Let’s be clear. Racial groupings are really very significant pieces of evidence. There’s huge amounts of genetics that correlates, huge amounts of culture that correlates, huge amounts of wider environment that correlates. It would be frankly astonishing if things like IQ, reaction time, hight, life expectancy, and rates of disease didn’t also correlate.
Culture and environment are not race. Therefore, if you’re studying race, those influences should be taken out of your scientific experiment. It’s extremely difficult to remove things like culture and environment from a study on IQ. The fact that so much is correlated with it doesn’t mean the results of studies intended to determine racial differences are significant so much as it means they’re a tangled mess of cause and effect which we likely haven’t sorted out adequately.
Why on earth do we want there to be no such correlation with IQ.
A. We don’t want black people to suffer needlessly.
B. We don’t want to encourage ourselves and others to be prejudiced against people when, regardless of what the average African’s IQ is, it is still both logically incorrect (hasty generalization) and ethically wrong to prejudge individual Africans. However, knowing how humans behave, we figure that if people believe Africans have lower IQs, that will result in an increase in prejudice.
We’re very happy to say there’s a correlation between race and hight, between race and life expectancy, between race and disease, between race and income. Why not race and IQ? Why do we want that to be false?
Actually, I bet some people are not happy saying that there are correlations there. This is one of those notions you might want to double check.
We don’t want to encourage ourselves and others to be prejudiced against people when, regardless of what the average African’s IQ is, it is still both logically incorrect (hasty generalization) and ethically wrong to prejudge individual Africans.
So what you’re saying is that it’s ethically wrong to use Bayesian reasoning.
It is dangerous to be half a rationalist. This applies to groups as well as individuals. No matter how good your process for arriving at beliefs, it is indeed unethical to go around spreading those beliefs to people that will predictably misunderstand and misuse them.
No matter how good your process for arriving at beliefs, it is indeed unethical to go around spreading those beliefs to people that will predictably misunderstand and misuse them.
Funny how the only people to make that argument tend to be people who don’t want to believe the beliefs in question, but have out of ways to ignore the evidence.
On a less meta level: what kind of “misunderstand and misuse” do you think is going to “predictably” happen.
Funny how the only people to make that argument tend to be people who don’t want to believe the beliefs in question, but have out of ways to ignore the evidence.
I, for one, do believe that the average African has lower IQ than the average European but don’t go around telling that to the wrong people.
On a less meta level: what kind of “misunderstand and misuse” do you think is going to “predictably” happen.
Underestimating how much the evidence race provides about an individual’s IQ can be screened off by other evidence about the individual, due to the confirmation bias and similar.
In the interest of clarity: I am not at all sure how to proceed in this particular case. History makes me wary of departing from the current Schelling point of assuming everybody is equal, but that’s not my point.
I am saying that a course of action based on Bayesian reasoning has no special immunity to being ethically wrong, and it is those actual results that are worth worrying about, not merely the epistemology.
If you intend to use an African prejudgment heuristic like 1 (below) rather than reacting as if you’ve done an equation that takes into account other relevant data like 2 (below), then I think your probability equation needs an upgrade.
1) African prejudgment heuristic: “The IQ test(s) said African’s IQs are lower than those of whites, therefore this specific African individual is likely to be relatively stupid compared to my white friends.”
2) Reasoning that Takes Relevant Data Into Account: “The IQ test(s) said African’s IQs are lower than those of whites. However, there are known flaws with IQ tests such as cultural bias, so that figure might be wrong. Most published research findings are false (PLOS Medicine), so I should apply healthy skepticism to all the research I read. This is not likely to be an accurate piece of data to use as a Bayesian prior. Once I’ve decided on a prior to use, I should then adjust for other relevant data (things I know about the specific individual).
Then, if one wants to behave rationally after one has decided what to believe, I think one must continue by thinking something like this:
“Considering things like...
A. …the fact my prior is likely to be inaccurate (there isn’t an accurate one for this subject as far as I’m aware)...
B. …the fact that even if the IQ study is correct and the IQ test it used was accurate, there’s a decent chance (25%) that this specific individual has an IQ above the African average—meaning I need to avoid the logical fallacy called hasty generalization...
C. …the risk of lost utility via damaging this individual’s reputation, emotional health or opportunities for success by pre-judging them...
D. …the risk that this makes for bad social signaling and witnesses may retaliate against me with one or more forms of social rejection if I pre-emptively treat an African like an idiot...
…do I really want to treat this person as if they are less intelligent?”
I think you may have reacted to my “I hope it is false.” statement or my “it’s ethically wrong to prejudge individual Africans” statement—but that shouldn’t matter to your probability calculation. What should matter is to get an accurate idea of reality. Along with saying other things, I also provided other factors which are relevant, as credible sources can confirm. If one wants to be a good Bayesian probabilist, after one specifies some prior probability, one must then update it in the light of new, relevant data. [1] This situation where you focused on one part of my comment and ignored the rest reminds me of those math problems where there’s an irrelevant statement thrown in just to distract you. I of course did not intend to distract you, but since you seem to think that Bayesian reasoning in this case means ignoring all the other data I presented, it appears that you have skipped the parts of the process where you ensure an accurate prior and update your prior with new, relevant data.
Life is really, really complicated. I doubt it’s ever wise to just grab a prior and run with it and I certainly hope that you do not reason this way.
Paulos, John Allen. The Mathematics of Changing Your Mind, New York Times (US). August 5, 2011; retrieved 2011-08-06
Reasoning that Takes Relevant Data Into Account: “The IQ test(s) said African’s IQs are lower than those of whites. However, there are known flaws with IQ tests such as cultural bias, so that figure might be wrong. Most published research findings are false (PLOS Medicine), so I should apply healthy skepticism to all the research I read. This is not likely to be an accurate piece of data to use as a Bayesian prior.
This reads like a classic case of motivated cognition. You don’t want to believe the conclusion; therefore, you selectively look for potential flaws then declare that there’s still a chance. The reason I believe the connection between race and intelligence is not just because of the tests but because more or less every relevant aspect of reality (e.g., the statistic on race and crime, the nearly complete lack of blacks in intelligence intensive fields, e.g., math, programing, the state of majority black countries) looks the way one would expect it to look if the connection existed.
Once I’ve decided on a prior to use, I should then adjust for other relevant data (things I know about the specific individual).
Who is claiming otherwise?
A. …the fact my prior is likely to be inaccurate (there isn’t an accurate one for this subject as far as I’m aware)...
Yes there is. You just don’t want to believe it exists.
B. …the fact that even if the IQ study is correct and the IQ test it used was accurate, there’s a decent chance (25%) that this specific individual has an IQ above the African average—meaning I need to avoid the logical fallacy called hasty generalization...
Actually the chance of this particular black being above the African average is 50% (more if I condition on the fact that he is in the USA). The probability that he is above the white average is significantly less. The probability that he is above some high cutoff is can be even lower.
C. …the risk of lost utility via damaging this individual’s reputation, emotional health or opportunities for success by pre-judging them...
This is a general argument against using evidence of any kind.
D. …the risk that this makes for bad social signaling and witnesses may retaliate against me with one or more forms of social rejection if I pre-emptively treat an African like an idiot...
This is potentially a problem, although here the problem is arguably with the witnesses behaving irrationally and/or participating in out of control signaling arms races than with the strategy itself.
This reads like a classic case of motivated cognition.
Did you stop to make distinction between me being influenced by motivated cognition and alternate explanations like:
Me seeing significant flaws in data that would otherwise support your conclusion. Part of this may be that I’ve spent a significant amount of time reading about IQ and giftedness and I have learned that there are a lot of pitfalls to doing IQ related research.
Me simply being unaware of relevant data. (This might be the case in the event that the people who supplied my data were influenced by motivated cognition or confirmation bias.)
You seeing motivated cognition in my words because of being influenced by motivated cognition yourself?
The reason I believe the connection between race and intelligence is not just because of the tests but because more or less every relevant aspect of reality (e.g., the statistic on race and crime, the nearly complete lack of blacks in intelligence intensive fields, e.g., math, programing, the state of majority black countries) looks the way one would expect it to look if the connection existed.
There is an alternate explanation for those which does not have the same issues that IQ tests and studies have: The effects of slavery and prejudice. We are certain that slavery and prejudice has influenced them, and that it has existed for a long time. To know this, one must only look at the KKK or investigate the history of black enslavement. Imagine a third world country. Imagine that an equal proportion of those inhabitants are removed and used as slaves. Imagine an equal proportion of them dying. Imagine that they’re freed, but all of them—not some but all—are freed into a situation of extreme poverty where they don’t even own a home or have the ability to read. Many still aren’t being taught to read. Consider also that even though there have been advances in medicine, poverty means you can’t afford health insurance or medical treatments. Don’t think that disability and chronic illness are uncommon—they’re not. Not even in America. They’re probably especially common for the poor. Don’t think that severe worker abuse ended with slavery, either—do some research on sweatshops in America sometime. Now take into account the effects of stress, and the human element—how those effects can compound into things like mental illnesses and drug addictions. Would you predict that the majority of these people who started out with literally nothing and without even the education to read would manage to avoid pitfalls like disability, mental illness, drug addiction and sweatshops and carve an opportunity to excel out of poverty and ignorance over the course of 150 years? I would not expect that. I would expect most of them to have fared poorly.
I don’t see a good way to tell the difference between a low IQ score due to actually being less intelligent versus a low IQ score due to nurture-related reasons such as the following:
Improper nutrition due to poverty.
Lack of education.
The effects of extreme stress (How are you supposed to focus on an IQ test when you’ve just been threatened by a gang?)
Suffering from medical conditions (these can cause memory symptoms, brain fog, and fatigue), mental conditions or drug addictions.
Having been parented by people that were mentally or physically ill, severely stressed, or addicted.
Cultural differences that cause arbitrary communication issues during testing.
The psychological effects of prejudice (may influence things like self-esteem and locus of control or result in learned helplessness, etc.)
If you want to attribute the IQ scores to race, not poverty or circumstances, then there needs to be a good way to distinguish between nurture and nature as a cause for low IQ scores. Do you have one?
Yes there is. You just don’t want to believe it exists.
If it’s true I want to believe it. However, it’s hard to believe it exists without a citation. Do you have one?
Actually the chance of this particular black being above the African average is 50% (more if I condition on the fact that he is in the USA).
I respect you more for being able to say something that supports my view better than it does yours. +1 karma for that. I still think the number is 25%, however I do not view this as a key point in our disagreement, so I will leave it at that.
This is a general argument against using evidence of any kind.
You appear to have taken that statement as an argument regarding what to believe. It was not. I deliberately put that part after the section where I was discussing deciding what to believe, and put it under “if one wants to behave rationally”.
I don’t see a good way to tell the difference between a low IQ score due to actually being less intelligent versus a low IQ score due to nurture-related reasons such as the following:
As the saying goes, “Life is an IQ test.” The predictive ability of IQ on income (and most other statistics of interest) is very similar for each race, which suggests that differences in measured IQ scores map onto differences in life outcomes. To the extent that a medical condition makes someone test poorly, it generally also makes them live poorly. (There are a handful of prominent exceptions to this- like dyslexia- which don’t significantly impact the main point.)
Now, where those IQ differences come from in the first place is an interesting question. Many people have looked at it, and the consensus answer for individual IQ differences is “somewhere between 50% and 80% of it is genetic,” and the extrapolation from that to group IQ differences is somewhat controversial but seems straightforward to me. A perhaps more interesting question is “what knobs do we have to adjust those IQ differences?”
However, it’s hard to believe it exists without a citation. Do you have one?
Here’s Jason Richwine explaining that all serious scientists have agreed on the basics of IQ for decades, and the media is completely mistaken on the state of reality and scientific consensus.
I still think the number is 25%, however I do not view this as a key point in our disagreement, so I will leave it at that.
It seems pretty relevant to me, because it looks like basic statistical innumeracy on your part, unless you think the IQ distribution of African Americans is tremendously skewed such that the mean intelligence is the 75th percentile of intelligence, rather than the 50th percentile like it would be in a symmetric distribution. (Or you think that the African average is higher than the African American average, which is very much not the case.)
That’s a saying? Are there also sayings “Life is a test of height.”, “Life is a test of immune system efficiency” and “Life is a test of facial symmetry”? We may as well round out the set. Anything of comparable test significance that I missed? Oh! “Life is a test of breast perkiness” and “Life is a test of capacity for situation-appropriate violence.”
(I actually agree with the point of the rest of the paragraph.)
Are there also sayings “Life is a test of height.”, “Life is a test of immune system efficiency” and “Life is a test of facial symmetry”? We may as well round out the set.
Those may be qualitatively similar but I would suggest they are quantitatively different. I would be surprised if facial symmetry did not correlate with income, health, social status, and so on, but I would expect the correlation to be much lower than the correlation with IQ. The saying means that most metrics of life success are moderately highly g-loaded, and so it makes sense that IQ correlates positively with basically everything good.
Those may be qualitatively similar but I would suggest they are quantitatively different.
I think you either drastically overestimate the impact of IQ or underestimate the predictive value of those other factors of life success—particularly in environments different to those experienced by the white male nerd class of first world countries. Regardless, the paragraph with that alleged saying redacted would be far more persuasive than the one with it included. It strongly undermines the credibility of your point. I currently agree with you despite the opening sentence, not because of it.
I think you either drastically overestimate the impact of IQ or underestimate the predictive value of those other factors of life success
Perhaps we should switch to numbers to make it easier to communicate. I haven’t looked at any numbers recently, so I don’t expect these guesses to be particularly accurate, but I’d guess IQ correlates .2 to .6 with most interesting measures of life success, and height correlates around 0 to 0.1 (after controlling for IQ). I’d guess facial symmetry has correlation >0.3 with other health-related things, a correlation around 0.4 with social things, and a correlation below 0.3 with the rest.
If you have a source handy that estimates these sorts of correlations, I’d like to see it, but I don’t think it’s important enough to spend time hunting something down.
I would be very surprised if IQ correlates at .6 with, say, wealth or income. Parental wealth and income possibly correlate no more than 0.5 to childrens’ incomes, and it would be frankly remarkable for IQ to be (1) transmitted intergenerationally to a large degree, and (2) more closely correlated to financial outcomes than one’s parent’s financial outcomes, since your parents often give you not only your genes, but your inheritance/early support, financial assumptions, and first set of career contacts.
Intelligence: Knowns and Unknowns claims that parental SES and IQ are correlated at .33; that parental SES explains one third of social status variance (which implies r=.58) and one fifth of income variance (r=.45); that IQ explains about a quarter of the social status variance (r=.5) and a sixth of the income variance (r=.4), and that also correcting for parental SES reduces the predictive ability of IQ by a quarter. I would expect more recent numbers to be broadly similar.
