People would assume that you probably have an IQ in the gifted range if you tell them that you read LW. In this case, everyone has an incentive to fudge the numbers.
Ok, now here is a motive! I still find it difficult to believe that:
Most of 1000 people care so much about status that they’re willing to prioritize it over truth, especially since this is LessWrong where we gather around the theme of rationality. If there’s anyplace you’d think it would be unlikely to find a lot of people lying about things on a survey, it’s here.
The people who take the survey know that their IQ contribution is going to be watered down by the 1000 other people taking the survey. Unless they have collaborated by PM and have made a pact to fudge their IQ test figures, these frequently math oriented people must know that fudging their IQ figure is going to have very, very little impact on the average that Yvain calculates. I do not know why they’d see the extra work as worthwhile considering the expected amount of impact. Thinking that fudging only one of the IQs is going to be worthwhile is essentially falling for a Pascal’s mugging.
Registration at LessWrong is free and it’s not exclusive. At all. How likely is it, do you think, that this group of rationality-loving people has reasoned that claiming to have joined a group that anybody can join is a good way to brag about their awesomeness?
I suppose you can argue that people who have karma on their accounts can point to that and say “I got karma in a gifted group” but lurkers don’t have that incentive. All lurkers can say is “I read LessWrong.” but that is harder to prove and even less meaningful than “I joined LessWrong”.
Putting the numbers where our mouths are:
If the average IQ for lurkers / people with low karma on LessWrong is pretty close to the average IQ for posters and/or people with karma on LessWrong, would you say that the likelihood of post-making/karma-bearing LessWrongers lying on the survey in order to increase other’s status perceptions of them is pretty low?
Do you want to get these numbers? I’ll probably get them later if you don’t, but I have a pile of LW messages and a bunch of projects going on right now so there will be a delay and a chance that I completely forget.
165 out of 549 responses without reported positive karma (30%) self-reported an IQ score; the average response was 138.44.
181 out of 518 responses with reported positive karma (34%) self-reported an IQ score; the average response was 138.25.
One of the curious features of the self-reports is how many of the IQs are divisible by 5. Among lurkers, we had 2 151s, 1 149, and 10 150s.
I think the average self-response is basically worthless, since it’s only a third of responders and they’re likely to be wildly optimistic.
So, what about the Raven’s test? In total, 188 responders with positive karma (36%), and 164 responders without positive karma (30%) took the Raven’s test, with averages of 126.9 and 124.4. Noteworthy is the new max and min- the highest scorer on the Raven’s test claimed to get 150, and the three sub-100 scores were 3, 18, and 66 (of which I suspect only the last isn’t a typo or error of some sort).
Only 121 users both self-reported IQ and took the Raven’s test. The correlation between their mean-adjusted self-reported IQ and mean-adjusted Raven’s test was an abysmal .2. Among posters with positive karma, the correlation was .45; among posters without positive karma, the correlation was -.11.
Thank you for these numbers, Vaniver! I should have thanked you sooner. I had become quite busy (partly with preparing my new endless September post) so I did not show up to thank you promptly. Sorry about that.
The people who take the survey know that their IQ contribution is going to be watered down by the 1000 other people taking the survey.
I have thought of that. But a person who wants to lie about his IQ would think this way: If I lie and other LWers do not, it is true that my impact on the average calculated IQ will be negligible, but at least it will not be negative; but if I lie and most other LWers also lie, then the collective upward bias will lead to a very positive result which would portray me in a good light when I associate myself with other LWers. So there is really no incentive to not lie.
(I’m not saying that they definitely lied; I’m merely pointing out that this is something to think about.)
How likely is it, do you think, that this group of rationality-loving people has reasoned that claiming to have joined a group that anybody can join is a good way to brag about their awesomeness?
Fair point; but very often the kind of clubs you join does indicate something about your personality and interests, regardless of whether you are actually an active/contributing member or not. Saying “I read LessWrong” or “I joined LessWrong” certainly signals to me that you are more intelligent than someone who joined, say, Justin Bieber’s fan club, or the Twilight fan-fiction club. And if there are numbers showing that LW readers tend to have IQs in the gifted range, naturally I would think that X is probably quite intelligent just by virtue of the fact that X reads LW.
One last point is that LWers might not be deliberately lying: Perhaps they were merely victim to the Dunning-Kruger effect when self-reporting IQs. I am not sure if there are any studies showing that intelligent people are generally less likely to fall prey to the Dunning-Kruger effect.
Last but not least, I would again like to suggest that future surveys include questions asking people how much time they spent on average preparing for exams such as the SAT and the ACT—as I pointed out previously, scores on such exams can be very significantly improved just by studying hard, whereas tests like iqtest.dk actually measure your native intelligence.
Not true. It would probably take at least 20 minutes to fudge all the stuff that has to be fudged. When you’re already fatigued from filling out survey questions, that’s even less desirable at that time. At best, this would be falling for a Pascal’s mugging. True that some people may. But would the majority of survey participants… at a site about rationality?
Perhaps they were merely victim to the Dunning-Kruger effect when self-reporting IQs
They were not asked to assess their own IQ they were asked to report the results of a real assessment. To report something other than the results of a real assessment is a type of lie in this case.
