Indeed. At the usual standard deviation norm of ~15, a 220 IQ would be 8 standard deviations out and make him ~1 in 8*10^14 (100 trillion).
Inasmuch as only 100 billion humans are estimated to have ever lived, the overwhelming majority of that having an average IQ far lower than 100 and so being essentially irrelevant, we can conclude that he is either lying or from the future.
220+ IQ scores DO happen—due to the fact that IQ tests cannot be made accurate for such an uncommon group of people, they’re far more common than they’re mathematically supposed to be. A collection of research on that can be found online right here:
I’ve actually talked with people in that IQ vicinity, and based on the absolutely sublime intelligent conversation they’re capable of providing, and considering the likelihood of specifically them being dishonest about that within the context of their other behaviors, I just don’t think they’re lying.
Superintelligent people do exist. And they have to actually BE somewhere, right? Where do they go?
Do you think that none of them would be attracted to a website like LessWrong? I think this site is likely to be a genius magnet.
If it turns out that this person’s IQ really is over 220, I totally want to have intelligent conversation with them. If you give people the benefit of the doubt in situations like this, sometimes the result is more than worth the effort to withhold judgment for a while.
P.S. Yes, I realize the claim is that it was estimated at over 220, not that they received that score. The obvious argument here is “What professional would estimate it that high knowing how rare those scores are SUPPOSED to be?” but if you’re not basing your estimation on observations about people who have received that score, all you are left with is attempting to deduce the characteristics of a person with such an IQ out of the numbers themselves, with no actual experience to base it on. Or, this person may be referring to the practice of adjusting a young child’s IQ score upward in order to reflect the age at which they took the test. For instance, if you are 2 years old and get an IQ of 100 on an IQ test, that’s really incredible. You definitely have to give that kid a higher score than 100. The only way I know of to get a score in the 200 ballpark is to have that sort of age adjustment done after taking the test with the highest limit before a certain age.
220+ IQ scores DO happen—due to the fact that IQ tests cannot be made accurate for such an uncommon group of people, they’re far more common than they’re mathematically supposed to be. A collection of research on that can be found online right here: http://hiqnews.megafoundation.org/Terman_Summary.htm …You definitely have to give that kid a higher score than 100. The only way I know of to get a score in the 200 ballpark is to have that sort of age adjustment done after taking the test with the highest limit before a certain age.
And that’s a limitation of the tests being ratio tests or not being normed on the sufficiently large population they’re supposed to be normed on. (Why are all the datapoints on that page so old?) That’s why modern IQ tests come with listed ceilings! ‘Past this point, who knows what it’s measuring if anything’. With a short test, even random guessing will eventually throw up some remarkable scores...
I’ve actually talked with people in that IQ vicinity, and based on the absolutely sublime intelligent conversation they’re capable of providing, and considering the likelihood of specifically them being dishonest about that within the context of their other behaviors, I just don’t think they’re lying.
Perfectly consistent with them having more earthly IQs >140. (If even that; I have been reading up on psychopathy lately, and one of the diagnosable traits is being gifted conversationalists and creators of emotional ‘bonds’, despite psychopathy being, if correlated with IQ at all, negatively correlated.)
Living people with 220+ IQs do exist—you say the words “modern IQ tests” as if the ratio tests were invented in the dark ages. This only changed in recent decades. Regardless of which method is the best, the fact that there are plenty of people still alive today who can honestly claim that they were given an IQ score that high means that this person is not automatically a liar or “from the future”.
“Perfectly consistent with them having more earthly IQs >140”
Which is perfectly consistent with them having IQs over 220, if you think about it...
And, no, they weren’t like the people with scores of 140. There are differences that they have that I have not encountered in anyone else. Things stood out.
Why are you bringing up psychopathy? That’s totally out of left field. Do you mean to imply that people claiming that IQ are probably psychopaths?
Living people with 220+ IQs do exist—you say the words “modern IQ tests” as if the ratio tests were invented in the dark ages. This only changed in recent decades.
Very well then, let us discuss the cases of people recorded to be hundreds of years old by less than modern documentation, like Methuselah as verified by the Book of Genesis. Wait, you don’t think people actually live to thousands of years? But you just said we can use datapoints from any kind of test we please!
