people with an IQ over ${ARBITRARY_CUTOFF}'. They need 'people who can do X and Y'. And IQs over 100 quickly stop having significant predictive power for mental abilities other than IQ tests. Even exams which focus on pass/fail of an aptitude rather than excellence show little 'racial' variance in general. The numbers in that paragraph also seem entirely fantastical to me. It basically claims not only that there are IQ differences between the "races", but that they are huge. Finally, and most importantly, the whole set of observations is meaningless in the way it is used. Even granted that there are such huge differences in IQ between various "racial" groups, the social justice response won't change - we need to focus on eliminating the social and environmental causes at the root of these differences. Murray could claim that the differences are genetic, but I am very curious what is the scientific methodology by which he defines these groups - how much genetic testing did he do determine whether one particular person is 'European', 'African', 'Latin', or 'Asian'? What genes in particular was he looking for that make someone European? (For that matter, are Italians and Spanish people "European" or "Latin"?? Is Eva Longoria, whose family came from Spain to America with a royal land grant 400 years ago "European" or "Latin"?).
    - marcusverus 5y ago
      > This is one mistake, and the danger of using IQ as anything more than a trend indicator. Employers don't need 'people with an IQ over ${ARBITRARY_CUTOFF}'. They need 'people who can do X and Y'. Even exams which focus on pass/fail of an aptitude rather than excellence show little 'racial' variance in general. The scientific literature is very clear on this point--IQ is predictive of job performance. If you need people who can do X & Y, IQ is the best predictor[0] of whether a given individual will be able to do X & Y. > And IQs over 100 quickly stop having significant predictive power for mental abilities other than IQ tests. I would love to see a source for this claim. > The numbers in that paragraph also seem entirely fantastical to me. It basically claims not only that there are IQ differences between the "races", but that they are huge. The numbers in the paragraph are partly due to differences in IQ, and partly due to the nature of distributions. IIRC, Murray claims a mean IQ difference of ~15 points between 'Africans' and 'Asians'. Here's an image[0] of two distributions with an equivalent mean difference. Note how, the farther you get from the center, the greater the proportion of red:black. At 1 standard deviation, the ratio is 3:2, at 2sds, the ratio is 4:1! > Even granted that there are such huge differences in IQ between various "racial" groups, the social justice response won't change - we need to focus on eliminating the social and environmental causes at the root of these differences. The point of this conversation is that, if these differences are due to genetics, then no amount of 'eliminating causes' will address the root cause of these differences. > Murray could claim that the differences are genetic, but I am very curious what is the scientific methodology by which he defines these groups - how much genetic testing did he do determine whether one particular person is 'European', 'African', 'Latin', or 'Asian'? What genes in particular was he looking for that make someone European? (For that matter, are Italians and Spanish people "European" or "Latin"?? Is Eva Longoria, whose family came from Spain to America with a royal land grant 400 years ago "European" or "Latin"?). All the data he uses relies on self-reporting of race. Self-reporting is always how race is assigned. Self-reporting is how we know that certain minorities are 'underrepresented' in certain jobs--folks self-report their race when hired. It's the same way we know that there are differences in incomes and in educational achievement between races--self-reporting. You can doubt the veracity of self-reporting of race, but that doubt would have to apply to all data with a racial dimension. And, anyway, if race was some sort of fuzzy, meaningless construct (as you appear to be implying), then we would expect there to be little correlation between race and anything, let alone IQ. And yet the data show these differences (according to Murray). Longoria would be 'Latin' (Hispanic). The whole race vs. ethnicity thing is weird, but he explains it well in the book. [0]https://www.jstor.org/stable/20182140 https://www.jstor.org/stable/20182140 [1]https://tamino.files.wordpress.com/2015/07/shiftmean.jpeg https://tamino.files.wordpress.com/2015/07/shiftmean.jpeg

by esyir 5y ago

tsimionescu 5y ago

But it's a bad idea to compare the world's fastest runners with people getting high paying jobs - that's the main point of the GP's comment. Even for something as clearly and uncontroversially genetically influenced as running ability, the group differences only really matter for the top 100 people out of 7 billion. If you look at top 100k runners, the group differences are almost entirely gone. Even more, the top 100k runners are probably within a second or two of each other - entirely irrelevant if you were to hire based on a bar of performance, and not some elitist concept of hiring the very best possible.