Thanks for the link. It’s a nice summary of the state of research a few years back, and anybody who’s interested in the topic should read it.
It is probably even more interesting to me because it tacks pretty hard away from the conclusions some people have drawn in this thread. The authors clearly did not believe that the well-attested differences in IQ testing across ethnic groups could be ascribed to genetic factors.
The authors clearly did not believe that the well-attested differences in IQ testing across ethnic groups could be ascribed to genetic factors.
I think that there’s not definitive evidence on the subject, even today. The definitive evidence would be if we knew the specifics of the causal link between genetics and IQ and had representative genetic samples from different racial groups, so we could look at the prevalence of various IQ genes and calculate what we’d estimate the average racial IQ to be for various groups from their genes. That’d give us an estimate of the genetic factors, and the difference between that and measured IQ would give us an estimate of the environmental factors.
My read of the field, though, is that the majority of the evidence points to the majority of the difference between racial groups being explained by genetic factors, and the trend has been that the hereditarian position has been growing more solid over time, especially as more and more people have been sequenced.
For example, African Americans represent a significant observational data source on ancestry and intelligence, since while the average African American has 80% African ancestry, that percentage can vary significantly from person to person. Intelligence: Knowns and Unknowns reports that a study that used blood groups to determine ancestry failed to find a correlation between European ancestry and IQ; more recent research claims to have found a correlation between European ancestry and IQ and controlled for the impact of skin color.
I haven’t looked at any numbers recently, so I don’t expect these guesses to be particularly accurate, but I’d guess IQ correlates .2 to .6 with most interesting measures of life success, and height correlates around 0 to 0.1 (after controlling for IQ).
Do you similarly control the IQ finding for height? For obvious reasons that is necessary for consistency.
I’d certainly expect (based on loose memories of studies encountered) height to predict more than ‘0 to 0.1’ and IQ to predict a heck of a lot less than 0.6, especially when considering populations that are not limited to first world nations. I wish it were otherwise, naturally. I’d love the world to be more biased in favour of my own highest stat.
In any case, I would consider it legitimate to readers who encounter “Life is an IQ test” delivered in response to the quoted context to be sufficient evidence that the comment is not worth reading and downvoting and ignoring it. It was only because I was more patient that normal that I bothered to read further and found you had a good point hidden in the detail. Do with that information what you will.
I’d love the world to be more biased in favour of my own highest stat
You can re-allocate some of that to Charisma if you work really hard (stand-up comedy is a learned skill) and if you have a British or Australian accent you get +1 just by coming to North America and talking. Provided you haven’t already maxed it, Strength is highly trainable, as are Dexterity and Constitution. Even HP, up till about 30 when bone density stops accumulating. Wisdom is extremely trainable, and there’s some evidence the world’s biased that direction, so I’d throw points there when in doubt.
Provided you haven’t already maxed it, Strength is highly trainable
In fact, even those who have reached their maximum strength potential can increase it by using the highly potent Potion of Potential Strength.
Even HP, up till about 30 when bone density stops accumulating.
HPs are primarily determined by Constitution (somewhat trainable) but in many worlds (including the real one) there is also a bonus from Strength stat. Better developed muscles are useful for absorbing damage non-critically.
Perhaps the most important stat is actually Willpower, which is also somewhat trainable and can be buffed with items and hired allies.
You can re-allocate some of that to Charisma if you work really hard (stand-up comedy is a learned skill) and if you have a British or Australian accent you get +1 just by coming to North America and talking.
Well spotted on the ‘favour’ usage! Yes, I’ve noticed a bonus there (my current girlfriend is North American), presumably the accent helped.
Do you similarly control the IQ finding for height? For obvious reasons that is necessary for consistency.
If you do a linear regression with both IQ and height as inputs, this automatically separates their (linear) effects.
But I’m not sure about the general point, because IQ is an estimate of a factor and height is an estimate of a single variable. My impression is that when you are looking for the individual effect of a component of a hidden factor, then you want to control for the factor before measuring the effect of the component, but when measuring the effect of the hidden factor you don’t control for the components. But height isn’t a component of the IQ calculation, though it does appear to be weakly related to intelligence.
I’d certainly expect (based on loose memories of studies encountered) height to predict more than ‘0 to 0.1’ and IQ to predict a heck of a lot less than 0.6, especially when considering populations that are not limited to first world nations.
I’m also having trouble remembering if the 0.05 number I remembered was an r or r^2, which would significantly impact the range. I’m finding rs of about .2 for the correlation between height and intelligence, and rs of about .2 for the correlation between height and income without controlling for intelligence. Haven’t found anything yet that does control for it. (IQ correlation with income, as mentioned in a cousin comment, is about .4.)
In any case, I would consider it legitimate to readers who encounter “Life is an IQ test” delivered in response to the quoted context to be sufficient evidence that the comment is not worth reading and downvoting and ignoring it.
It’s still not clear to me what about the saying you find objectionable. The best guess I have is that you took it to mean that the g-loading of life success was comparable to Raven’s, when I meant that they were g-loaded at all. The heart of that paragraph:
To the extent that a medical condition makes someone test poorly, it generally also makes them live poorly.
seems to me like a fair explanation of the saying, and why it’s confused to think about nurture factors influencing intelligence as separate from actual intelligence.
As a stand-alone statement, I would probably leave this alone. But as a response to “What about nurture”, the first thing that comes to mind is:
Has Vaniver adequately corrected for the just world fallacy?
The predictive ability of IQ on income (and most other statistics of interest) is very similar for each race, which suggests that differences in measured IQ scores map onto differences in life outcomes.
Ok, that’s interesting, but it does nothing to rule out nurture factors that would impact both IQ and income.
Many people have looked at it, and the consensus answer for individual IQ differences is “somewhere between 50% and 80% of it is genetic”
I agree that IQ is mostly genetic and that IQ does seem to correlate with a lot of factors. I’m not saying IQ does not exist. What I am saying is that, specifically when it comes to black people, there are other factors that are definitely influencing performance and IQ scores. Therefore I reject claims about IQ and race that haven’t controlled for known factors.
Here’s Jason Richwine explaining that all serious scientists have agreed on the basics of IQ for decades
Actually, when I read that section (it starts with “What scholars”), I parsed it like this:
Jason explicitly says that there’s a scientific consensus on many issues that seem controversial to journalists.
Jason states that virtually all psychologists (not scientists) believe there is a general mental ability factor (That he’s not saying is specifically connected to race).
Then, without qualifying these statements with anything along the lines of “most x believe”, he states: “In terms of group differences, people of northeast Asian descent have higher average IQ scores than people of European lineage, who in turn have higher average scores than people of sub-Saharan African descent.”
I will not assume that this sequence of claims means that the group differences statement is also something scientists have a consensus about. If I did, that would be a non-sequitur.
Also, below that, he writes:
“It is possible that genetic factors could influence IQ differences among ethnic groups, but many scientists are withholding judgment until DNA studies are able to link specific gene combinations with IQ.”
This is where I stop reading the article because it is clear to me that it does not say “there’s a scientific consensus that there’s a link between race and IQ”. If you have a credible source for that claim, I’ll be curious about it. No more Politico articles please.
It seems pretty relevant to me
It might or might not have been an error. In any case, I’m not going to go digging for that right now because I still think knowing the percentage is irrelevant to the current point. Whether the figure is 50% or 25%, it is still true that a significant proportion of people will have an IQ above average and therefore it would be hasty generalization to assume that a person of a certain group was an idiot. That is one way in which the exact number is irrelevant. However, that point about hasty generalization is much more irrelevant at this particular moment because we haven’t even decided on a prior, let alone have we got a decent posterior—so the step where we have a concern about making a hasty generalization based on our probabilities should be in this disagreement’s future. If it becomes relevant, I will dig around, but not right now.
Ok, that’s interesting, but it does nothing to rule out nurture factors that would impact both IQ and income.
It’s not clear to me why you would be interested in nurture factors. There are two things going on here: the ability of IQ to measure intelligence, and the historical causes of intelligence.
With the exception of disorders that prevent people from testing well without significantly impacting life outcomes, the historical causes of intelligence don’t appear to have much to do with the ability of IQ to measure intelligence. A nurture factor (like, for example, being breastfed as a child or being struck on the head) actually alters someone’s intelligence, and their intelligence influences both their test scores and their income.
What I am saying is that, specifically when it comes to black people, there are other factors that are definitely influencing performance and IQ scores. Therefore I reject claims about IQ and race that haven’t controlled for known factors.
Again, this looks like it’s mixing up the historical causes and the predictive ability. If the predictive ability is the same independent of race (it is), then it doesn’t matter why the racial IQ averages are the way they are. What we would need to show to discount IQ measurements is that the IQ measurements are not as predictive for members of one race than another.
As an example of a real bias like this, girls tend to get better grades than standardized test scores alone would predict. In order to get an accurate estimate of what a girl’s grades would be from her standardized test scores, you need to adjust upwards because she’s a girl. Symmetrically, boys score better than one would expect from their grades, and so when predicting scores one needs to adjust upwards. When moving in the opposite direction, one would need to adjust downwards; a girl’s grades overestimate her standardized test scores. But note that this doesn’t mean we throw out the data- it’s still predictive! We just adjust it the correct quantitative amount.
Now, do we know the historical causes of that effect? I’m not familiar with that field, but it seems like there are lots of plausible theories that probably have support. Even without knowing the causes, though, we can use our estimates of the size of the effect in order to predict more accurately.
virtually all psychologists (not scientists)
What would you call a scientist who studies intelligence? (I suppose I should also make clear that by “serious” I mean a scientist speaking confidently in their field of expertise.)
Then, without qualifying these statements with anything along the lines of “most x believe”, he states
One does not say “most scientists believe that hydrogen has one proton,” one says “hydrogen has one proton.”
If you have a credible source for that claim, I’ll be curious about it. No more Politico articles please.
Here’s the APA report he references. The group means section starts on page 16.
In general, though, asking for citations like this is really frustrating, because it doesn’t seem like the true rejection. The linked Richwine article referenced more serious sources that you could find if interested, and even if you didn’t notice that Googling “racial IQ averages” leads to this as the fourth hit, and if sufficiently motivated you could find the paper that journalist was writing about, and so on.
But if you’re not curious enough to seek out this information, and you don’t seem to have updated on the other information I’ve provided, what reason do I have to expect that the difference between my position and your position is that I have citations, and as soon as I share them you’ll adopt my position?
Whether the figure is 50% or 25%, it is still true that a significant proportion of people will have an IQ above average and therefore it would be hasty generalization to assume that a person of a certain group was an idiot.
Sure. When you think in distributions, an estimate generally comes with both a mode and a precision (or, relatedly, the standard deviation). Knowing someone is African American gives you an estimate with a mode of 85 and standard deviation of 15, which has a non-trivial but small chance of being over 120. Knowing someone got a 120 on a recent IQ test gives you an estimate with a mode slightly south of 120 and a standard deviation of probably 2-5, depending on the precision of the test.
In Africa? It so happens that the world is much bigger than the USA and the people in sub-Saharan Africa test for IQ pretty much the same as African-Americans.
then there needs to be a good way to distinguish between nurture and nature as a cause for low IQ scores. Do you have one?
Sure, you can control for wealth/economic status. Or you can go and test poor peasants in China and poor peasants in Africa. You seem to think that this is a white-vs-black US problem. It’s not. The highest-average-IQ large group of people is East Asians, like Han Chinese—not Caucasian whites.
I still think the number is 25%
I am curious—how do you figure out that in a distribution close to normal only 25% are higher than the mean?
Yeah. I suspect there are two reasons for that. First, malnutrition as a child can drive your IQ down and malnutrition is much more common in sub-Saharan Africa. And second, many African-Americans have some white ancestors. Look at Obama, for example—he self-identifies as African-American though only half his genes come from Africans.
Ethiopians have lots of Caucasian admixture too. (But once we know that both genes and environment play an important role, working out which fraction of the variance in IQs is due to each to within three significant figures doesn’t sound terribly interesting to me.)
But once we know that both genes and environment play an important role
How is that not self-evident given the edge cases (puppies going to human schools / children growing up in a sensory-deprivation tank, both not doing well on IQ tests)? Regarding the significant figures, we need to keep in mind those are to be interpreted as “this is how much of the variance factor X explains given a certain scenario”. They will vary across e.g. nations:
In a homogeneous environment (e.g. classless society, higher Gini-index), genes will acount for more of the variance than in a mixed environment with people of the same genetic makeup. IOW, as you e.g. change the school system, or who marries whom, so you change those relative weights of nature v. nurture.
You might say “well, given typical circumstances and typical gene pool variances”, but consider that the discussion is in any case comparing e.g. the US to sub-saharan Africa (or whereever), which absolutely cannot have the same relative weights for their respective nature versus nurture, unless the different gene variances in tribal societies and the different “school” environment somehow equalled out, a dubious proposition.
So the horrible European experience of being a Chinese colony has shocked the Caucasian population into scoring noticeably lower on the IQ tests than the Chinese?
In Africa? It so happens that the world is much bigger than the USA and the people in sub-Saharan Africa test for IQ pretty much the same as African-Americans.
It is far from hard to see how sub-saharan africa has been stripped and degraded over the centuries in a way that, say, rural China wasn’t.
And I am far from convinced that IQ, being an extremely culturally contextual measure, can be disentangled from modes of thought that lend themselves to abstract pattern analysis being more or less common in different traditions.
It is far from hard to see how sub-saharan africa has been stripped and degraded over the centuries in a way that, say, rural China wasn’t.
Centuries..? Central Africa was explored by Europeans (that is, the first Europeans appeared there) in the mid-XIX century. Most of the African countries were independent by late 60s early 70s of the XX century. Notable chunks like Ethiopia were never colonized (Ethiopia was occupied by Italy for a short time in 1930s and early 40s).
But anyway, it is hard for me to see. Can you provide data?
IQ, being an extremely culturally contextual measure
So your thesis is that the Chinese culture is very suited to “abstract pattern analysis”, the European culture is moderately suited and the Black culture… oh wait there is no single unified Black culture, so the Black population is unique in that its culture doesn’t matter but it was so similarly oppressed in Africa and the US that they ended up with similar reduced IQ. But the Ashkenazi Jews, though they were oppressed in Europe, ended up with a higher-than-Caucasian IQ.