I would again like to suggest that future surveys include questions asking people how much time they spent on average preparing for exams
That’s a suggestion for Yvain. I don’t assist with the surveys.
Ok, now here is a motive! I still find it difficult to believe that:
Most of 1000 people care so much about status that they’re willing to prioritize it over truth, especially since this is LessWrong where we gather around the theme of rationality. If there’s anyplace you’d think it would be unlikely to find a lot of people lying about things on a survey, it’s here.
The people who take the survey know that their IQ contribution is going to be watered down by the 1000 other people taking the survey. Unless they have collaborated by PM and have made a pact to fudge their IQ test figures, these frequently math oriented people must know that fudging their IQ figure is going to have very, very little impact on the average that Yvain calculates. I do not know why they’d see the extra work as worthwhile considering the expected amount of impact. Thinking that fudging only one of the IQs is going to be worthwhile is essentially falling for a Pascal’s mugging.
Registration at LessWrong is free and it’s not exclusive. At all. How likely is it, do you think, that this group of rationality-loving people has reasoned that claiming to have joined a group that anybody can join is a good way to brag about their awesomeness?
I suppose you can argue that people who have karma on their accounts can point to that and say “I got karma in a gifted group” but lurkers don’t have that incentive. All lurkers can say is “I read LessWrong.” but that is harder to prove and even less meaningful than “I joined LessWrong”.
Putting the numbers where our mouths are:
If the average IQ for lurkers / people with low karma on LessWrong is pretty close to the average IQ for posters and/or people with karma on LessWrong, would you say that the likelihood of post-making/karma-bearing LessWrongers lying on the survey in order to increase other’s status perceptions of them is pretty low?
Do you want to get these numbers? I’ll probably get them later if you don’t, but I have a pile of LW messages and a bunch of projects going on right now so there will be a delay and a chance that I completely forget.
From the public dataset:
165 out of 549 responses without reported positive karma (30%) self-reported an IQ score; the average response was 138.44.
181 out of 518 responses with reported positive karma (34%) self-reported an IQ score; the average response was 138.25.
One of the curious features of the self-reports is how many of the IQs are divisible by 5. Among lurkers, we had 2 151s, 1 149, and 10 150s.
I think the average self-response is basically worthless, since it’s only a third of responders and they’re likely to be wildly optimistic.
So, what about the Raven’s test? In total, 188 responders with positive karma (36%), and 164 responders without positive karma (30%) took the Raven’s test, with averages of 126.9 and 124.4. Noteworthy is the new max and min- the highest scorer on the Raven’s test claimed to get 150, and the three sub-100 scores were 3, 18, and 66 (of which I suspect only the last isn’t a typo or error of some sort).
Only 121 users both self-reported IQ and took the Raven’s test. The correlation between their mean-adjusted self-reported IQ and mean-adjusted Raven’s test was an abysmal .2. Among posters with positive karma, the correlation was .45; among posters without positive karma, the correlation was -.11.
Thank you for these numbers, Vaniver! I should have thanked you sooner. I had become quite busy (partly with preparing my new endless September post) so I did not show up to thank you promptly. Sorry about that.
You’re welcome!
I have thought of that. But a person who wants to lie about his IQ would think this way: If I lie and other LWers do not, it is true that my impact on the average calculated IQ will be negligible, but at least it will not be negative; but if I lie and most other LWers also lie, then the collective upward bias will lead to a very positive result which would portray me in a good light when I associate myself with other LWers. So there is really no incentive to not lie.
(I’m not saying that they definitely lied; I’m merely pointing out that this is something to think about.)
Fair point; but very often the kind of clubs you join does indicate something about your personality and interests, regardless of whether you are actually an active/contributing member or not. Saying “I read LessWrong” or “I joined LessWrong” certainly signals to me that you are more intelligent than someone who joined, say, Justin Bieber’s fan club, or the Twilight fan-fiction club. And if there are numbers showing that LW readers tend to have IQs in the gifted range, naturally I would think that X is probably quite intelligent just by virtue of the fact that X reads LW.
One last point is that LWers might not be deliberately lying: Perhaps they were merely victim to the Dunning-Kruger effect when self-reporting IQs. I am not sure if there are any studies showing that intelligent people are generally less likely to fall prey to the Dunning-Kruger effect.
Last but not least, I would again like to suggest that future surveys include questions asking people how much time they spent on average preparing for exams such as the SAT and the ACT—as I pointed out previously, scores on such exams can be very significantly improved just by studying hard, whereas tests like iqtest.dk actually measure your native intelligence.
Not true. It would probably take at least 20 minutes to fudge all the stuff that has to be fudged. When you’re already fatigued from filling out survey questions, that’s even less desirable at that time. At best, this would be falling for a Pascal’s mugging. True that some people may. But would the majority of survey participants… at a site about rationality?
They were not asked to assess their own IQ they were asked to report the results of a real assessment. To report something other than the results of a real assessment is a type of lie in this case.
That’s a suggestion for Yvain. I don’t assist with the surveys.