Whatever cutoff you choose to exclude things like Genesis or scientific results from hundreds of years ago while still including largely obsolete ratio tests, I will shift it slightly to include only better IQ tests. I think this is perfectly legitimate, as one should strive to use the best available data, and regard your ‘but old obsolete scores!’ as quibbling.
Which is perfectly consistent with them having IQs over 220, if you think about it...
And it’s also consistent with IQs over 9000!!!
Occam’s razor. Use it, love it. The base rate of IQs like 140 are by definition higher than >220.
There are differences that they have that I have not encountered in anyone else. Things stood out.
“But I was so impressed, don’t you understand?” You’ll pardon me if I ignore some rubbish anecdotes about them seeming like shining special snowflakes.
Why are you bringing up psychopathy? That’s totally out of left field. Do you mean to imply that people claiming that IQ are probably psychopaths?
My argument was perfectly clear: brilliant conversation is far from a flawless indicator of intelligence. That you don’t understand why I would bring up an example of how this indicator can fail catastrophically or interpret it as implying that...
More fun base-rate reasoning: psychopaths make up 1-2% of the population, and most are great manipulators; the top 1% of the population IQ-wise is sometimes taken as being the genius fragment; even if we assume the 1% IQ are all gifted conversationalists, if all we know about someone is their gifted conversation, we wind up inferring that they are equally or more likely to be a psychopath than a genius!
Looking closer, I think there are several points of confusion. Neither of us carefully distinguished between meanings like “An IQ score / estimation that really was that high but not accurate.” versus “An IQ score / estimation that high that is NOT accurate.” and I guess that neither of us noticed the possibility for multiple meanings. We also did not address the possibility that this person’s (inaccurate) IQ could be that high while incidentally, the person does have an intelligence level to truly match an IQ of 220. That sort of person would be more likely to get an IQ of 220 on a test/estimate, no? There may be people with that (true) IQ who also have a nearby (inaccurate) IQ result to match it. After all, that is what the developmental psychologists are aiming for—that they’re right sometimes but not all the time is possible.
Given that IQ is a measurement of intelligence, and there is no way to measure a number that is that high, all 4 apply. If you are still confused about the meaning of the verb “measure”, here it is:
to ascertain the extent, dimensions, quantity, capacity, etc., of, especially by comparison with a standard
Agreed: You can’t create an accurate IQ test for that.
Disagreed: You can’t generate an estimate IQ that high. Estimates can be made, but their accuracy can’t be verified. Estimates are, by definition, an approximation. An estimated IQ that high is not automatically invalid, as long as it’s expressed as an estimate, the statement can still be true.
Disagreed: None of the IQ tests developed provide any method to generate an IQ that high. Some IQ tests can be adjusted upward based on a child’s age. I’m not claiming it’s accurate. I’m claiming it exists.
Disagreed: You can’t have an IQ that high. Saying a person can’t get an IQ score that high is not the same thing as saying they can’t get an IQ that high. You can be born with an intelligence level beyond what is statistically probable. Whether it can be measured may be another story. But it can still happen. William Sidis is my cite for that—his estimated IQ was 250-300. If you read enough about developmental psychology, you’ll probably agree that Sidis not only qualifies as a person having abilities consistent with what an person with an IQ of > 220 should have, but that he had such great abilities that by using him as an example I am doing something more along the lines of killing an ant with a nuke. You can claim that because it’s improbable, it can’t happen, but that’s would be an appeal to probability (or a reverse of that appeal, if you want to be really technical.)
Since IQ scores are calculated to conform to a normal distribution, your IQ is in exact correspondence with your percentile ranking in the general population. If f(x) is the CDF of the Normal(100,15) distribution, and your percentile ranking is p, then your “true IQ” is precisely the value of x such that f(x)=p. Since there are only roughly 7 billion people alive, the smartest person on Earth is ranked above 99.99999998578% of other people, which corresponds to an IQ of 195. If we include all people that have ever lived (I believe the number is roughly 100 billion), the smartest would have an IQ of 201.
William Sidis’s IQ of 250-300 is using a different scale, which is no longer used: it claims that his intellectual age, as a child at the time he took the exam (if in fact he did, there is some doubt about this) was 2.5 to 3 times his physical age. This isn’t really a meaningful assessment of how smart he was as an adult, nor does it have any relationship to the modern system of IQ scores (on a normal curve, such a score is 1 in 10^23, which I think we all agree is ridiculous).