I don’t see a good way to tell the difference between a low IQ score due to actually being less intelligent versus a low IQ score due to nurture-related reasons such as the following:
If your point is that it’s not clear to what extent the difference in intelligence is due to nature or nurture, I agree but would like to point out that for many applications it doesn’t matter.
When you’re deciding what to replace X with in the following statement, it most certainly does matter:
“X have a lower IQ on average.”
You can choose “People of African descent” or you can choose “People from poor backgrounds” or “People with serious health conditions” or “People with drug addictions” or any number of other things.
When attempting to determine how best to help a school in a black ghetto that is failing, and you’re choosing between spending money on remedial courses or on a school nutrition program, you will most certainly benefit from having this knowledge.
Conversely, I can’t think of any applications for which tying IQ to race is useful. Would you name three examples?
Also, I’m still interested in seeing the source that you believe is an accurate prior regarding race and IQ. Do you happen to have that information available?
Conversely, I can’t think of any applications for which tying IQ to race is useful.
If the results of the racial IQ studies are true, then that is very important because it disproves the doctrine of ethnic cognitive equality. Many people, especially in America, have this idea that all ethnic groups must have exactly equal average cognitive ability, and that if one or more ethnic groups perform below average on a test of aptitude, that is taken as strong evidence that the test is invalid and racially biased and thus cannot be used.
For this reason, many aptitude tests are severely restricted in their use since they are considered racist. This in turn would have a negative economic impact if these tests are actually valid, since employers and colleges are forced to use other, less effective means to vet candidates.
Actually, the legal rationale for restricting the use of such tests in certain kinds of hiring is not that they’re invalid. If you proved to the courts that they were “valid,” meaning an accurate reflection of crystallized intelligence/abstract reasoning/g/whatever, this would not undermine the central legal argument against them, which is that they produce disparate impacts on protected classes.
they produce disparate impacts on protected classes
There is the “business necessity” defense to disparate impact accusations. If the courts were to accept that IQ tests correctly reflect g/intelligence that defense will be much more applicable.
I’m pretty sure the courts have allowed that IQ-like tests are acceptable in many situations for many types of employment. It’s not a hypothetical. I guess I’m saying the question of the “validity of the tests” is a red herring, even if it’s an ideological hot potato. I think the main debate these days is not at all about the validity of the tests, it’s a debate over business necessity versus disparate impact.
I am not aware of that “main debate”. In the US, at least, political climate makes it impossible to discuss race issues in public. The courts, of course, have to decide these issues, but that hardly constitutes debate.
In the US, at least, political climate makes it impossible to discuss race issues in public.
Race issues are discussed constantly in the U.S. — often, but not always, under guises such as “immigration” or “the War on Drugs” or “failing schools”.
However, certain views are broadly discredited, for instance those which attribute or imply differences in the moral value of people’s lives on the basis of their race.
Well then, we’ve come to stating that the pertinent legal question is whether the use of IQ tests in hiring falls under “business necessity”. I don’t know of any answer to that other than “it depends”.
Though the issue of whether a job really requires high IQ is an interesting one...
I think the main debate these days is not at all about the validity of the tests, it’s a debate over business necessity versus disparate impact.
Which is still ridiculous. It’s been known for generations that IQ has a positive impact on basically every job, which should imply that the default is to assume business necessity for IQ tests.
Even if this were true, it would not follow that there is no countervailing incentive to remove barriers to employment for disadvantaged classes of people. Is it not possible that society has an interest in broad employment, especially among people disadvantaged by such tests? Two thoughts:
1) IQ tests have a history of being used deliberately to weed out applicants of certain races. This was not an incidental effect: it was the entire purpose of the test, much like literacy tests for voting. The odds of them being used this way again, were changes made in the law, seem extremely high.
2) It is interesting that LW sees so many rational arguments for policies that would give more resources to whites or Asians, especially white or Asian males with high test scores who may not have gone to college. While these arguments are phrased as both logical and obvious, LW rarely (ever?) entertains the easily constructed, similarly phrased arguments that would push resources away from LW’s typical membership. For example: “It’s been known for generations that physical strength has a positive impact statistically on outcomes in basically every sort of violent encounter, so as a default, in a world where couples and families could be attacked, people should assume a necessity for bigger, more muscular men as romantic partners.” Or how’s this: “It’s been known for generations that religious identification with the in-group eases working relationships and obviates friction over expressions of belief, so employers should as a default prefer employees share their religions.”
Is it not possible that society has an interest in broad employment, especially among people disadvantaged by such tests?
Very possible. I would take the Steve Sailer approach here, of acknowledging underlying differences and making the best of the situation. Let’s step away from race and just talk about tracking in schools- by the time someone is 12, we have a pretty good guess what their eventual social strata / broad kind of career will be.
In countries like Germany, they respond to this with different high schools- someone who will be a technician can go to a technical school, and someone who will be an engineer can go to an academic school. Both get work suited for their intellectual ability and interests, and so the first isn’t drowning and the second isn’t bored. (Relevant here is the finding that getting rid of shop classes increases the high school dropout rate in America- turns out that for an easily identifiable group of students, the primary benefit they get out of high school is a place to practice basic handyman skills!)
In the US, we get lunacy like “whether or not someone takes the first optional math class is a very strong predictor of whether or not they go to college. Let’s make that class mandatory for graduating high school!” which makes everyone involved worse off, as the students not pointed at college now find it more difficult to graduate high school.
For example: “It’s been known for generations that physical strength has a positive impact statistically on outcomes in basically every sort of violent encounter, so as a default, in a world where couples and families could be attacked, people should assume a necessity for bigger, more muscular men as romantic partners.”
I’m not sure this would see significant disagreement here on LW. The main response I would give is yes, but the preference is miscalibrated. Ceteris paribus, a stronger partner is likely to be better (assuming they aren’t prone to domestic violence), but my reflective preferences would give a weight to athleticism that’s orders of magnitude lower than the weight my attraction heuristics give athleticism. This mismatch seems to be because those heuristics were tuned in an era when the chance of being the victim (or beneficiary!) of violent crime was orders of magnitude higher than they currently are.
Is it not possible that society has an interest in broad employment, especially among people disadvantaged by such tests?
Of course it has. But the issue is that the society isn’t going to come out and say that—it will deliberately distort the map and make claims that are not true in reality.
The argument being made isn’t “50% of people are below median intelligence, we still need to and can productively employ them”, the argument is “we will pretend that all groups of people are exactly equally smart and if you say otherwise we’ll sue your ass into the ground”.
people should assume a necessity for bigger, more muscular men as romantic partners
Nope, not true since Mr.Colt made an equalizer :-) But I’ll agree that firearms training and ownership can be a reasonable plus in looking for a romantic partner. Well, unless his name is Pistorius...
Of course it has. But the issue is that the society isn’t going to come out and say that—it will deliberately distort the map and make claims that are not true in reality.
So, reason dictates that… “we” should shove our offended senses of intellectual consistency and naively understood “honesty” up our collective butt, and just do whatever helps people.
And we should absolutely not help people “equally”! Whatever you think of the abstract moral/political ideal of equality, in practical terms people’s circumstances in any society are so unequal that symmetrical treatment makes no sense. Any policy that does not identify the most vulnerable and marginalized groups and offer them targeted aid and protection is not “fair”, it’s not “impartial”… it’s basically a waste of resources, failing to seek out the greatest marginal utility for its subjects. So, ironically, it becomes a core left-wing idea that people should not be approached as identical or treated in an “equal” manner. Bam!
“In its majestic equality, the law forbids rich and poor alike to sleep under bridges, beg in the streets and steal loaves of bread.”—Anatole France on “legal equality”
I think your problem is that you’re so focused on the “fairest” way to divide a fixed set of goods, that you’re forgetting that the decisions in question also have a large effect on the amount of goods available.
“It’s been known for generations that religious identification with the in-group eases working relationships and obviates friction over expressions of belief, so employers should as a default prefer employees share their religions.”
That’s actually an interesting argument. I wouldn’t mind seeing it expanded, if you happen to have real numbers lying around.
Though some obvious confounders do come to mind: in a really diverse religious environment (like, for example, the Silicon Valley tech scene), you’re giving up quite a bit in talent if you recruit only from your co-religionists. And if you weight it less heavily, I’d be very surprised if the response looked linear: I wouldn’t expect a workplace that’s (say) 50% Christian with the rest split between atheists, Hindus, and Buddhists to be that much more harmonious than one with equal numbers of all of the above plus the odd Wiccan or Discordian. It might actually be worse under some circumstances, although this is rank speculation.
A lot of the current research focuses on “trust” inside groups. This is not exactly double-blinded climate controlled stuff, as you might expect, just brave and smart social psychologists doing their best. I find it highly plausible and confirmatory of many centuries of non-scientific observations about insularity. Disclaimer: I AM NOT SAYING DISTRUST OF PEOPLE OF OTHER BELIEF SYSTEMS IS GOOD, JUST THAT IT HAPPENS.
I know of no studies on friction over expression of religious beliefs. I do kind of take as a given that there are fewer HR complaints when everybody’s got the same Sacred Heart/Darwin amphibian/Santa Muerte/COEXIST bumper sticker.
Though some obvious confounders do come to mind...
Granted that there are huge trade-offs for religious homogeneity, and I think that it’s almost always a bad business decision (exceptions: semi-utopian communes? survival in Hobbesian chaos? new colonies without hope of reinforcement?) It was just an exemplary argument of a sort made less often than, you know, arguments about race and IQ.
Even if this were true, it would not follow that there is no countervailing incentive to remove barriers to employment for disadvantaged classes of people.
To refer back to the OP, why is the relevant disadvantaged class “black people” rather than “people with low IQ” or even “people unqualified for the job”?
why is the relevant disadvantaged class “black people”
As far as it goes, I’m in favor of preserving opportunities for all sorts of people to work, because it’s humanizing and it makes people happy. We’re all in favor of that, right?
But I also don’t think there’s been historic, organized pressure to keep low-IQ people from finding useful labor, and while such people deserve the protection of the law, it’s not illuminating to compare their plight to a group of people who were denied the ability to find employment they were very capable of using intimidation, violence, and bad-faith law....
...tools which, and this is sad, were very much still in use when the Civil Right Act was passed, and would still be in use today if it had never been passed.
Racial “classes”—not sets of corresponding genetic polymorphisms, which science tells us about, but race as we understand it in America, which is both more and less complicated—were not created by the Civil Rights Act, or the civil rights movement. They were created long before that, to justify cruelty, and to deny the continuing effect of that social construction would have been, in the the judgment of the majority of our Congress in 1964 and our Supreme Court since then, counterproductive.
All other things being equal, is anyone disagreeing with this?
Not at all. It’s a very rationalist sort of argument. There are many like it. I think it would be terrific if we spent more time exploring those, possibly at the expense of focusing heavily on arguments that seem a little less than disinterested.
If you proved to the courts that they were “valid,” meaning an accurate reflection of crystallized intelligence/abstract reasoning/g/whatever, this would not undermine the central legal argument against them, which is that they produce disparate impacts on protected classes.
Yes, and what is the justification for the disparate impact doctrine?
And for that matter what is the justification for declaring certain classes “protected”?
The American legal justification for the disparate impact doctrine, and for declaring race a protected category, is the 1964 Civil Rights Act, and the legislative justification for that was a history of massive mistreatment of individuals based on skin color.
I gather from the thrust of arguments in this thread that you may be strongly opposed to government protection of racial minorities in the United States, and that you may not believe that racial bigotry is—or possibly even was—a problem that needed legal redress. It is worthwhile to note that the legal basis for these doctrines is well established and, through the wonders of litigation, much studied and highly nuanced. That does not speak to any philosophical objections you have but, frankly, no philosophical objections you make have any bearing on the legal justification.
legal basis for these doctrines is well established
Um, the legal basis is the act of Congress. That’s all, you don’t need studies and nuances. Whatever Congress says and the President signs is the law of the land. Unless SCOTUS objects, of course.
This is a somewhat fundamentalist view of the law, and I am guessing many federal judges at all levels, and regulatory bodies of technical experts, would add something to your definition. I agree with you that the statutory basis for these court rulings is very clear.
But it’s also pretty clear that the doctrine of disparate impact, which is what he asked about, has been clarified and nuanced through litigation of those statutes. My point was that over many decades, the courts have not overturned this doctrine due to any philosophical objections of litigants.
But it’s also pretty clear that the doctrine of disparate impact, which is what he asked about, has been clarified and nuanced through litigation of those statutes.
Yes, of course, though it has nothing to do with legal basis—it’s interpretation of the law which is what the court system does all the time.
over many decades, the courts have not overturned this doctrine due to any philosophical objections of litigants.
Courts do not do that. A philosophical objection is not a legal objection—a court can overturn a law only by deciding that it is unconstitutional.
But I am unsure what is the point that you are making. Is it that both politically and legally the Civil Rights Act is untouchable in the US? Sure, but that’s pretty obvious…
Sorry, I meant the two questions in different senses, I should have made that clearer.
The American legal justification for the disparate impact doctrine, (..) is the 1964 Civil Rights Act,
The Civil Rights Acts didn’t specify disparate impact as opposed to disparate treatment.
and the legislative justification for that was a history of massive mistreatment of individuals based on skin color.
I understand the motivation, but I don’t think the ever increasing (and rather arbitrary) list of protected groups is a workable approach. Not to mention the “some groups are more equal than others” problem implicit in having a specific list of “protected groups”.
That does not speak to any philosophical objections you have but, frankly, no philosophical objections you make have any bearing on the legal justification.
If you look at the history of law, philosophical arguments end up influencing legal arguments all the time.
If you look at the history of law, philosophical arguments end up influencing legal arguments all the time.
I absolutely agree. It is conceivable that in the future, arguments could change the courts’ regard for this doctrine. But it is unlikely. The law has been in place for fifty years, and the doctrine has seen a ton of challenges in court.
Actually, the most useful application for individual businesses in this case would be (in the event that IQ tests are good at predicting who will be a good worker) to request IQ scores as part of a job application, not to discriminate based on race—this is not to say that it would be useful for society as a whole. I am not sure what it would do to society as a whole. On the one hand, if there’s a correlation between race and IQ, more people of each race with a low IQ might find themselves worse off. However, if employers become more willing to hire black people after testing their IQs, it could be a great boon to blacks and actually serve as a way to encourage people to judge each person based on individual characteristics as opposed to rejecting them for being part of a group. Simply tossing away all of the people of whatever race because the others have a low IQ would probably, in practice, not work very well—this is because they’re selecting from a pool of people who are qualified in the first place, and the process of becoming qualified acts as a filter. Most of the people they’re interviewing who are qualified are probably also intelligent enough to do the job—so you’d have too man false negatives this way.