inglor_cz 5y ago
Socially, who makes it to the top hundred matters a lot. We can know only so many people by name: Nobel awardees, composers like Bach and Dvořák. Our own brain capacity for knowledge makes us concentrate on the top 100s. While the 100k top scientists do matter scientifically, they do not matter culturally.
marcusverus 5y ago
> Even for something as clearly and uncontroversially genetically influenced as running ability, the group differences only really matter for the top 100 people out of 7 billion. Sure, but there is no social contention arising from racial differences in the world's fastest runners. What gets us in trouble are the the racial differences in high-prestige jobs. To really understand the impact of differences in distributions, you need to look at some numbers. The excerpt below comes from Douglas Murray's new book. I am not going to attempt to convince you that racial differences in IQ exist. But he makes a strong case that if such racial differences did exist, the impact on the distribution of high-prestige jobs would be absolutely enormous: "...Why are there so few minorities in these high-prestige jobs? It's a numbers game in which the odds are against a Latin (he uses geographic indicators in place of the colloquial 'black' and 'white', and 'latin' in place of 'hispanic') achieving one of those positions are high and the odds against an African are prohibitive, even if we assume that there is no racism whatsoever among the employers for high-prestige jobs. To illustrate, I'll use the cohort of young Americans ages 25-29, the age at which the potential candidates for such jobs are coming out of law schools, medical schools, business schools, and graduate STEM departments. In 2019, there were 23.2 million Americans in that age group. About 228,000 people in that age group can be expected to have IQs of 135 or over. The racial distribution of Americans ages 25-29 in 2019 was more multiracial than among the older population. Only 54% were European while 20 percent were Latin, 15 percent were African, and 6 percent were Asian. But that reduced dominance of Europeans in total population doesn't make a lot of difference in the 135+ pool. Employers seeking these exceptionally intelligent young adults were choosing from a pool that contained about 2,800 Africans [~1%] and 9,500 Latins [~3%] compared to 50,700 Asians [*~22%*] and 160,000 Europeans [~70%]." [~3% being 'other'] So, while the marginal IQ differences between individuals do not matter not one whit, when considered at the population level, these differences can hardly be ignored.
- tsimionescu 5y ago
  > But that reduced dominance of Europeans in total population doesn't make a lot of difference in the 135+ pool. Employers seeking these exceptionally intelligent young adults were choosing from a pool [...] This is one mistake, and the danger of using IQ as anything more than a trend indicator. Employers don't need 'people with an IQ over ${ARBITRARY_CUTOFF}'. They need 'people who can do X and Y'. And IQs over 100 quickly stop having significant predictive power for mental abilities other than IQ tests. Even exams which focus on pass/fail of an aptitude rather than excellence show little 'racial' variance in general. The numbers in that paragraph also seem entirely fantastical to me. It basically claims not only that there are IQ differences between the "races", but that they are huge. Finally, and most importantly, the whole set of observations is meaningless in the way it is used. Even granted that there are such huge differences in IQ between various "racial" groups, the social justice response won't change - we need to focus on eliminating the social and environmental causes at the root of these differences. Murray could claim that the differences are genetic, but I am very curious what is the scientific methodology by which he defines these groups - how much genetic testing did he do determine whether one particular person is 'European', 'African', 'Latin', or 'Asian'? What genes in particular was he looking for that make someone European? (For that matter, are Italians and Spanish people "European" or "Latin"?? Is Eva Longoria, whose family came from Spain to America with a royal land grant 400 years ago "European" or "Latin"?).
  - marcusverus 5y ago
    > This is one mistake, and the danger of using IQ as anything more than a trend indicator. Employers don't need 'people with an IQ over ${ARBITRARY_CUTOFF}'. They need 'people who can do X and Y'. Even exams which focus on pass/fail of an aptitude rather than excellence show little 'racial' variance in general. The scientific literature is very clear on this point--IQ is predictive of job performance. If you need people who can do X & Y, IQ is the best predictor[0] of whether a given individual will be able to do X & Y. > And IQs over 100 quickly stop having significant predictive power for mental abilities other than IQ tests. I would love to see a source for this claim. > The numbers in that paragraph also seem entirely fantastical to me. It basically claims not only that there are IQ differences between the "races", but that they are huge. The numbers in the paragraph are partly due to differences in IQ, and partly due to the nature of distributions. IIRC, Murray claims a mean IQ difference of ~15 points between 'Africans' and 'Asians'. Here's an image[0] of two distributions with an equivalent mean difference. Note how, the farther you get from the center, the greater the proportion of red:black. At 1 standard deviation, the ratio is 3:2, at 2sds, the ratio is 4:1! > Even granted that there are such huge differences in IQ between various "racial" groups, the social justice response won't change - we need to focus on eliminating the social and environmental causes at the root of these differences. The point of this conversation is that, if these differences are due to genetics, then no amount of 'eliminating causes' will address the root cause of these differences. > Murray could claim that the differences are genetic, but I am very curious what is the scientific methodology by which he defines these groups - how much genetic testing did he do determine whether one particular person is 'European', 'African', 'Latin', or 'Asian'? What genes in particular was he looking for that make someone European? (For that matter, are Italians and Spanish people "European" or "Latin"?? Is Eva Longoria, whose family came from Spain to America with a royal land grant 400 years ago "European" or "Latin"?). All the data he uses relies on self-reporting of race. Self-reporting is always how race is assigned. Self-reporting is how we know that certain minorities are 'underrepresented' in certain jobs--folks self-report their race when hired. It's the same way we know that there are differences in incomes and in educational achievement between races--self-reporting. You can doubt the veracity of self-reporting of race, but that doubt would have to apply to all data with a racial dimension. And, anyway, if race was some sort of fuzzy, meaningless construct (as you appear to be implying), then we would expect there to be little correlation between race and anything, let alone IQ. And yet the data show these differences (according to Murray). Longoria would be 'Latin' (Hispanic). The whole race vs. ethnicity thing is weird, but he explains it well in the book. [0]https://www.jstor.org/stable/20182140 https://www.jstor.org/stable/20182140 [1]https://tamino.files.wordpress.com/2015/07/shiftmean.jpeg https://tamino.files.wordpress.com/2015/07/shiftmean.jpeg