I suppose that if you mean that it is possible to make an invalid extrapolation based on age adjustment, then you are right, it is indeed possible, if meaningless. It is not, however, how IQ is measured (by comparing how well you did vs other people who took the same test).
Disagreed: You can’t have an IQ that high. Saying a person can’t get an IQ score that high is not the same thing as saying they can’t get an IQ that high.
It is exactly the same thing, because IQ is not intelligence, it’s one (not very accurate) way to measure it. Thus there is no such thing as person’s IQ, only person’s IQ score. Just because IQ is commonly confused with intelligence, does not mean it is the same thing.
You can estimate someone’s IQ testing skills from one or more of their IQ test scores, and this skill is indeed a property of (your model of) the individual, not of a piece of paper with the number on it, and this skill is correlated with other measures of intelligence. This skill can conceivably be shorted to “person’s IQ”. However, there is no standard procedure of calculating anything like that (should it be the average? mode? geometric mean? maximum? minimum? any of those have merits, depending on your model of how the hypothetical innate IQ testing skill is translated into IQ test scores).
William Sidis is my cite for that—his estimated IQ was 250-300
Funny that you bring him up. Says Wikipedia:
never before have I found a topic so satiated with lies, myths, half-truths, exaggerations, and other forms of misinformation as is in the history behind William Sidis
I suppose that if you mean that it is possible to make an invalid extrapolation based on age adjustment, then you are right, it is indeed possible, if meaningless.
It is possible, and they’ve done it. Whether it is meaningless is outside the scope of our particular debate as, if I remember correctly, we started arguing after Gwern said that claiming an IQ of 220+ is proof that a person is “either a lying or from the future”.
So, this supports my point that such a claim is not proof that a person is lying. I’ll just disregard the “from the future” comment for now. ;)
It is not, however, how IQ is measured (by comparing how well you did vs other people who took the same test).
Not sure if you’re saying “IQ test scores aren’t generated based on how you compare.” or, within the context of the previous sentence “IQ tests are not scored using age adjustment.”
For the former, you’re partly right and partly wrong. The ratio tests were scored using a bell curve. If your IQ was relatively close to normal, you’d get a score that would tell you how you compare to average. If your IQ was super high like 160, it was likely to be inaccurate, because those people are rare.
As for age adjustments—of course they make age adjustments. Otherwise, how would they test children of different ages on the same test? If a child of the age of 3 gets the same questions right as children who are 10 years old, do you give the ten year olds a toddler’s score or do you give the toddler a score closer to that of a ten year old? It would be inappropriate to imply that the ten year olds and the toddler have the same amount of intelligence.
I imagine this is fairly meaningful at least when it concerns testing average children of different ages, since it’s not too difficult to find lots of children to make the test more accurate with. When it comes to testing child prodigies, the exact score (say “exactly 223” or something) would be meaningless, but the fact that, say, a 7 year old got a perfect score on an IQ test suitable for adults, we’ll say, that would be very meaningful—though their score should be taken as more of a ballpark figure than an exact measurement.
Saying a person can’t get an IQ score that high is not the same thing as saying they can’t get an IQ that high.
It is exactly the same thing, because IQ is not intelligence, it’s one (not very accurate) way to measure it. Thus there is no such thing as person’s IQ, only person’s IQ score. Just because IQ is commonly confused with intelligence, does not mean it is the same thing.
It depends on how you use the word in the sentence. You make a distinction between IQ and intelligence which is good, but I am making a different distinction. Even if I haven’t measured the number of degrees Fahrenheit in a particular igloo near the North pole, that does not mean it has no temperature or that it’s temperature does not correspond to a specific number of degrees Fahrenheit. This is more like the debate “If a tree falls in a forest does it make a sound?”—my answer is yes because I’m using a definition that involves physics, disrupted air waves and decibels. Just because you didn’t measure the number of decibels doesn’t mean they weren’t there.
the point
Anyway, your point is “it’s physically impossible to measure an IQ score that high” but that does nothing to refute my point that “this claim is not proof the person is a liar.”
brilliant conversation is far from a flawless indicator of intelligence
Maybe, but if someone talks to me about quantum field theory and actually makes sense, my posterior probability that their IQ is < 80 suddenly goes down to epsilon.