I would say also that if aptitude tests are restricted because of racist connotations, it’s because people tied IQ to race.
Can you think of any applications for tying IQ to race that do not have the above issues?
Actually, the most useful application for individual businesses in this case would be (in the event that IQ tests are good at predicting who will be a good worker) to request IQ scores as part of a job application, not to discriminate based on race
The problem is that this is currently illegal in the USA. This was in fact precisely the point of the parent comment.
this is because they’re selecting from a pool of people who are qualified in the first place,
Also, I’m still interested in seeing the source that you believe is an accurate prior regarding race and IQ. Do you happen to have that information available?
The Bell Curve book is a standard source. Otherwise a quick look at Wikipedia provides this:
Rushton & Jensen (2005) write that, in the United States, self-identified blacks and whites have been the subjects of the greatest number of studies. They state that the black-white IQ difference is about 15 to 18 points or 1 to 1.1 standard deviations (SDs), which implies that between 11 and 16 percent of the black population have an IQ above 100 (the general population median). The black-white IQ difference is largest on those components of IQ tests that are claimed best to represent the general intelligence factor g.[11][non-primary source needed] The 1996 APA report “Intelligence: Knowns and Unknowns” and the 1994 editorial statement “Mainstream Science on Intelligence” gave more or less similar estimates.[42][43] Roth et al. (2001), in a review of the results of a total of 6,246,729 participants on other tests of cognitive ability or aptitude, found a difference in mean IQ scores between blacks and whites of 1.1 SD. Consistent results were found for college and university application tests such as the Scholastic Aptitude Test (N = 2.4 million) and Graduate Record Examination (N = 2.3 million), as well as for tests of job applicants in corporate sections (N = 0.5 million) and in the military (N = 0.4 million).[44]
Ok thanks. However, I am aware that “most published research is wrong” (PLOS Medicine) and know that there are factors that need to be controlled for in studies on race and IQ (in the second numbered list). Do you also claim that these factors were controlled for, that the key study or studies have been replicated, and that this is quality data that generally avoids research pitfalls? That’s what I am looking for.
Yes, I’ve read Ioannidis. However you’re using this quote here as a rather blatant aid to your confirmation bias. There have been many, many studies which all show the same thing. These are findings which have been confirmed, re-confirmed, and confirmed once again.
Precisely because these results are so controversial they have been the subject of very thorough checking, vetting, and multiple attempts to debunk them. The results survived all this. What, do you think that for the last 50 years no one really tried to find holes in the studies showing racial IQ differences? Many highly qualified people tried. The results still stand.
Yes, I’ve read Ioannidis. However you’re using this quote here as a rather blatant aid to your confirmation bias.
I think everyone should consider that published research findings are likely to be wrong each time they are seeking research findings. If you agree that we should be skeptical about research findings, why do you think that asking questions about whether the research controlled for multiple factors, was replicated etc. should be taken as evidence of confirmation bias? Maybe you disagree that we should be skeptical about research findings?
There have been many, many studies which all show the same thing.
Every single one? I would find that hard to believe for any topic, especially one as politically charged and controversial as this one, where both sides have a motive to bias research in their particular direction. If that is true, I would find it surprising. Assuming you were referring to the results of a meta-analysis, would you point to that meta-analysis please?
Precisely because these results are so controversial they have been the subject of very thorough checking, vetting, and multiple attempts to debunk them. The results survived all this. What, do you think that for the last 50 years no one really tried to find holes in the studies showing racial IQ differences? Many highly qualified people tried. The results still stand.
Are you saying that studies used for “The Bell Curve” did take into account the factors I mentioned, were replicated and / or may contain a meta-analysis that states that all the studies that could be found had similar findings?
If you aren’t specific about what measures were taken to ensure quality in the information you’re providing, I have no way to make the distinction between a matter of opinion and a matter of fact when you claim things like “The results survived all this.” Please be specific about what particular quality features the data in The Bell Curve provides.
The Bell Curve
I started checking out this book because of your high praise and was surprised to find this:
Can you explain why you seem to be disagreeing with me, when both myself and The Bell Curve agree that we don’t have a good way to tell whether IQ differences are nature or nurture? (Note: In addition to that, my view is also influenced by skepticism about research in general and an understanding that although IQ tests are correlated with various things, they have some limitations.)
That there are population level differences in IQ is not controversial (except in the sense that evolution is controversial because more than 30% of Americans don’t believe in it).
That IQ is a useful proxy for general intelligence and a useful tool in determining life outcomes is not controversial.
That IQ is heritable is not controversial.
That the differences are genetic is controversial but the data does seem to suggest that much of the difference is indeed genetic and another portion is biological (pre-natal care, early nutrition).
Did you read the two paragraphs following your quoted sentence? It seems to me that they more or less settle the matter, and resolve your grayness regarding environment and genetics.
What I suspect is going on is that Epiphany is statistically innumerate (as suggested by her rather hilarious statement about 25% of Africans being above average), but doesn’t want to loose status by admitting she doesn’t understand the arguments.
Both of the citations I was given by you guys said clearly that they were uncertain about the connection between race and IQ. That is the reason I don’t agree—because even your citations do not agree. I assume those are the best citations you have, so that your citations do not agree with you makes your belief look very bad indeed.
Also, by arguing that the reason I don’t agree is because I am statistically innumerate and that the reason I don’t agree is because I’m too inept to understand, you have made an ad hominem fallacy. Attacking the person does zilch to support your argument.
I can’t believe I just saw an ad hominem attack on LessWrong. That is the the most obvious behavior that one avoids if one wants to have a rational debate.
Both of the citations I was given by you guys said clearly that they were uncertain about the connection between race and IQ. That is the reason I don’t agree—because even your citations do not agree.
I don’t think you’re correctly distinguishing between multiple related claims.
Claim A is that IQ distributions vary by race. This is supported by mountains of evidence. This alone is sufficient to justify using race as a factor when predicting IQ. This is the original point under discussion, as you argued against using race as a factor when predicting IQ both here and here.
Claim B is that differences in measured IQs overestimate the actual differences in intelligence or life outcomes between races. There is substantial evidence against this claim, and it is only weakly related to the original point under discussion. (Were it true, it would suggest that estimating IQ is not as important when doing between-race comparisons as other estimations, but does not impact IQ estimation.)
Claim C is that X% of difference in racial average IQ is due to genetic factors. It is currently not clear what X is for any particular between-race comparison, which the citations reflect. This is unrelated to the original point under discussion.
Also, by arguing that the reason I don’t agree is because I am statistically innumerate and that the reason I don’t agree is because I’m too inept to understand, you have made an ad hominem fallacy. Attacking the person does zilch to support your argument.
Not quite. The arguments you’ve made recently are mostly social arguments- “you say Y, but your citation says Z, how do you account for that!”- rather than technical arguments.
The arguments seem vacuous to a technical expert, because Y and Z turn out to be totally compatible, but may still seem impressive to a non-expert who is unfamiliar with the relationship between Y and Z. Similarly, a non-expert doesn’t know what sort of claims do and don’t need citations, and so may see a virtuous skeptic against credulous believers, rather than a crank who defends their perpetual motion machine by insisting on a citation for the 2nd Law of Thermodynamics. (This is not to say that the 2nd Law of Thermodynamics is taken on faith; what it means is that there are some issues where ignorance, confusion, or denial is solid evidence against being a curious and open-minded scholar.)
An effective social response is to poke at the technical content of your claims at a much more basic level (i.e. charges of innumeracy). If you aren’t familiar with means and medians, and disengage when someone presses you on a basic statistical point, then anyone who is familiar with statistics can use that to gauge your level of technical ability. (And if you aren’t familiar with stats 101, why have a confident opinion on an inherently statistical topic?)
IQ distributions vary by race. This alone is sufficient to justify using race as a factor when predicting IQ.
Contrast this with Epiphany’s comment earlier on
If you want to attribute the IQ scores to race, not poverty or circumstances [emphasis mine] (...)
I guess it mostly comes down to “race” being such a charged term. Epiphany seems to be fighting “Someone being an African American predicts a lower than US-average IQ” because that has a lot of negative connotations, whereas you may reply “Well, but it’s technically correct, even if did turn out (unlikely/no idea) the correlation only exists because race predicts/includes poverty/culture/social customs, which in turn causally [e|a]ffect IQ”, while she would say “But then it’s not race!”.
Maybe something like “I refuse to use race as a predictor (even if I could) because that’s mindkilling/misleading, unless poverty/culture/social customs have all been controlled for. Since there is no scientific consensus on race when poverty/culture/social customs have all been controlled for, we shouldn’t speculate and rather work on the latter confounders, which can probably be improved with socially acceptable policies.” would capture most of her point and be more palatable?
you may reply “Well, but it’s technically correct, even if did turn out (unlikely/no idea) the correlation only exists because race predicts/includes poverty/culture/social customs, which in turn causally [e|a]ffect IQ”, while she would say “But then it’s not race!”.
I agree that there are a handful of valuable points one can make about the underlying causal diagrams. Suppose the diagram is X<-Y->Z: then the correlation between X and Z is indirect and acausal, where we should not expect that modifying X or Z will modify the other one. If the diagram is instead X->Y->Z, then the correlation between X and Z is indirect but modifying X may modify Z.
I disagree that those subtle points are the ones under discussion. It seems to me that this discussion is indirectly about the social acceptability of noticing the correlation between race and IQ, and that the technical points are used mostly for obfuscation. Suppose we were uncertain which of the two causal diagrams I described above were correct; we would still be certain that X and Z were correlated if Y is unmeasured, because that is the case in both causal diagrams, and responding to the claim that X and Z are correlated with “we don’t know if X causes Y or Y causes X” is irrelevant.
Since there is no scientific consensus on race when poverty/culture/social customs have all been controlled for, we shouldn’t speculate and rather work on the latter confounders, which can probably be improved with socially acceptable policies.
I would agree that it’s easier to intervene in wealth, culture, and customs than intelligence, and that we already know of several obvious ways where interventions in those three can positively impact intelligence. I don’t think Epiphany and I would agree on what interventions are most beneficial there, but that’s a separate conversation that’s not worth having here.
Arguments ad hominem are inappropriate for deciding the truth of the matter. They are entirely appropriate for deciding whether you want to take someone seriously or even just to talk to a person.
Conversely, I can’t think of any applications for which tying IQ to race is useful.
You might be in a situation where you need to decide how to allocate money to help a black school or a white school. If white people have higher IQs, and if money is worse at improving the performance of students who do poorly because of IQ than it is at improving the performance of students who do poorly for other reasons, then you should allocate the money to the white school.
You might be in a situation where you need to hire a white person or a black person and have no information about their IQs, but you would prefer an employee with a higher IQ. You then should hire the white person.
Of course, this is exactly why using IQ this way is a bad idea.
You might be in a situation where you need to hire a white person or a black person and have no information about their IQs, but you would prefer an employee with a higher IQ. You then should hire the white person.
How often do you know someone’s race but nothing else whatsoever about them?
Of course, this is exactly why using IQ this way is a bad idea.
How often do you know someone’s race but nothing else whatsoever about them?
It’s an additional piece of information and allows you to do more of an update. If black people, on the average, have lower IQs than white people, then (for instance) black people with college degrees are still likely to have lower IQs than white people with college degrees, even if college raises the likely IQ for both groups. It is also possible to have traits such that given those traits race is no predictor at all, but those would be balanced out—if race is no predictor of IQ among people with PhDs, it must be a stronger predictor for people without PhDs than it is for the general populace.
If black people, on the average, have lower IQs than white people, then (for instance) black people with college degrees are still likely to have lower IQs than white people with college degrees, even if college raises the likely IQ for both groups.
Not necessarily. You haven’t accounted for differences in variance between the two groups.
To make a very rough analogy: it seems widely believed that there is greater variance in intelligence among men than among women; surely such differences are imaginable among racial groups. (For one thing, there’s more genetic variation among Africans than among other human populations — which makes sense, given that all other human populations descended from small subsets of Africans.)
Nor for selection effects on who gets to go to college — for instance, there seem to be a lot of pretty dopey people who have unusual bonuses to their chances to get into college on account of their parents being wealthy.
Nor for socioeconomic differences in general, which are substantial between whites and blacks in America. To make another very rough analogy: If two people reach the same measurement of achievement, but one has to overcome greater obstacles to get there, we would often take this as an indicator that that person had greater ability: a runner who runs a five-minute mile while carrying ten pounds of lead weights is a better runner than one who makes the same achievement carrying no weight.
Not necessarily. You haven’t accounted for differences in variance between the two groups.
Yes I have—see the next comment. If blacks with college degrees don’t have lower IQs than whites with college degrees, and blacks in general have lower IQ, then blacks without college degrees must have even lower IQs so that the average is still lower IQ.
Furthermore, taking other factors into consideration can result in worse discrimination. Consider the trait that most obviously makes up for the difference—actually taking an IQ test. Blacks may have lower IQ than whites, but blacks who take an IQ test and score X don’t have lower IQ than whites who score X at all. But if I were to hire people based on the trait “scoring X on an IQ test”, and X is at the high end, I may end up hiring a lot more whites than blacks—a small difference in IQ translates to a large difference in the number of people at the tail end of the distribution. If people with IQ 145 are 10 times as common as people with IQ 150 and the black curve is only shifted by 5 points, and I want to hire people with IQ 150, I may end up hiring whites to blacks at a 10 to 1 ratio compared to the proportion of blacks who apply.
Well, seeing an unknown man approaching you at night, granted this is more about criminality than IQ but the correlation is the same.
Once you specify where I am, who I am with, what kind of body language the man is using, how big he is, and what he is wearing, further specifying what race he is wouldn’t matter that much.
Also thinking about whether affirmative action and the desperate impact doctrine are reasonable ideas.
I can’t recall anyone on LW advocating those, so you might be attacking a straw man.
Once you specify where I am, who I am with, what kind of body language the man is using, how big he is, and what he is wearing, further specifying what race he is wouldn’t matter that much.
Is that true? Depending on the “where I am” part?
There’s only so much you can tell about someone from “what kind of body language the man is using, how big he is, and what he is wearing”, after all. In the right racially-segregated society, could it provide valuable additional data?
Also thinking about whether affirmative action and the desperate impact doctrine are reasonable ideas.
I can’t recall anyone on LW advocating those, so you might be attacking a straw man.
That’s because they rarely come up. In any case my point is that these doctrines are in place in the USA and the false belief that race is uncorrelated with anything important.
I mean data about individuals like resumes and qualifications That racial-group info correlates with important things is unimportant, unless it correlates significantly more than individual data. However, the reverse is the case.
