The mean IQ reported in the latest LW survey will be 137 or less (probability 75%).
Thought I’d squeeze that in before Yvain posts the results. Some background. In the 2012 survey LWers reported having a mean IQ of 138.7. I found (and find) that unlikely, and reckon that average was inflated through a mixture of selective reporting, mis-remembering, perhaps the occasional lie, and people taking old or otherwise iffy tests. However, the 2013 survey should suffer much less from that last problem because (if I remember rightly) it asks only about formally assessed IQs. So I expect the 2013 IQ average to be less (because less inflated) than the 2012 IQ average.
I’m not super confident about this, because there’s some countervailing evidence (using self-reported SAT & ACT scores from the 2012 survey as IQ proxies suggests higher IQ estimates, not lower estimates), and because a selection effect could actually worsen the inflation in the 2013 survey (people with higher IQs might be more likely to have their IQs formally tested; if so, asking only for properly measured IQs will selectively pick out high-IQ people). But my hunch is that these issues won’t matter so much.
(As well as the headline result that the average turned out to be 138.2, see also section V.B, and Vaniver’s comment. Also, it looks like I may have been wrong to take Yvain’s SAT-and-ACT-to-IQ conversions from last year at face value in the parent comment.)
No idea (Yvain hasn’t said yet), but presumably some time this year!
But I can try to guess. In 2012, Yvain apparently closed the survey on November 26 and reported the results on December 7. In 2011, the survey closed on December 3 and Yvain reported results on December 5. And in 2009, Yvain announced the survey on May 3 and posted its results on May 12. So for those surveys, it’s been 11 days, 2 days, and ≈9 days between when the survey closed and when Yvain posted results, and I’d guess that 2-11 days after Yvain closes last year’s survey (which he hasn’t yet), he’ll post its results.
(Maybe I should make that a formal prediction? “2-11 days after Yvain closes the current LW survey, he’ll post the results (80%).”)
A 138 average doesn’t seem far-fetched at all to me. A little bit of self-serving bias is inevitable, but I highly doubt the real average is e.g. in the 120s. This random website I found says that the average IQ of an Ivy League student is 142. I go to a school that isn’t as good as most Ivys but is better than some of them. I would guess the average IQ of a student here is 135-ish. The average LW poster seems much, much smarter to me than the average person at my school.
Yeah, I’ve moved a bit towards your sort of position because of the 2013⁄14 results. That said, I don’t have an impression of LWers being way, way smarter than other students I encounter in real life.
(I’m also still leery of the poor correlation between education level and IQ, which cropped up again in the latest survey. To go into tedious detail: among those aged ≥29, 25 people with a high school education or less gave a mean IQ of 139.5, and 155 people with more education gave a mean IQ of 140.2. And Nornagest’s suggestion to look at the high end now gives a less statistically significant result than last year. The 24 oldsters with PhDs gave a mean IQ of 142.4, and the other 156 non-PhDs gave a mean IQ of 139.8.)
The mean IQ reported in the latest LW survey will be 137 or less (probability 75%).
Thought I’d squeeze that in before Yvain posts the results. Some background. In the 2012 survey LWers reported having a mean IQ of 138.7. I found (and find) that unlikely, and reckon that average was inflated through a mixture of selective reporting, mis-remembering, perhaps the occasional lie, and people taking old or otherwise iffy tests. However, the 2013 survey should suffer much less from that last problem because (if I remember rightly) it asks only about formally assessed IQs. So I expect the 2013 IQ average to be less (because less inflated) than the 2012 IQ average.
I’m not super confident about this, because there’s some countervailing evidence (using self-reported SAT & ACT scores from the 2012 survey as IQ proxies suggests higher IQ estimates, not lower estimates), and because a selection effect could actually worsen the inflation in the 2013 survey (people with higher IQs might be more likely to have their IQs formally tested; if so, asking only for properly measured IQs will selectively pick out high-IQ people). But my hunch is that these issues won’t matter so much.
Prediction failed.
(As well as the headline result that the average turned out to be 138.2, see also section V.B, and Vaniver’s comment. Also, it looks like I may have been wrong to take Yvain’s SAT-and-ACT-to-IQ conversions from last year at face value in the parent comment.)
When will we get to see the results of the LW survey?
No idea (Yvain hasn’t said yet), but presumably some time this year!
But I can try to guess. In 2012, Yvain apparently closed the survey on November 26 and reported the results on December 7. In 2011, the survey closed on December 3 and Yvain reported results on December 5. And in 2009, Yvain announced the survey on May 3 and posted its results on May 12. So for those surveys, it’s been 11 days, 2 days, and ≈9 days between when the survey closed and when Yvain posted results, and I’d guess that 2-11 days after Yvain closes last year’s survey (which he hasn’t yet), he’ll post its results.
(Maybe I should make that a formal prediction? “2-11 days after Yvain closes the current LW survey, he’ll post the results (80%).”)
A 138 average doesn’t seem far-fetched at all to me. A little bit of self-serving bias is inevitable, but I highly doubt the real average is e.g. in the 120s. This random website I found says that the average IQ of an Ivy League student is 142. I go to a school that isn’t as good as most Ivys but is better than some of them. I would guess the average IQ of a student here is 135-ish. The average LW poster seems much, much smarter to me than the average person at my school.
Yeah, I’ve moved a bit towards your sort of position because of the 2013⁄14 results. That said, I don’t have an impression of LWers being way, way smarter than other students I encounter in real life.
(I’m also still leery of the poor correlation between education level and IQ, which cropped up again in the latest survey. To go into tedious detail: among those aged ≥29, 25 people with a high school education or less gave a mean IQ of 139.5, and 155 people with more education gave a mean IQ of 140.2. And Nornagest’s suggestion to look at the high end now gives a less statistically significant result than last year. The 24 oldsters with PhDs gave a mean IQ of 142.4, and the other 156 non-PhDs gave a mean IQ of 139.8.)