But how do you know that? Plenty of nutters sound convincing on quantum matters (as judging by the sales into millions of such folk as Deepak Chopra and abominations like The Dancing Wu-li Masters), so I assume you have some expertise in the matter—and now you’re just judging based on that. (And what if they sound convincing on a topic you have no expertise in...)
I think one of the main reason they “sound“ convincing (though the readers’ ignorance is also a necessary condition) is motivated cognition: the kind of people who read such books would like to believe what they say. Lose that, and your strength as a rationalist kicks in. (And anyway, I don’t think Chopra et al. are idiots; they are either misguided or bullshitting the readers for fun and profit.)
And what if they sound convincing on a topic you have no expertise in
I’d have to test that. Anyone willing to give me a few paragraphs of either something “serious” or crackpottery (or a spoof à la Sokal), without telling me which it is, about a topic other than physics?
More fun base-rate reasoning: psychopaths make up 1-2% of the population, and most are great manipulators; the top 1% of the population IQ-wise is sometimes taken as being the genius fragment; even if we assume the 1% IQ are all gifted conversationalists, if all we know about someone is their gifted conversation, we wind up inferring that they are equally or more likely to be a psychopath than a genius!
I’m actually somewhat curious about the degree overlap between those two groups.
Well, assuming complete independence would just be 1% 1%; but there does seem to be a slight negative correlation between psychopathy & IQ. Complicating matters is Hare and Babiak’s research into business psychopaths, where they estimate they are over-represented by a factor of 2-3 or so, suggesting that the negative correlation may be skewed by the ‘failures’ in prison samples (which are the samples for most studies, for obvious reasons); smarter psychopaths are far less likely to resort to violence\*, further hindering identification (since impulsive violence is one of the major diagnostic hallmarks). To the extent that the gifted 1% avoid business, that may restore a negative correlation / underepresentation. Finally, psychopath’s impulsivity and few long-range goals or efforts (another part of the diagnosis along with glibness/manipulation) suggests that to the extent genuine objective achievements cause you to be considered a gifted 1%, we can expect still more underrepresentation*.
So guesstimating further, I’d say in a population of 300m people (eg. the US), we could expect substantially fewer than 30,000 gifted psychopaths (0.01 0.01 300,000,000). Phew!
On the other hand, they would be the ones who would do the most damage and be least likely to ever be diagnosed, so we may never know for sure...
* Which makes me wonder about high IQ societies, now that I think about it. My vague impression was that they tend to collect those with poor social skills, and also with fewer objective accomplishments & success. So if you meet someone in a high IQ societies who seems very charming and empathetic but lacks objective accomplishments, just how much does this increase the psychopath possibility over the 1% base rate..? ** Covered multiple times in the Handbook; first relevant paper seems to be “Psychopathy and Aggression”, Porter & Woodworth.
IQ is defined to be a normal distribution with mu = 100 and sigma = 15, so “IQ 220” means ‘99.9999999999999th percentile¹’; if more than a person in 10^15 gets such a score, then the test is miscalibrated. (But most tests are, beyond a few standard deviations away from the mean.)
I didn’t count the nines, I just copied and pasted the output of pr norm(8) in gnuplot and moved the decimal point.
Indeed. At the usual standard deviation norm of ~15, a 220 IQ would be 8 standard deviations out and make him ~1 in 8*10^14 (100 trillion).
Inasmuch as only 100 billion humans are estimated to have ever lived, the overwhelming majority of that having an average IQ far lower than 100 and so being essentially irrelevant, we can conclude that he is either lying or from the future.
220+ IQ scores DO happen—due to the fact that IQ tests cannot be made accurate for such an uncommon group of people, they’re far more common than they’re mathematically supposed to be. A collection of research on that can be found online right here:
http://hiqnews.megafoundation.org/Terman_Summary.htm
I’ve actually talked with people in that IQ vicinity, and based on the absolutely sublime intelligent conversation they’re capable of providing, and considering the likelihood of specifically them being dishonest about that within the context of their other behaviors, I just don’t think they’re lying.
Superintelligent people do exist. And they have to actually BE somewhere, right? Where do they go?
Do you think that none of them would be attracted to a website like LessWrong? I think this site is likely to be a genius magnet.
If it turns out that this person’s IQ really is over 220, I totally want to have intelligent conversation with them. If you give people the benefit of the doubt in situations like this, sometimes the result is more than worth the effort to withhold judgment for a while.