First I don’t understand the distinction your drawing between “individual data” and presumably “group data” since the people with a particular qualification are a group and conversely skin color, say, is a property of an individual.
Back to the point: in the great-grandparent I was talking about affirmative action and the disparate impact. The logic of those is based on concluding that racism happened on the basis of disparate outcomes. This logic relies on the implicit premise that race isn’t correlated with anything important. I don’t see how anything you wrote in your two comments that addresses this issue.
Well, seeing an unknown man approaching you at night
Actually, it is far more prudent to avoid a stranger approaching me at night, regardless of his race—depending on the environment I am in.
If he is approaching from a dark alley, I will head away from him, whatever his race. If he approaches me at a party full of friends, I will speak to him.
The crime statistics are not so incredibly different for blacks and whites that you can simply trust all of the whites.
All of that looks rationalization of a pre-determined belief.
No it doesn’t. The conclusion supported is far too close to the ‘middle ground’ on the issue to readily pattern match to rationalization and seems to be a description of considered reasoning from plausibly held premises. The ‘rationalisation’ and ‘pre-determined belief’ charges could be credibly made in response to many of the comments in this thread but doesn’t apply to the grandparent.
NOTE: I don’t entirely agree with with either Epiphany’s position or Eugine’s position. In particular Epiphany seems a little too disillusioned with research while Eugine has somewhat too much passion on the race/IQ political correctness subject to keep his claims balanced enough that I could support them despite agreeing that there are almost certainly IQ differences between groups selected by just about any significant feature—not that this seems like an especially useful thing to place emphasis on. I seem to recall some credible claims about higher average mathematical intelligence in Ashkenazi Jews for example.
Regardless of whether I agree with the position being argued with, the parent is making what seems to me to be a false accusation and one of a kind that derails the flow of discourse.
No it doesn’t. The conclusion supported is far too close to the ‘middle ground’ on the issue to readily pattern match to rationalization and seems to be a description of considered reasoning from plausibly held premises. The ‘rationalisation’ and ‘pre-determined belief’ charges could be credibly made in response to many of the comments in this thread but doesn’t apply to the grandparent.
Respectfully disagree, though we may be focusing on different parts. It worries me that your analysis seems to focus on the conclusion drawn rather than the procedures followed.
Consider this bit of the great-grandparent:
2) Reasoning that Takes Relevant Data Into Account: “The IQ test(s) said African’s IQs are lower than those of whites. However, there are known flaws with IQ tests such as cultural bias, so that figure might be wrong. Most published research findings are false (PLOS Medicine), so I should apply healthy skepticism to all the research I read. This is not likely to be an accurate piece of data to use as a Bayesian prior. Once I’ve decided on a prior to use, I should then adjust for other relevant data (things I know about the specific individual).
This looks like pure rationalization to me, with many inaccuracies and irrelevancies, all of which support the intended conclusion. (An unbiased sloppy thinker should be expected to make mistakes in both directions simultaneously; when the mistakes all point one way it suggests bias.) The most egregious irrelevance is the attempt to discredit one of the most replicated findings in social science with a study that showed that prominently promoted recent research often fails to replicate. The last sentence is bizarre by its addition- would someone using race as Bayesian evidence not update on other information? (The later reference on Bayesian updating is similarly bizarre.)
The rest of the great-grandparent discusses how, even if we had an estimate, we should be careful how we use that estimate. Of course—who would argue against using estimates carefully?--but irrelevant to the question of whether or not it is ethical to use Bayesian reasoning.
It is almost always bad Bayes, or any other kind of reasoning, to make judgements about individuals based on group characeristics, since there is almost always information about them as individuals available which is almost always more reliable.
Group level information is still useful for shrinkage of estimates and correcting for the always-present unreliability in individual estimates; see for example the long conversation between me and Vaniver on LW somewhere where we work through how you would shrink males and females’ SAT scores based on the College Board’s published reliability numbers.
And E(X’s deserved SAT score|X’s measured SAT score; X is male) - E(X’s deserved SAT score|X’s measured SAT score; X is female) was, like, four points? I still think people’s System 1 are likely to overestimate this difference if they know about the correlation more than underestimate it if they don’t.
And E(X’s deserved SAT score|X’s measured SAT score; X is male) - E(X’s deserved SAT score|X’s measured SAT score; X is female) was, like, four points?
A 4 point adjustment (or more) across all candidates based solely on 1 binary variable (gender) and a trivial centuries-old bit of statistical reasoning seems like a fairly impressive output, and likely to make a difference on the margin for thousands of applications out of the millions sent each year.
I still think people’s System 1 are likely to overestimate this difference if they know about the correlation more than underestimate it if they don’t.
And E(X’s deserved SAT score|X’s measured SAT score; X is male) - E(X’s deserved SAT score|X’s measured SAT score; X is female) was, like, four points?
For applicants scoring 800 on a hypothetical normally distributed math SAT, yep. For normally distributed tests, shrinkage is linear based on the difference between the group mean and the measured mean, and so it’s smaller for less extreme scores.
(For some reason, I’m having difficulty finding the link to the actual conversation; I think Google search is not going into deep comment threads, and the search function is based off the site, rather than just a database of my comments. Anyone remember helpful keywords further upthread to get a link to the actual conversation?)
I still think people’s System 1 are likely to overestimate this difference if they know about the correlation more than underestimate it if they don’t.
Saying “we shouldn’t explicitly calculate something because some people might implicitly calculate that thing incorrectly” sounds to me like going in the exact wrong direction.
(For some reason, I’m having difficulty finding the link to the actual conversation; I think Google search is not going into deep comment threads, and the search function is based off the site, rather than just a database of my comments. Anyone remember helpful keywords further upthread to get a link to the actual conversation?)
Culture and environment are not race. Therefore, if you’re studying race, those influences should be taken out of your scientific experiment. It’s extremely difficult to remove things like culture and environment from a study on IQ. The fact that so much is correlated with it doesn’t mean the results of studies intended to determine racial differences are significant so much as it means they’re a tangled mess of cause and effect which we likely haven’t sorted out adequately.
A. We don’t want black people to suffer needlessly.
B. We don’t want to encourage ourselves and others to be prejudiced against people when, regardless of what the average African’s IQ is, it is still both logically incorrect (hasty generalization) and ethically wrong to prejudge individual Africans. However, knowing how humans behave, we figure that if people believe Africans have lower IQs, that will result in an increase in prejudice.
Actually, I bet some people are not happy saying that there are correlations there. This is one of those notions you might want to double check.
So what you’re saying is that it’s ethically wrong to use Bayesian reasoning.
It is dangerous to be half a rationalist. This applies to groups as well as individuals. No matter how good your process for arriving at beliefs, it is indeed unethical to go around spreading those beliefs to people that will predictably misunderstand and misuse them.
Funny how the only people to make that argument tend to be people who don’t want to believe the beliefs in question, but have out of ways to ignore the evidence.
On a less meta level: what kind of “misunderstand and misuse” do you think is going to “predictably” happen.
I, for one, do believe that the average African has lower IQ than the average European but don’t go around telling that to the wrong people.
Underestimating how much the evidence race provides about an individual’s IQ can be screened off by other evidence about the individual, due to the confirmation bias and similar.
In the interest of clarity: I am not at all sure how to proceed in this particular case. History makes me wary of departing from the current Schelling point of assuming everybody is equal, but that’s not my point.
I am saying that a course of action based on Bayesian reasoning has no special immunity to being ethically wrong, and it is those actual results that are worth worrying about, not merely the epistemology.
If you intend to use an African prejudgment heuristic like 1 (below) rather than reacting as if you’ve done an equation that takes into account other relevant data like 2 (below), then I think your probability equation needs an upgrade.
1) African prejudgment heuristic: “The IQ test(s) said African’s IQs are lower than those of whites, therefore this specific African individual is likely to be relatively stupid compared to my white friends.”
2) Reasoning that Takes Relevant Data Into Account: “The IQ test(s) said African’s IQs are lower than those of whites. However, there are known flaws with IQ tests such as cultural bias, so that figure might be wrong. Most published research findings are false (PLOS Medicine), so I should apply healthy skepticism to all the research I read. This is not likely to be an accurate piece of data to use as a Bayesian prior. Once I’ve decided on a prior to use, I should then adjust for other relevant data (things I know about the specific individual).
Then, if one wants to behave rationally after one has decided what to believe, I think one must continue by thinking something like this:
“Considering things like...
A. …the fact my prior is likely to be inaccurate (there isn’t an accurate one for this subject as far as I’m aware)...
B. …the fact that even if the IQ study is correct and the IQ test it used was accurate, there’s a decent chance (25%) that this specific individual has an IQ above the African average—meaning I need to avoid the logical fallacy called hasty generalization...
C. …the risk of lost utility via damaging this individual’s reputation, emotional health or opportunities for success by pre-judging them...
D. …the risk that this makes for bad social signaling and witnesses may retaliate against me with one or more forms of social rejection if I pre-emptively treat an African like an idiot...
…do I really want to treat this person as if they are less intelligent?”
I think you may have reacted to my “I hope it is false.” statement or my “it’s ethically wrong to prejudge individual Africans” statement—but that shouldn’t matter to your probability calculation. What should matter is to get an accurate idea of reality. Along with saying other things, I also provided other factors which are relevant, as credible sources can confirm. If one wants to be a good Bayesian probabilist, after one specifies some prior probability, one must then update it in the light of new, relevant data. [1] This situation where you focused on one part of my comment and ignored the rest reminds me of those math problems where there’s an irrelevant statement thrown in just to distract you. I of course did not intend to distract you, but since you seem to think that Bayesian reasoning in this case means ignoring all the other data I presented, it appears that you have skipped the parts of the process where you ensure an accurate prior and update your prior with new, relevant data.
Life is really, really complicated. I doubt it’s ever wise to just grab a prior and run with it and I certainly hope that you do not reason this way.
Paulos, John Allen. The Mathematics of Changing Your Mind, New York Times (US). August 5, 2011; retrieved 2011-08-06
This reads like a classic case of motivated cognition. You don’t want to believe the conclusion; therefore, you selectively look for potential flaws then declare that there’s still a chance. The reason I believe the connection between race and intelligence is not just because of the tests but because more or less every relevant aspect of reality (e.g., the statistic on race and crime, the nearly complete lack of blacks in intelligence intensive fields, e.g., math, programing, the state of majority black countries) looks the way one would expect it to look if the connection existed.
Who is claiming otherwise?
Yes there is. You just don’t want to believe it exists.
Actually the chance of this particular black being above the African average is 50% (more if I condition on the fact that he is in the USA). The probability that he is above the white average is significantly less. The probability that he is above some high cutoff is can be even lower.
This is a general argument against using evidence of any kind.
This is potentially a problem, although here the problem is arguably with the witnesses behaving irrationally and/or participating in out of control signaling arms races than with the strategy itself.
Did you stop to make distinction between me being influenced by motivated cognition and alternate explanations like:
Me seeing significant flaws in data that would otherwise support your conclusion. Part of this may be that I’ve spent a significant amount of time reading about IQ and giftedness and I have learned that there are a lot of pitfalls to doing IQ related research.
Me simply being unaware of relevant data. (This might be the case in the event that the people who supplied my data were influenced by motivated cognition or confirmation bias.)
You seeing motivated cognition in my words because of being influenced by motivated cognition yourself?
There is an alternate explanation for those which does not have the same issues that IQ tests and studies have: The effects of slavery and prejudice. We are certain that slavery and prejudice has influenced them, and that it has existed for a long time. To know this, one must only look at the KKK or investigate the history of black enslavement. Imagine a third world country. Imagine that an equal proportion of those inhabitants are removed and used as slaves. Imagine an equal proportion of them dying. Imagine that they’re freed, but all of them—not some but all—are freed into a situation of extreme poverty where they don’t even own a home or have the ability to read. Many still aren’t being taught to read. Consider also that even though there have been advances in medicine, poverty means you can’t afford health insurance or medical treatments. Don’t think that disability and chronic illness are uncommon—they’re not. Not even in America. They’re probably especially common for the poor. Don’t think that severe worker abuse ended with slavery, either—do some research on sweatshops in America sometime. Now take into account the effects of stress, and the human element—how those effects can compound into things like mental illnesses and drug addictions. Would you predict that the majority of these people who started out with literally nothing and without even the education to read would manage to avoid pitfalls like disability, mental illness, drug addiction and sweatshops and carve an opportunity to excel out of poverty and ignorance over the course of 150 years? I would not expect that. I would expect most of them to have fared poorly.
I don’t see a good way to tell the difference between a low IQ score due to actually being less intelligent versus a low IQ score due to nurture-related reasons such as the following:
Improper nutrition due to poverty.
Lack of education.
The effects of extreme stress (How are you supposed to focus on an IQ test when you’ve just been threatened by a gang?)
Suffering from medical conditions (these can cause memory symptoms, brain fog, and fatigue), mental conditions or drug addictions.
Having been parented by people that were mentally or physically ill, severely stressed, or addicted.
Cultural differences that cause arbitrary communication issues during testing.
The psychological effects of prejudice (may influence things like self-esteem and locus of control or result in learned helplessness, etc.)
If you want to attribute the IQ scores to race, not poverty or circumstances, then there needs to be a good way to distinguish between nurture and nature as a cause for low IQ scores. Do you have one?
If it’s true I want to believe it. However, it’s hard to believe it exists without a citation. Do you have one?
I respect you more for being able to say something that supports my view better than it does yours. +1 karma for that. I still think the number is 25%, however I do not view this as a key point in our disagreement, so I will leave it at that.
You appear to have taken that statement as an argument regarding what to believe. It was not. I deliberately put that part after the section where I was discussing deciding what to believe, and put it under “if one wants to behave rationally”.
As the saying goes, “Life is an IQ test.” The predictive ability of IQ on income (and most other statistics of interest) is very similar for each race, which suggests that differences in measured IQ scores map onto differences in life outcomes. To the extent that a medical condition makes someone test poorly, it generally also makes them live poorly. (There are a handful of prominent exceptions to this- like dyslexia- which don’t significantly impact the main point.)
Now, where those IQ differences come from in the first place is an interesting question. Many people have looked at it, and the consensus answer for individual IQ differences is “somewhere between 50% and 80% of it is genetic,” and the extrapolation from that to group IQ differences is somewhat controversial but seems straightforward to me. A perhaps more interesting question is “what knobs do we have to adjust those IQ differences?”
Here’s Jason Richwine explaining that all serious scientists have agreed on the basics of IQ for decades, and the media is completely mistaken on the state of reality and scientific consensus.
It seems pretty relevant to me, because it looks like basic statistical innumeracy on your part, unless you think the IQ distribution of African Americans is tremendously skewed such that the mean intelligence is the 75th percentile of intelligence, rather than the 50th percentile like it would be in a symmetric distribution. (Or you think that the African average is higher than the African American average, which is very much not the case.)
That’s a saying? Are there also sayings “Life is a test of height.”, “Life is a test of immune system efficiency” and “Life is a test of facial symmetry”? We may as well round out the set. Anything of comparable test significance that I missed? Oh! “Life is a test of breast perkiness” and “Life is a test of capacity for situation-appropriate violence.”
(I actually agree with the point of the rest of the paragraph.)
EDIT: sorry, misunderstood your comment.
Those may be qualitatively similar but I would suggest they are quantitatively different. I would be surprised if facial symmetry did not correlate with income, health, social status, and so on, but I would expect the correlation to be much lower than the correlation with IQ. The saying means that most metrics of life success are moderately highly g-loaded, and so it makes sense that IQ correlates positively with basically everything good.
I think you either drastically overestimate the impact of IQ or underestimate the predictive value of those other factors of life success—particularly in environments different to those experienced by the white male nerd class of first world countries. Regardless, the paragraph with that alleged saying redacted would be far more persuasive than the one with it included. It strongly undermines the credibility of your point. I currently agree with you despite the opening sentence, not because of it.
Perhaps we should switch to numbers to make it easier to communicate. I haven’t looked at any numbers recently, so I don’t expect these guesses to be particularly accurate, but I’d guess IQ correlates .2 to .6 with most interesting measures of life success, and height correlates around 0 to 0.1 (after controlling for IQ). I’d guess facial symmetry has correlation >0.3 with other health-related things, a correlation around 0.4 with social things, and a correlation below 0.3 with the rest.
If you have a source handy that estimates these sorts of correlations, I’d like to see it, but I don’t think it’s important enough to spend time hunting something down.
I would be very surprised if IQ correlates at .6 with, say, wealth or income. Parental wealth and income possibly correlate no more than 0.5 to childrens’ incomes, and it would be frankly remarkable for IQ to be (1) transmitted intergenerationally to a large degree, and (2) more closely correlated to financial outcomes than one’s parent’s financial outcomes, since your parents often give you not only your genes, but your inheritance/early support, financial assumptions, and first set of career contacts.
Intelligence: Knowns and Unknowns claims that parental SES and IQ are correlated at .33; that parental SES explains one third of social status variance (which implies r=.58) and one fifth of income variance (r=.45); that IQ explains about a quarter of the social status variance (r=.5) and a sixth of the income variance (r=.4), and that also correcting for parental SES reduces the predictive ability of IQ by a quarter. I would expect more recent numbers to be broadly similar.
Thanks for the link. It’s a nice summary of the state of research a few years back, and anybody who’s interested in the topic should read it.
It is probably even more interesting to me because it tacks pretty hard away from the conclusions some people have drawn in this thread. The authors clearly did not believe that the well-attested differences in IQ testing across ethnic groups could be ascribed to genetic factors.
You’re welcome!
I think that there’s not definitive evidence on the subject, even today. The definitive evidence would be if we knew the specifics of the causal link between genetics and IQ and had representative genetic samples from different racial groups, so we could look at the prevalence of various IQ genes and calculate what we’d estimate the average racial IQ to be for various groups from their genes. That’d give us an estimate of the genetic factors, and the difference between that and measured IQ would give us an estimate of the environmental factors.
My read of the field, though, is that the majority of the evidence points to the majority of the difference between racial groups being explained by genetic factors, and the trend has been that the hereditarian position has been growing more solid over time, especially as more and more people have been sequenced.
For example, African Americans represent a significant observational data source on ancestry and intelligence, since while the average African American has 80% African ancestry, that percentage can vary significantly from person to person. Intelligence: Knowns and Unknowns reports that a study that used blood groups to determine ancestry failed to find a correlation between European ancestry and IQ; more recent research claims to have found a correlation between European ancestry and IQ and controlled for the impact of skin color.
Do you similarly control the IQ finding for height? For obvious reasons that is necessary for consistency.
I’d certainly expect (based on loose memories of studies encountered) height to predict more than ‘0 to 0.1’ and IQ to predict a heck of a lot less than 0.6, especially when considering populations that are not limited to first world nations. I wish it were otherwise, naturally. I’d love the world to be more biased in favour of my own highest stat.
In any case, I would consider it legitimate to readers who encounter “Life is an IQ test” delivered in response to the quoted context to be sufficient evidence that the comment is not worth reading and downvoting and ignoring it. It was only because I was more patient that normal that I bothered to read further and found you had a good point hidden in the detail. Do with that information what you will.
You can re-allocate some of that to Charisma if you work really hard (stand-up comedy is a learned skill) and if you have a British or Australian accent you get +1 just by coming to North America and talking. Provided you haven’t already maxed it, Strength is highly trainable, as are Dexterity and Constitution. Even HP, up till about 30 when bone density stops accumulating. Wisdom is extremely trainable, and there’s some evidence the world’s biased that direction, so I’d throw points there when in doubt.
In fact, even those who have reached their maximum strength potential can increase it by using the highly potent Potion of Potential Strength.
HPs are primarily determined by Constitution (somewhat trainable) but in many worlds (including the real one) there is also a bonus from Strength stat. Better developed muscles are useful for absorbing damage non-critically.
Perhaps the most important stat is actually Willpower, which is also somewhat trainable and can be buffed with items and hired allies.
Well spotted on the ‘favour’ usage! Yes, I’ve noticed a bonus there (my current girlfriend is North American), presumably the accent helped.
If you do a linear regression with both IQ and height as inputs, this automatically separates their (linear) effects.
But I’m not sure about the general point, because IQ is an estimate of a factor and height is an estimate of a single variable. My impression is that when you are looking for the individual effect of a component of a hidden factor, then you want to control for the factor before measuring the effect of the component, but when measuring the effect of the hidden factor you don’t control for the components. But height isn’t a component of the IQ calculation, though it does appear to be weakly related to intelligence.
I’m also having trouble remembering if the 0.05 number I remembered was an r or r^2, which would significantly impact the range. I’m finding rs of about .2 for the correlation between height and intelligence, and rs of about .2 for the correlation between height and income without controlling for intelligence. Haven’t found anything yet that does control for it. (IQ correlation with income, as mentioned in a cousin comment, is about .4.)
It’s still not clear to me what about the saying you find objectionable. The best guess I have is that you took it to mean that the g-loading of life success was comparable to Raven’s, when I meant that they were g-loaded at all. The heart of that paragraph:
seems to me like a fair explanation of the saying, and why it’s confused to think about nurture factors influencing intelligence as separate from actual intelligence.
As a stand-alone statement, I would probably leave this alone. But as a response to “What about nurture”, the first thing that comes to mind is:
Has Vaniver adequately corrected for the just world fallacy?
Ok, that’s interesting, but it does nothing to rule out nurture factors that would impact both IQ and income.
I agree that IQ is mostly genetic and that IQ does seem to correlate with a lot of factors. I’m not saying IQ does not exist. What I am saying is that, specifically when it comes to black people, there are other factors that are definitely influencing performance and IQ scores. Therefore I reject claims about IQ and race that haven’t controlled for known factors.
Actually, when I read that section (it starts with “What scholars”), I parsed it like this:
Jason explicitly says that there’s a scientific consensus on many issues that seem controversial to journalists.
Jason states that virtually all psychologists (not scientists) believe there is a general mental ability factor (That he’s not saying is specifically connected to race).
Then, without qualifying these statements with anything along the lines of “most x believe”, he states: “In terms of group differences, people of northeast Asian descent have higher average IQ scores than people of European lineage, who in turn have higher average scores than people of sub-Saharan African descent.”
I will not assume that this sequence of claims means that the group differences statement is also something scientists have a consensus about. If I did, that would be a non-sequitur.
Also, below that, he writes:
“It is possible that genetic factors could influence IQ differences among ethnic groups, but many scientists are withholding judgment until DNA studies are able to link specific gene combinations with IQ.”
This is where I stop reading the article because it is clear to me that it does not say “there’s a scientific consensus that there’s a link between race and IQ”. If you have a credible source for that claim, I’ll be curious about it. No more Politico articles please.
It might or might not have been an error. In any case, I’m not going to go digging for that right now because I still think knowing the percentage is irrelevant to the current point. Whether the figure is 50% or 25%, it is still true that a significant proportion of people will have an IQ above average and therefore it would be hasty generalization to assume that a person of a certain group was an idiot. That is one way in which the exact number is irrelevant. However, that point about hasty generalization is much more irrelevant at this particular moment because we haven’t even decided on a prior, let alone have we got a decent posterior—so the step where we have a concern about making a hasty generalization based on our probabilities should be in this disagreement’s future. If it becomes relevant, I will dig around, but not right now.
It’s not clear to me why you would be interested in nurture factors. There are two things going on here: the ability of IQ to measure intelligence, and the historical causes of intelligence.
With the exception of disorders that prevent people from testing well without significantly impacting life outcomes, the historical causes of intelligence don’t appear to have much to do with the ability of IQ to measure intelligence. A nurture factor (like, for example, being breastfed as a child or being struck on the head) actually alters someone’s intelligence, and their intelligence influences both their test scores and their income.
Again, this looks like it’s mixing up the historical causes and the predictive ability. If the predictive ability is the same independent of race (it is), then it doesn’t matter why the racial IQ averages are the way they are. What we would need to show to discount IQ measurements is that the IQ measurements are not as predictive for members of one race than another.
As an example of a real bias like this, girls tend to get better grades than standardized test scores alone would predict. In order to get an accurate estimate of what a girl’s grades would be from her standardized test scores, you need to adjust upwards because she’s a girl. Symmetrically, boys score better than one would expect from their grades, and so when predicting scores one needs to adjust upwards. When moving in the opposite direction, one would need to adjust downwards; a girl’s grades overestimate her standardized test scores. But note that this doesn’t mean we throw out the data- it’s still predictive! We just adjust it the correct quantitative amount.
Now, do we know the historical causes of that effect? I’m not familiar with that field, but it seems like there are lots of plausible theories that probably have support. Even without knowing the causes, though, we can use our estimates of the size of the effect in order to predict more accurately.
What would you call a scientist who studies intelligence? (I suppose I should also make clear that by “serious” I mean a scientist speaking confidently in their field of expertise.)
One does not say “most scientists believe that hydrogen has one proton,” one says “hydrogen has one proton.”
Here’s the APA report he references. The group means section starts on page 16.
In general, though, asking for citations like this is really frustrating, because it doesn’t seem like the true rejection. The linked Richwine article referenced more serious sources that you could find if interested, and even if you didn’t notice that Googling “racial IQ averages” leads to this as the fourth hit, and if sufficiently motivated you could find the paper that journalist was writing about, and so on.
But if you’re not curious enough to seek out this information, and you don’t seem to have updated on the other information I’ve provided, what reason do I have to expect that the difference between my position and your position is that I have citations, and as soon as I share them you’ll adopt my position?
Sure. When you think in distributions, an estimate generally comes with both a mode and a precision (or, relatedly, the standard deviation). Knowing someone is African American gives you an estimate with a mode of 85 and standard deviation of 15, which has a non-trivial but small chance of being over 120. Knowing someone got a 120 on a recent IQ test gives you an estimate with a mode slightly south of 120 and a standard deviation of probably 2-5, depending on the precision of the test.
In Africa? It so happens that the world is much bigger than the USA and the people in sub-Saharan Africa test for IQ pretty much the same as African-Americans.
Sure, you can control for wealth/economic status. Or you can go and test poor peasants in China and poor peasants in Africa. You seem to think that this is a white-vs-black US problem. It’s not. The highest-average-IQ large group of people is East Asians, like Han Chinese—not Caucasian whites.
I am curious—how do you figure out that in a distribution close to normal only 25% are higher than the mean?
Actually much of sub-Saharan Africa has average IQ around 70, whereas African-Americans average around 85.
Yeah. I suspect there are two reasons for that. First, malnutrition as a child can drive your IQ down and malnutrition is much more common in sub-Saharan Africa. And second, many African-Americans have some white ancestors. Look at Obama, for example—he self-identifies as African-American though only half his genes come from Africans.
Ethiopians have lots of Caucasian admixture too. (But once we know that both genes and environment play an important role, working out which fraction of the variance in IQs is due to each to within three significant figures doesn’t sound terribly interesting to me.)
How is that not self-evident given the edge cases (puppies going to human schools / children growing up in a sensory-deprivation tank, both not doing well on IQ tests)? Regarding the significant figures, we need to keep in mind those are to be interpreted as “this is how much of the variance factor X explains given a certain scenario”. They will vary across e.g. nations:
In a homogeneous environment (e.g. classless society, higher Gini-index), genes will acount for more of the variance than in a mixed environment with people of the same genetic makeup. IOW, as you e.g. change the school system, or who marries whom, so you change those relative weights of nature v. nurture.
You might say “well, given typical circumstances and typical gene pool variances”, but consider that the discussion is in any case comparing e.g. the US to sub-saharan Africa (or whereever), which absolutely cannot have the same relative weights for their respective nature versus nurture, unless the different gene variances in tribal societies and the different “school” environment somehow equalled out, a dubious proposition.
cough colonialism cough
So the horrible European experience of being a Chinese colony has shocked the Caucasian population into scoring noticeably lower on the IQ tests than the Chinese?
I was referring to:
It is far from hard to see how sub-saharan africa has been stripped and degraded over the centuries in a way that, say, rural China wasn’t.
And I am far from convinced that IQ, being an extremely culturally contextual measure, can be disentangled from modes of thought that lend themselves to abstract pattern analysis being more or less common in different traditions.
Centuries..? Central Africa was explored by Europeans (that is, the first Europeans appeared there) in the mid-XIX century. Most of the African countries were independent by late 60s early 70s of the XX century. Notable chunks like Ethiopia were never colonized (Ethiopia was occupied by Italy for a short time in 1930s and early 40s).
But anyway, it is hard for me to see. Can you provide data?
So your thesis is that the Chinese culture is very suited to “abstract pattern analysis”, the European culture is moderately suited and the Black culture… oh wait there is no single unified Black culture, so the Black population is unique in that its culture doesn’t matter but it was so similarly oppressed in Africa and the US that they ended up with similar reduced IQ. But the Ashkenazi Jews, though they were oppressed in Europe, ended up with a higher-than-Caucasian IQ.
Um...