P.S. Yes, I realize the claim is that it was estimated at over 220, not that they received that score. The obvious argument here is “What professional would estimate it that high knowing how rare those scores are SUPPOSED to be?” but if you’re not basing your estimation on observations about people who have received that score, all you are left with is attempting to deduce the characteristics of a person with such an IQ out of the numbers themselves, with no actual experience to base it on. Or, this person may be referring to the practice of adjusting a young child’s IQ score upward in order to reflect the age at which they took the test. For instance, if you are 2 years old and get an IQ of 100 on an IQ test, that’s really incredible. You definitely have to give that kid a higher score than 100. The only way I know of to get a score in the 200 ballpark is to have that sort of age adjustment done after taking the test with the highest limit before a certain age.
And that’s a limitation of the tests being ratio tests or not being normed on the sufficiently large population they’re supposed to be normed on. (Why are all the datapoints on that page so old?) That’s why modern IQ tests come with listed ceilings! ‘Past this point, who knows what it’s measuring if anything’. With a short test, even random guessing will eventually throw up some remarkable scores...
Perfectly consistent with them having more earthly IQs >140. (If even that; I have been reading up on psychopathy lately, and one of the diagnosable traits is being gifted conversationalists and creators of emotional ‘bonds’, despite psychopathy being, if correlated with IQ at all, negatively correlated.)
Living people with 220+ IQs do exist—you say the words “modern IQ tests” as if the ratio tests were invented in the dark ages. This only changed in recent decades. Regardless of which method is the best, the fact that there are plenty of people still alive today who can honestly claim that they were given an IQ score that high means that this person is not automatically a liar or “from the future”.
“Perfectly consistent with them having more earthly IQs >140”
Which is perfectly consistent with them having IQs over 220, if you think about it...
And, no, they weren’t like the people with scores of 140. There are differences that they have that I have not encountered in anyone else. Things stood out.
Why are you bringing up psychopathy? That’s totally out of left field. Do you mean to imply that people claiming that IQ are probably psychopaths?
Very well then, let us discuss the cases of people recorded to be hundreds of years old by less than modern documentation, like Methuselah as verified by the Book of Genesis. Wait, you don’t think people actually live to thousands of years? But you just said we can use datapoints from any kind of test we please!
Whatever cutoff you choose to exclude things like Genesis or scientific results from hundreds of years ago while still including largely obsolete ratio tests, I will shift it slightly to include only better IQ tests. I think this is perfectly legitimate, as one should strive to use the best available data, and regard your ‘but old obsolete scores!’ as quibbling.
And it’s also consistent with IQs over 9000!!!
Occam’s razor. Use it, love it. The base rate of IQs like 140 are by definition higher than >220.
“But I was so impressed, don’t you understand?” You’ll pardon me if I ignore some rubbish anecdotes about them seeming like shining special snowflakes.
My argument was perfectly clear: brilliant conversation is far from a flawless indicator of intelligence. That you don’t understand why I would bring up an example of how this indicator can fail catastrophically or interpret it as implying that...
More fun base-rate reasoning: psychopaths make up 1-2% of the population, and most are great manipulators; the top 1% of the population IQ-wise is sometimes taken as being the genius fragment; even if we assume the 1% IQ are all gifted conversationalists, if all we know about someone is their gifted conversation, we wind up inferring that they are equally or more likely to be a psychopath than a genius!
...And able to convince you that they have IQ > 220, regardless of whether it means anything,
I wish I could do that, imagine how much “funding” I could get for my perpetual motion machine.
Some, but apparently not that much.
Damn, back to selling bridges.
Upvoted (also) for this.
Looking closer, I think there are several points of confusion. Neither of us carefully distinguished between meanings like “An IQ score / estimation that really was that high but not accurate.” versus “An IQ score / estimation that high that is NOT accurate.” and I guess that neither of us noticed the possibility for multiple meanings. We also did not address the possibility that this person’s (inaccurate) IQ could be that high while incidentally, the person does have an intelligence level to truly match an IQ of 220. That sort of person would be more likely to get an IQ of 220 on a test/estimate, no? There may be people with that (true) IQ who also have a nearby (inaccurate) IQ result to match it. After all, that is what the developmental psychologists are aiming for—that they’re right sometimes but not all the time is possible.
I thought this was great, BTW:
Note that it is physically impossible to measure IQ > 200:
It’s not clear whether you’re saying:
You can’t create an accurate IQ test for that.