If your point is that it’s not clear to what extent the difference in intelligence is due to nature or nurture, I agree but would like to point out that for many applications it doesn’t matter.
“X have a lower IQ on average.”
You can choose “People of African descent” or you can choose “People from poor backgrounds” or “People with serious health conditions” or “People with drug addictions” or any number of other things.
When attempting to determine how best to help a school in a black ghetto that is failing, and you’re choosing between spending money on remedial courses or on a school nutrition program, you will most certainly benefit from having this knowledge.
Conversely, I can’t think of any applications for which tying IQ to race is useful. Would you name three examples?
Also, I’m still interested in seeing the source that you believe is an accurate prior regarding race and IQ. Do you happen to have that information available?
If the results of the racial IQ studies are true, then that is very important because it disproves the doctrine of ethnic cognitive equality. Many people, especially in America, have this idea that all ethnic groups must have exactly equal average cognitive ability, and that if one or more ethnic groups perform below average on a test of aptitude, that is taken as strong evidence that the test is invalid and racially biased and thus cannot be used.
For this reason, many aptitude tests are severely restricted in their use since they are considered racist. This in turn would have a negative economic impact if these tests are actually valid, since employers and colleges are forced to use other, less effective means to vet candidates.
Actually, the legal rationale for restricting the use of such tests in certain kinds of hiring is not that they’re invalid. If you proved to the courts that they were “valid,” meaning an accurate reflection of crystallized intelligence/abstract reasoning/g/whatever, this would not undermine the central legal argument against them, which is that they produce disparate impacts on protected classes.
There is the “business necessity” defense to disparate impact accusations. If the courts were to accept that IQ tests correctly reflect g/intelligence that defense will be much more applicable.
I’m pretty sure the courts have allowed that IQ-like tests are acceptable in many situations for many types of employment. It’s not a hypothetical. I guess I’m saying the question of the “validity of the tests” is a red herring, even if it’s an ideological hot potato. I think the main debate these days is not at all about the validity of the tests, it’s a debate over business necessity versus disparate impact.
I am not aware of that “main debate”. In the US, at least, political climate makes it impossible to discuss race issues in public. The courts, of course, have to decide these issues, but that hardly constitutes debate.
Race issues are discussed constantly in the U.S. — often, but not always, under guises such as “immigration” or “the War on Drugs” or “failing schools”.
However, certain views are broadly discredited, for instance those which attribute or imply differences in the moral value of people’s lives on the basis of their race.
Fair enough. For “main debate” please read “pertinent legal question.”
Well then, we’ve come to stating that the pertinent legal question is whether the use of IQ tests in hiring falls under “business necessity”. I don’t know of any answer to that other than “it depends”.
Though the issue of whether a job really requires high IQ is an interesting one...
Which is still ridiculous. It’s been known for generations that IQ has a positive impact on basically every job, which should imply that the default is to assume business necessity for IQ tests.
Even if this were true, it would not follow that there is no countervailing incentive to remove barriers to employment for disadvantaged classes of people. Is it not possible that society has an interest in broad employment, especially among people disadvantaged by such tests? Two thoughts:
1) IQ tests have a history of being used deliberately to weed out applicants of certain races. This was not an incidental effect: it was the entire purpose of the test, much like literacy tests for voting. The odds of them being used this way again, were changes made in the law, seem extremely high.
2) It is interesting that LW sees so many rational arguments for policies that would give more resources to whites or Asians, especially white or Asian males with high test scores who may not have gone to college. While these arguments are phrased as both logical and obvious, LW rarely (ever?) entertains the easily constructed, similarly phrased arguments that would push resources away from LW’s typical membership. For example: “It’s been known for generations that physical strength has a positive impact statistically on outcomes in basically every sort of violent encounter, so as a default, in a world where couples and families could be attacked, people should assume a necessity for bigger, more muscular men as romantic partners.” Or how’s this: “It’s been known for generations that religious identification with the in-group eases working relationships and obviates friction over expressions of belief, so employers should as a default prefer employees share their religions.”
Very possible. I would take the Steve Sailer approach here, of acknowledging underlying differences and making the best of the situation. Let’s step away from race and just talk about tracking in schools- by the time someone is 12, we have a pretty good guess what their eventual social strata / broad kind of career will be.
In countries like Germany, they respond to this with different high schools- someone who will be a technician can go to a technical school, and someone who will be an engineer can go to an academic school. Both get work suited for their intellectual ability and interests, and so the first isn’t drowning and the second isn’t bored. (Relevant here is the finding that getting rid of shop classes increases the high school dropout rate in America- turns out that for an easily identifiable group of students, the primary benefit they get out of high school is a place to practice basic handyman skills!)
In the US, we get lunacy like “whether or not someone takes the first optional math class is a very strong predictor of whether or not they go to college. Let’s make that class mandatory for graduating high school!” which makes everyone involved worse off, as the students not pointed at college now find it more difficult to graduate high school.
I’m not sure this would see significant disagreement here on LW. The main response I would give is yes, but the preference is miscalibrated. Ceteris paribus, a stronger partner is likely to be better (assuming they aren’t prone to domestic violence), but my reflective preferences would give a weight to athleticism that’s orders of magnitude lower than the weight my attraction heuristics give athleticism. This mismatch seems to be because those heuristics were tuned in an era when the chance of being the victim (or beneficiary!) of violent crime was orders of magnitude higher than they currently are.
Of course it has. But the issue is that the society isn’t going to come out and say that—it will deliberately distort the map and make claims that are not true in reality.
The argument being made isn’t “50% of people are below median intelligence, we still need to and can productively employ them”, the argument is “we will pretend that all groups of people are exactly equally smart and if you say otherwise we’ll sue your ass into the ground”.
Nope, not true since Mr.Colt made an equalizer :-) But I’ll agree that firearms training and ownership can be a reasonable plus in looking for a romantic partner. Well, unless his name is Pistorius...
So, reason dictates that… “we” should shove our offended senses of intellectual consistency and naively understood “honesty” up our collective butt, and just do whatever helps people.
And we should absolutely not help people “equally”! Whatever you think of the abstract moral/political ideal of equality, in practical terms people’s circumstances in any society are so unequal that symmetrical treatment makes no sense. Any policy that does not identify the most vulnerable and marginalized groups and offer them targeted aid and protection is not “fair”, it’s not “impartial”… it’s basically a waste of resources, failing to seek out the greatest marginal utility for its subjects. So, ironically, it becomes a core left-wing idea that people should not be approached as identical or treated in an “equal” manner. Bam!
“In its majestic equality, the law forbids rich and poor alike to sleep under bridges, beg in the streets and steal loaves of bread.”—Anatole France on “legal equality”
I think your problem is that you’re so focused on the “fairest” way to divide a fixed set of goods, that you’re forgetting that the decisions in question also have a large effect on the amount of goods available.
That’s actually an interesting argument. I wouldn’t mind seeing it expanded, if you happen to have real numbers lying around.
Though some obvious confounders do come to mind: in a really diverse religious environment (like, for example, the Silicon Valley tech scene), you’re giving up quite a bit in talent if you recruit only from your co-religionists. And if you weight it less heavily, I’d be very surprised if the response looked linear: I wouldn’t expect a workplace that’s (say) 50% Christian with the rest split between atheists, Hindus, and Buddhists to be that much more harmonious than one with equal numbers of all of the above plus the odd Wiccan or Discordian. It might actually be worse under some circumstances, although this is rank speculation.
A lot of the current research focuses on “trust” inside groups. This is not exactly double-blinded climate controlled stuff, as you might expect, just brave and smart social psychologists doing their best. I find it highly plausible and confirmatory of many centuries of non-scientific observations about insularity. Disclaimer: I AM NOT SAYING DISTRUST OF PEOPLE OF OTHER BELIEF SYSTEMS IS GOOD, JUST THAT IT HAPPENS.
Atheism associated with lack of “trustworthiness signals” by believers.
Religious in-group trust and cooperation is higher.
I know of no studies on friction over expression of religious beliefs. I do kind of take as a given that there are fewer HR complaints when everybody’s got the same Sacred Heart/Darwin amphibian/Santa Muerte/COEXIST bumper sticker.
Granted that there are huge trade-offs for religious homogeneity, and I think that it’s almost always a bad business decision (exceptions: semi-utopian communes? survival in Hobbesian chaos? new colonies without hope of reinforcement?) It was just an exemplary argument of a sort made less often than, you know, arguments about race and IQ.
To refer back to the OP, why is the relevant disadvantaged class “black people” rather than “people with low IQ” or even “people unqualified for the job”?
As far as it goes, I’m in favor of preserving opportunities for all sorts of people to work, because it’s humanizing and it makes people happy. We’re all in favor of that, right?
But I also don’t think there’s been historic, organized pressure to keep low-IQ people from finding useful labor, and while such people deserve the protection of the law, it’s not illuminating to compare their plight to a group of people who were denied the ability to find employment they were very capable of using intimidation, violence, and bad-faith law....
...tools which, and this is sad, were very much still in use when the Civil Right Act was passed, and would still be in use today if it had never been passed.
Racial “classes”—not sets of corresponding genetic polymorphisms, which science tells us about, but race as we understand it in America, which is both more and less complicated—were not created by the Civil Rights Act, or the civil rights movement. They were created long before that, to justify cruelty, and to deny the continuing effect of that social construction would have been, in the the judgment of the majority of our Congress in 1964 and our Supreme Court since then, counterproductive.
Not at all. It’s a very rationalist sort of argument. There are many like it. I think it would be terrific if we spent more time exploring those, possibly at the expense of focusing heavily on arguments that seem a little less than disinterested.
Yes, and what is the justification for the disparate impact doctrine?
And for that matter what is the justification for declaring certain classes “protected”?
Are you asking rhetorically?
The American legal justification for the disparate impact doctrine, and for declaring race a protected category, is the 1964 Civil Rights Act, and the legislative justification for that was a history of massive mistreatment of individuals based on skin color.
I gather from the thrust of arguments in this thread that you may be strongly opposed to government protection of racial minorities in the United States, and that you may not believe that racial bigotry is—or possibly even was—a problem that needed legal redress. It is worthwhile to note that the legal basis for these doctrines is well established and, through the wonders of litigation, much studied and highly nuanced. That does not speak to any philosophical objections you have but, frankly, no philosophical objections you make have any bearing on the legal justification.
Um, the legal basis is the act of Congress. That’s all, you don’t need studies and nuances. Whatever Congress says and the President signs is the law of the land. Unless SCOTUS objects, of course.
This is a somewhat fundamentalist view of the law, and I am guessing many federal judges at all levels, and regulatory bodies of technical experts, would add something to your definition. I agree with you that the statutory basis for these court rulings is very clear.
But it’s also pretty clear that the doctrine of disparate impact, which is what he asked about, has been clarified and nuanced through litigation of those statutes. My point was that over many decades, the courts have not overturned this doctrine due to any philosophical objections of litigants.
Yes, of course, though it has nothing to do with legal basis—it’s interpretation of the law which is what the court system does all the time.
Courts do not do that. A philosophical objection is not a legal objection—a court can overturn a law only by deciding that it is unconstitutional.
But I am unsure what is the point that you are making. Is it that both politically and legally the Civil Rights Act is untouchable in the US? Sure, but that’s pretty obvious…
Sorry, I meant the two questions in different senses, I should have made that clearer.
The Civil Rights Acts didn’t specify disparate impact as opposed to disparate treatment.
I understand the motivation, but I don’t think the ever increasing (and rather arbitrary) list of protected groups is a workable approach. Not to mention the “some groups are more equal than others” problem implicit in having a specific list of “protected groups”.
If you look at the history of law, philosophical arguments end up influencing legal arguments all the time.
I absolutely agree. It is conceivable that in the future, arguments could change the courts’ regard for this doctrine. But it is unlikely. The law has been in place for fifty years, and the doctrine has seen a ton of challenges in court.
So? Far older legal doctrines have been overturned by courts.
I said it was conceivable but unlikely. You disagree?
Unlikely, over what timescale? Yes, I agree this is unlikely to change next year.
Actually, the most useful application for individual businesses in this case would be (in the event that IQ tests are good at predicting who will be a good worker) to request IQ scores as part of a job application, not to discriminate based on race—this is not to say that it would be useful for society as a whole. I am not sure what it would do to society as a whole. On the one hand, if there’s a correlation between race and IQ, more people of each race with a low IQ might find themselves worse off. However, if employers become more willing to hire black people after testing their IQs, it could be a great boon to blacks and actually serve as a way to encourage people to judge each person based on individual characteristics as opposed to rejecting them for being part of a group. Simply tossing away all of the people of whatever race because the others have a low IQ would probably, in practice, not work very well—this is because they’re selecting from a pool of people who are qualified in the first place, and the process of becoming qualified acts as a filter. Most of the people they’re interviewing who are qualified are probably also intelligent enough to do the job—so you’d have too man false negatives this way.
I would say also that if aptitude tests are restricted because of racist connotations, it’s because people tied IQ to race.
Can you think of any applications for tying IQ to race that do not have the above issues?
The problem is that this is currently illegal in the USA. This was in fact precisely the point of the parent comment.
When is this ever the case in practice?
The Bell Curve book is a standard source. Otherwise a quick look at Wikipedia provides this:
Ok thanks. However, I am aware that “most published research is wrong” (PLOS Medicine) and know that there are factors that need to be controlled for in studies on race and IQ (in the second numbered list). Do you also claim that these factors were controlled for, that the key study or studies have been replicated, and that this is quality data that generally avoids research pitfalls? That’s what I am looking for.
Yes, I’ve read Ioannidis. However you’re using this quote here as a rather blatant aid to your confirmation bias. There have been many, many studies which all show the same thing. These are findings which have been confirmed, re-confirmed, and confirmed once again.
Precisely because these results are so controversial they have been the subject of very thorough checking, vetting, and multiple attempts to debunk them. The results survived all this. What, do you think that for the last 50 years no one really tried to find holes in the studies showing racial IQ differences? Many highly qualified people tried. The results still stand.
I think everyone should consider that published research findings are likely to be wrong each time they are seeking research findings. If you agree that we should be skeptical about research findings, why do you think that asking questions about whether the research controlled for multiple factors, was replicated etc. should be taken as evidence of confirmation bias? Maybe you disagree that we should be skeptical about research findings?
Every single one? I would find that hard to believe for any topic, especially one as politically charged and controversial as this one, where both sides have a motive to bias research in their particular direction. If that is true, I would find it surprising. Assuming you were referring to the results of a meta-analysis, would you point to that meta-analysis please?