You can’t generate an estimate IQ that high.
None of the IQ tests developed provide any method to generate an IQ that high.
You can’t have an IQ that high.
Given that IQ is a measurement of intelligence, and there is no way to measure a number that is that high, all 4 apply. If you are still confused about the meaning of the verb “measure”, here it is:
Agreed: You can’t create an accurate IQ test for that.
Disagreed: You can’t generate an estimate IQ that high. Estimates can be made, but their accuracy can’t be verified. Estimates are, by definition, an approximation. An estimated IQ that high is not automatically invalid, as long as it’s expressed as an estimate, the statement can still be true.
Disagreed: None of the IQ tests developed provide any method to generate an IQ that high. Some IQ tests can be adjusted upward based on a child’s age. I’m not claiming it’s accurate. I’m claiming it exists.
Disagreed: You can’t have an IQ that high. Saying a person can’t get an IQ score that high is not the same thing as saying they can’t get an IQ that high. You can be born with an intelligence level beyond what is statistically probable. Whether it can be measured may be another story. But it can still happen. William Sidis is my cite for that—his estimated IQ was 250-300. If you read enough about developmental psychology, you’ll probably agree that Sidis not only qualifies as a person having abilities consistent with what an person with an IQ of > 220 should have, but that he had such great abilities that by using him as an example I am doing something more along the lines of killing an ant with a nuke. You can claim that because it’s improbable, it can’t happen, but that’s would be an appeal to probability (or a reverse of that appeal, if you want to be really technical.)
Since IQ scores are calculated to conform to a normal distribution, your IQ is in exact correspondence with your percentile ranking in the general population. If f(x) is the CDF of the Normal(100,15) distribution, and your percentile ranking is p, then your “true IQ” is precisely the value of x such that f(x)=p. Since there are only roughly 7 billion people alive, the smartest person on Earth is ranked above 99.99999998578% of other people, which corresponds to an IQ of 195. If we include all people that have ever lived (I believe the number is roughly 100 billion), the smartest would have an IQ of 201.
William Sidis’s IQ of 250-300 is using a different scale, which is no longer used: it claims that his intellectual age, as a child at the time he took the exam (if in fact he did, there is some doubt about this) was 2.5 to 3 times his physical age. This isn’t really a meaningful assessment of how smart he was as an adult, nor does it have any relationship to the modern system of IQ scores (on a normal curve, such a score is 1 in 10^23, which I think we all agree is ridiculous).
I suppose that if you mean that it is possible to make an invalid extrapolation based on age adjustment, then you are right, it is indeed possible, if meaningless. It is not, however, how IQ is measured (by comparing how well you did vs other people who took the same test).
It is exactly the same thing, because IQ is not intelligence, it’s one (not very accurate) way to measure it. Thus there is no such thing as person’s IQ, only person’s IQ score. Just because IQ is commonly confused with intelligence, does not mean it is the same thing.
You can estimate someone’s IQ testing skills from one or more of their IQ test scores, and this skill is indeed a property of (your model of) the individual, not of a piece of paper with the number on it, and this skill is correlated with other measures of intelligence. This skill can conceivably be shorted to “person’s IQ”. However, there is no standard procedure of calculating anything like that (should it be the average? mode? geometric mean? maximum? minimum? any of those have merits, depending on your model of how the hypothetical innate IQ testing skill is translated into IQ test scores).
Funny that you bring him up. Says Wikipedia:
It is possible, and they’ve done it. Whether it is meaningless is outside the scope of our particular debate as, if I remember correctly, we started arguing after Gwern said that claiming an IQ of 220+ is proof that a person is “either a lying or from the future”.
So, this supports my point that such a claim is not proof that a person is lying. I’ll just disregard the “from the future” comment for now. ;)
Not sure if you’re saying “IQ test scores aren’t generated based on how you compare.” or, within the context of the previous sentence “IQ tests are not scored using age adjustment.”
For the former, you’re partly right and partly wrong. The ratio tests were scored using a bell curve. If your IQ was relatively close to normal, you’d get a score that would tell you how you compare to average. If your IQ was super high like 160, it was likely to be inaccurate, because those people are rare.