Are you saying that studies used for “The Bell Curve” did take into account the factors I mentioned, were replicated and / or may contain a meta-analysis that states that all the studies that could be found had similar findings?
If you aren’t specific about what measures were taken to ensure quality in the information you’re providing, I have no way to make the distinction between a matter of opinion and a matter of fact when you claim things like “The results survived all this.” Please be specific about what particular quality features the data in The Bell Curve provides.
I started checking out this book because of your high praise and was surprised to find this:
On page 270, The Bell Curve clearly states: “The debate about whether and how much genes and environment have to do with ethnic differences remains unresolved”
Can you explain why you seem to be disagreeing with me, when both myself and The Bell Curve agree that we don’t have a good way to tell whether IQ differences are nature or nurture? (Note: In addition to that, my view is also influenced by skepticism about research in general and an understanding that although IQ tests are correlated with various things, they have some limitations.)
That there are population level differences in IQ is not controversial (except in the sense that evolution is controversial because more than 30% of Americans don’t believe in it).
That IQ is a useful proxy for general intelligence and a useful tool in determining life outcomes is not controversial.
That IQ is heritable is not controversial.
That the differences are genetic is controversial but the data does seem to suggest that much of the difference is indeed genetic and another portion is biological (pre-natal care, early nutrition).
http://occidentalascent.wordpress.com/2012/06/10/the-facts-that-need-to-be-explained/
Every objection you’ve listed so far has been addressed in exhaustive detail in the above link.
Sigh.
I don’t believe you’re listening and really have no inclination to play the “yes, but” game. Neither do I feel the need to prove anything to you.
You can believe whatever you want to believe, it’s just that such an attitude looks strange here.
That is not my attitude. I have been asking you for research. Did you see what I discovered about “The Bell Curve”? What do you say about that?
Did you read the two paragraphs following your quoted sentence? It seems to me that they more or less settle the matter, and resolve your grayness regarding environment and genetics.
What I suspect is going on is that Epiphany is statistically innumerate (as suggested by her rather hilarious statement about 25% of Africans being above average), but doesn’t want to loose status by admitting she doesn’t understand the arguments.
Both of the citations I was given by you guys said clearly that they were uncertain about the connection between race and IQ. That is the reason I don’t agree—because even your citations do not agree. I assume those are the best citations you have, so that your citations do not agree with you makes your belief look very bad indeed.
Also, by arguing that the reason I don’t agree is because I am statistically innumerate and that the reason I don’t agree is because I’m too inept to understand, you have made an ad hominem fallacy. Attacking the person does zilch to support your argument.
I can’t believe I just saw an ad hominem attack on LessWrong. That is the the most obvious behavior that one avoids if one wants to have a rational debate.
I don’t think you’re correctly distinguishing between multiple related claims.
Claim A is that IQ distributions vary by race. This is supported by mountains of evidence. This alone is sufficient to justify using race as a factor when predicting IQ. This is the original point under discussion, as you argued against using race as a factor when predicting IQ both here and here.
Claim B is that differences in measured IQs overestimate the actual differences in intelligence or life outcomes between races. There is substantial evidence against this claim, and it is only weakly related to the original point under discussion. (Were it true, it would suggest that estimating IQ is not as important when doing between-race comparisons as other estimations, but does not impact IQ estimation.)
Claim C is that X% of difference in racial average IQ is due to genetic factors. It is currently not clear what X is for any particular between-race comparison, which the citations reflect. This is unrelated to the original point under discussion.
Not quite. The arguments you’ve made recently are mostly social arguments- “you say Y, but your citation says Z, how do you account for that!”- rather than technical arguments.
The arguments seem vacuous to a technical expert, because Y and Z turn out to be totally compatible, but may still seem impressive to a non-expert who is unfamiliar with the relationship between Y and Z. Similarly, a non-expert doesn’t know what sort of claims do and don’t need citations, and so may see a virtuous skeptic against credulous believers, rather than a crank who defends their perpetual motion machine by insisting on a citation for the 2nd Law of Thermodynamics. (This is not to say that the 2nd Law of Thermodynamics is taken on faith; what it means is that there are some issues where ignorance, confusion, or denial is solid evidence against being a curious and open-minded scholar.)
An effective social response is to poke at the technical content of your claims at a much more basic level (i.e. charges of innumeracy). If you aren’t familiar with means and medians, and disengage when someone presses you on a basic statistical point, then anyone who is familiar with statistics can use that to gauge your level of technical ability. (And if you aren’t familiar with stats 101, why have a confident opinion on an inherently statistical topic?)
Contrast this with Epiphany’s comment earlier on
I guess it mostly comes down to “race” being such a charged term. Epiphany seems to be fighting “Someone being an African American predicts a lower than US-average IQ” because that has a lot of negative connotations, whereas you may reply “Well, but it’s technically correct, even if did turn out (unlikely/no idea) the correlation only exists because race predicts/includes poverty/culture/social customs, which in turn causally [e|a]ffect IQ”, while she would say “But then it’s not race!”.
Maybe something like “I refuse to use race as a predictor (even if I could) because that’s mindkilling/misleading, unless poverty/culture/social customs have all been controlled for. Since there is no scientific consensus on race when poverty/culture/social customs have all been controlled for, we shouldn’t speculate and rather work on the latter confounders, which can probably be improved with socially acceptable policies.” would capture most of her point and be more palatable?
Sorry for the verbosity.
I agree that there are a handful of valuable points one can make about the underlying causal diagrams. Suppose the diagram is X<-Y->Z: then the correlation between X and Z is indirect and acausal, where we should not expect that modifying X or Z will modify the other one. If the diagram is instead X->Y->Z, then the correlation between X and Z is indirect but modifying X may modify Z.
I disagree that those subtle points are the ones under discussion. It seems to me that this discussion is indirectly about the social acceptability of noticing the correlation between race and IQ, and that the technical points are used mostly for obfuscation. Suppose we were uncertain which of the two causal diagrams I described above were correct; we would still be certain that X and Z were correlated if Y is unmeasured, because that is the case in both causal diagrams, and responding to the claim that X and Z are correlated with “we don’t know if X causes Y or Y causes X” is irrelevant.
When I read “there is no scientific consensus,” it sounds like “but there’s still a chance, right?”
I would agree that it’s easier to intervene in wealth, culture, and customs than intelligence, and that we already know of several obvious ways where interventions in those three can positively impact intelligence. I don’t think Epiphany and I would agree on what interventions are most beneficial there, but that’s a separate conversation that’s not worth having here.
Because for Epiphany it’s not a statistical topic, it’s an issue of fairness and equality—she fights for Great Justice!
No doubt this, too, is an ad hominem, in some sense.
Arguments ad hominem are inappropriate for deciding the truth of the matter. They are entirely appropriate for deciding whether you want to take someone seriously or even just to talk to a person.
We are in agreement.
You might be in a situation where you need to decide how to allocate money to help a black school or a white school. If white people have higher IQs, and if money is worse at improving the performance of students who do poorly because of IQ than it is at improving the performance of students who do poorly for other reasons, then you should allocate the money to the white school.
You might be in a situation where you need to hire a white person or a black person and have no information about their IQs, but you would prefer an employee with a higher IQ. You then should hire the white person.
Of course, this is exactly why using IQ this way is a bad idea.
How often do you know someone’s race but nothing else whatsoever about them?
What?
It’s an additional piece of information and allows you to do more of an update. If black people, on the average, have lower IQs than white people, then (for instance) black people with college degrees are still likely to have lower IQs than white people with college degrees, even if college raises the likely IQ for both groups. It is also possible to have traits such that given those traits race is no predictor at all, but those would be balanced out—if race is no predictor of IQ among people with PhDs, it must be a stronger predictor for people without PhDs than it is for the general populace.
Not necessarily. You haven’t accounted for differences in variance between the two groups.
To make a very rough analogy: it seems widely believed that there is greater variance in intelligence among men than among women; surely such differences are imaginable among racial groups. (For one thing, there’s more genetic variation among Africans than among other human populations — which makes sense, given that all other human populations descended from small subsets of Africans.)
Nor for selection effects on who gets to go to college — for instance, there seem to be a lot of pretty dopey people who have unusual bonuses to their chances to get into college on account of their parents being wealthy.
Nor for socioeconomic differences in general, which are substantial between whites and blacks in America. To make another very rough analogy: If two people reach the same measurement of achievement, but one has to overcome greater obstacles to get there, we would often take this as an indicator that that person had greater ability: a runner who runs a five-minute mile while carrying ten pounds of lead weights is a better runner than one who makes the same achievement carrying no weight.
Yes I have—see the next comment. If blacks with college degrees don’t have lower IQs than whites with college degrees, and blacks in general have lower IQ, then blacks without college degrees must have even lower IQs so that the average is still lower IQ.
Furthermore, taking other factors into consideration can result in worse discrimination. Consider the trait that most obviously makes up for the difference—actually taking an IQ test. Blacks may have lower IQ than whites, but blacks who take an IQ test and score X don’t have lower IQ than whites who score X at all. But if I were to hire people based on the trait “scoring X on an IQ test”, and X is at the high end, I may end up hiring a lot more whites than blacks—a small difference in IQ translates to a large difference in the number of people at the tail end of the distribution. If people with IQ 145 are 10 times as common as people with IQ 150 and the black curve is only shifted by 5 points, and I want to hire people with IQ 150, I may end up hiring whites to blacks at a 10 to 1 ratio compared to the proportion of blacks who apply.
Why would IQ matter more than academic and job performance?
Well, seeing an unknown man approaching you at night, granted this is more about criminality than IQ but the correlation is the same.
Also thinking about whether affirmative action and the desperate impact doctrine are reasonable ideas.
Once you specify where I am, who I am with, what kind of body language the man is using, how big he is, and what he is wearing, further specifying what race he is wouldn’t matter that much.
I can’t recall anyone on LW advocating those, so you might be attacking a straw man.
Is that true? Depending on the “where I am” part?
There’s only so much you can tell about someone from “what kind of body language the man is using, how big he is, and what he is wearing”, after all. In the right racially-segregated society, could it provide valuable additional data?
That’s because they rarely come up. In any case my point is that these doctrines are in place in the USA and the false belief that race is uncorrelated with anything important.
Nobody has yet shown that racial-group data is more correlated with important things than individual data.
What do you mean by “individual data”?
Also, how is this relevant to my point?
I mean data about individuals like resumes and qualifications That racial-group info correlates with important things is unimportant, unless it correlates significantly more than individual data. However, the reverse is the case.
First I don’t understand the distinction your drawing between “individual data” and presumably “group data” since the people with a particular qualification are a group and conversely skin color, say, is a property of an individual.
Back to the point: in the great-grandparent I was talking about affirmative action and the disparate impact. The logic of those is based on concluding that racism happened on the basis of disparate outcomes. This logic relies on the implicit premise that race isn’t correlated with anything important. I don’t see how anything you wrote in your two comments that addresses this issue.
Actually, it is far more prudent to avoid a stranger approaching me at night, regardless of his race—depending on the environment I am in.
If he is approaching from a dark alley, I will head away from him, whatever his race. If he approaches me at a party full of friends, I will speak to him.
The crime statistics are not so incredibly different for blacks and whites that you can simply trust all of the whites.
All of that looks like rationalization of a pre-determined belief.
No it doesn’t. The conclusion supported is far too close to the ‘middle ground’ on the issue to readily pattern match to rationalization and seems to be a description of considered reasoning from plausibly held premises. The ‘rationalisation’ and ‘pre-determined belief’ charges could be credibly made in response to many of the comments in this thread but doesn’t apply to the grandparent.
NOTE: I don’t entirely agree with with either Epiphany’s position or Eugine’s position. In particular Epiphany seems a little too disillusioned with research while Eugine has somewhat too much passion on the race/IQ political correctness subject to keep his claims balanced enough that I could support them despite agreeing that there are almost certainly IQ differences between groups selected by just about any significant feature—not that this seems like an especially useful thing to place emphasis on. I seem to recall some credible claims about higher average mathematical intelligence in Ashkenazi Jews for example.
Regardless of whether I agree with the position being argued with, the parent is making what seems to me to be a false accusation and one of a kind that derails the flow of discourse.
Respectfully disagree, though we may be focusing on different parts. It worries me that your analysis seems to focus on the conclusion drawn rather than the procedures followed.
Consider this bit of the great-grandparent:
This looks like pure rationalization to me, with many inaccuracies and irrelevancies, all of which support the intended conclusion. (An unbiased sloppy thinker should be expected to make mistakes in both directions simultaneously; when the mistakes all point one way it suggests bias.) The most egregious irrelevance is the attempt to discredit one of the most replicated findings in social science with a study that showed that prominently promoted recent research often fails to replicate. The last sentence is bizarre by its addition- would someone using race as Bayesian evidence not update on other information? (The later reference on Bayesian updating is similarly bizarre.)
The rest of the great-grandparent discusses how, even if we had an estimate, we should be careful how we use that estimate. Of course—who would argue against using estimates carefully?--but irrelevant to the question of whether or not it is ethical to use Bayesian reasoning.
Is it inconceivable that this could ever be the case?
It is almost always bad Bayes, or any other kind of reasoning, to make judgements about individuals based on group characeristics, since there is almost always information about them as individuals available which is almost always more reliable.
Group level information is still useful for shrinkage of estimates and correcting for the always-present unreliability in individual estimates; see for example the long conversation between me and Vaniver on LW somewhere where we work through how you would shrink males and females’ SAT scores based on the College Board’s published reliability numbers.
And E(X’s deserved SAT score|X’s measured SAT score; X is male) - E(X’s deserved SAT score|X’s measured SAT score; X is female) was, like, four points? I still think people’s System 1 are likely to overestimate this difference if they know about the correlation more than underestimate it if they don’t.
A 4 point adjustment (or more) across all candidates based solely on 1 binary variable (gender) and a trivial centuries-old bit of statistical reasoning seems like a fairly impressive output, and likely to make a difference on the margin for thousands of applications out of the millions sent each year.
And your evidence for this is...?
For applicants scoring 800 on a hypothetical normally distributed math SAT, yep. For normally distributed tests, shrinkage is linear based on the difference between the group mean and the measured mean, and so it’s smaller for less extreme scores.
(For some reason, I’m having difficulty finding the link to the actual conversation; I think Google search is not going into deep comment threads, and the search function is based off the site, rather than just a database of my comments. Anyone remember helpful keywords further upthread to get a link to the actual conversation?)
Saying “we shouldn’t explicitly calculate something because some people might implicitly calculate that thing incorrectly” sounds to me like going in the exact wrong direction.
Here is Wei Dai’s tool for searching LW comments.