As for age adjustments—of course they make age adjustments. Otherwise, how would they test children of different ages on the same test? If a child of the age of 3 gets the same questions right as children who are 10 years old, do you give the ten year olds a toddler’s score or do you give the toddler a score closer to that of a ten year old? It would be inappropriate to imply that the ten year olds and the toddler have the same amount of intelligence.
I imagine this is fairly meaningful at least when it concerns testing average children of different ages, since it’s not too difficult to find lots of children to make the test more accurate with. When it comes to testing child prodigies, the exact score (say “exactly 223” or something) would be meaningless, but the fact that, say, a 7 year old got a perfect score on an IQ test suitable for adults, we’ll say, that would be very meaningful—though their score should be taken as more of a ballpark figure than an exact measurement.
It depends on how you use the word in the sentence. You make a distinction between IQ and intelligence which is good, but I am making a different distinction. Even if I haven’t measured the number of degrees Fahrenheit in a particular igloo near the North pole, that does not mean it has no temperature or that it’s temperature does not correspond to a specific number of degrees Fahrenheit. This is more like the debate “If a tree falls in a forest does it make a sound?”—my answer is yes because I’m using a definition that involves physics, disrupted air waves and decibels. Just because you didn’t measure the number of decibels doesn’t mean they weren’t there.
the point
Anyway, your point is “it’s physically impossible to measure an IQ score that high” but that does nothing to refute my point that “this claim is not proof the person is a liar.”
It’s a red herring.
Maybe I should verify my info on him then.
Maybe, but if someone talks to me about quantum field theory and actually makes sense, my posterior probability that their IQ is < 80 suddenly goes down to epsilon.
But how do you know that? Plenty of nutters sound convincing on quantum matters (as judging by the sales into millions of such folk as Deepak Chopra and abominations like The Dancing Wu-li Masters), so I assume you have some expertise in the matter—and now you’re just judging based on that. (And what if they sound convincing on a topic you have no expertise in...)
I think one of the main reason they “sound“ convincing (though the readers’ ignorance is also a necessary condition) is motivated cognition: the kind of people who read such books would like to believe what they say. Lose that, and your strength as a rationalist kicks in. (And anyway, I don’t think Chopra et al. are idiots; they are either misguided or bullshitting the readers for fun and profit.)
I’d have to test that. Anyone willing to give me a few paragraphs of either something “serious” or crackpottery (or a spoof à la Sokal), without telling me which it is, about a topic other than physics?
I’m actually somewhat curious about the degree overlap between those two groups.
Well, assuming complete independence would just be 1% 1%; but there does seem to be a slight negative correlation between psychopathy & IQ. Complicating matters is Hare and Babiak’s research into business psychopaths, where they estimate they are over-represented by a factor of 2-3 or so, suggesting that the negative correlation may be skewed by the ‘failures’ in prison samples (which are the samples for most studies, for obvious reasons); smarter psychopaths are far less likely to resort to violence\*, further hindering identification (since impulsive violence is one of the major diagnostic hallmarks). To the extent that the gifted 1% avoid business, that may restore a negative correlation / underepresentation. Finally, psychopath’s impulsivity and few long-range goals or efforts (another part of the diagnosis along with glibness/manipulation) suggests that to the extent genuine objective achievements cause you to be considered a gifted 1%, we can expect still more underrepresentation*.
So guesstimating further, I’d say in a population of 300m people (eg. the US), we could expect substantially fewer than 30,000 gifted psychopaths (0.01 0.01 300,000,000). Phew!
On the other hand, they would be the ones who would do the most damage and be least likely to ever be diagnosed, so we may never know for sure...
* Which makes me wonder about high IQ societies, now that I think about it. My vague impression was that they tend to collect those with poor social skills, and also with fewer objective accomplishments & success. So if you meet someone in a high IQ societies who seems very charming and empathetic but lacks objective accomplishments, just how much does this increase the psychopath possibility over the 1% base rate..?
** Covered multiple times in the Handbook; first relevant paper seems to be “Psychopathy and Aggression”, Porter & Woodworth.
IQ is defined to be a normal distribution with mu = 100 and sigma = 15, so “IQ 220” means ‘99.9999999999999th percentile¹’; if more than a person in 10^15 gets such a score, then the test is miscalibrated. (But most tests are, beyond a few standard deviations away from the mean.)
I didn’t count the nines, I just copied and pasted the output of
pr norm(8)in gnuplot and moved the decimal point.