If you ask an LLM what the personality profile of rationalists is, it'll probably point you to the 2012 LW Survey results, which reports the following Big Five scores:
Trait
Median percentile
Corresponding z-score
Openness
65th
+0.39
Conscientiousness
30th
−0.52
Extraversion
22nd
−0.77
Agreeableness
38th
−0.31
Neuroticism
27th
−0.61
So, it looks like rats are more Open than average, and less Conscientious, Extraverted, Agreeable, and Neurotic than average. Even so, the differences aren't that large.
Except, I don't buy that. Rats are weird. So, for a trait like Openness, I'd expect rats to be much higher than just 0.39 SD above the population mean. There must be something going on.
The problem is the online personality test's norms. The 2012 survey instructed respondents to take the OutOfService BFI-44 test[1]and enter the percentiles it gave them. If the test's norms come from people who take personality tests online rather than the general population, those percentiles could make rats look more ordinary than they are.
To check, we'd want to compare LessWrong's raw scores with those of a more representative sample. Unfortunately, the survey only collected percentiles. Fortunately, VincentYu found the means and SDs the test used, which allows me to recover most of the trait scores in the public survey data.[2]
For my representative sample, I'll use Bogg & Vo (2014). They administered the BFI-44 to a weighted, probability-based U.S. sample (N = 1,015).
So, how do LessWrong's raw scores compare with the online test's assumed means and the U.S. sample's means? Here they are:
As you can see, the online test assumed higher average Openness and lower average Conscientiousness and Agreeableness than the U.S. sample, making LessWrong's differences on those traits look smaller. And while the online test made rats look less Neurotic than average, their mean is almost exactly the U.S. mean.
So, using the U.S. means and SDs, the median LessWrong score on each trait comes out to roughly:
88th percentile (+1.2 SD) on Openness
10th percentile (−1.3 SD) on Conscientiousness
27th percentile (−0.6 SD) on Extraversion
28th percentile (−0.6 SD) on Agreeableness
50th percentile (0 SD) on Neuroticism
These revised scores are much closer to what I'd expect. LessWrong covers ideas well outside the norm, and the Openness score now reflects that. Also, the revised Conscientiousness score explains all the posts about akraisa.
Yes, that's the name of the website. Interestingly enough, it's still in service years later. ↩︎
VincentYu reported the test's assumed means (SDs) on the 1–5 scale: O 3.85 (0.65), C 3.40 (0.76), E 3.30 (0.88), A 3.66 (0.70), and N 3.15 (0.85). I inverted the Gaussian percentile mapping over the BFI's possible scores, allowing for rounding in the published parameters. Of 2,112 nonmissing trait responses, 1,982 (94%) had a unique match; 95 were ambiguous and 35 unmatched. The charts omit the latter two groups. ↩︎
If you ask an LLM what the personality profile of rationalists is, it'll probably point you to the 2012 LW Survey results, which reports the following Big Five scores:
Trait
Median percentile
Corresponding z-score
Openness
65th
+0.39
Conscientiousness
30th
−0.52
Extraversion
22nd
−0.77
Agreeableness
38th
−0.31
Neuroticism
27th
−0.61
So, it looks like rats are more Open than average, and less Conscientious, Extraverted, Agreeable, and Neurotic than average. Even so, the differences aren't that large.
Except, I don't buy that. Rats are weird. So, for a trait like Openness, I'd expect rats to be much higher than just 0.39 SD above the population mean. There must be something going on.
The problem is the online personality test's norms. The 2012 survey instructed respondents to take the OutOfService BFI-44 test[1]and enter the percentiles it gave them. If the test's norms come from people who take personality tests online rather than the general population, those percentiles could make rats look more ordinary than they are.
To check, we'd want to compare LessWrong's raw scores with those of a more representative sample. Unfortunately, the survey only collected percentiles. Fortunately, VincentYu found the means and SDs the test used, which allows me to recover most of the trait scores in the public survey data.[2]
For my representative sample, I'll use Bogg & Vo (2014). They administered the BFI-44 to a weighted, probability-based U.S. sample (N = 1,015).
So, how do LessWrong's raw scores compare with the online test's assumed means and the U.S. sample's means? Here they are:
As you can see, the online test assumed higher average Openness and lower average Conscientiousness and Agreeableness than the U.S. sample, making LessWrong's differences on those traits look smaller. And while the online test made rats look less Neurotic than average, their mean is almost exactly the U.S. mean.
So, using the U.S. means and SDs, the median LessWrong score on each trait comes out to roughly:
These revised scores are much closer to what I'd expect. LessWrong covers ideas well outside the norm, and the Openness score now reflects that. Also, the revised Conscientiousness score explains all the posts about akraisa.
Yes, that's the name of the website. Interestingly enough, it's still in service years later. ↩︎
VincentYu reported the test's assumed means (SDs) on the 1–5 scale: O 3.85 (0.65), C 3.40 (0.76), E 3.30 (0.88), A 3.66 (0.70), and N 3.15 (0.85). I inverted the Gaussian percentile mapping over the BFI's possible scores, allowing for rounding in the published parameters. Of 2,112 nonmissing trait responses, 1,982 (94%) had a unique match; 95 were ambiguous and 35 unmatched. The charts omit the latter two groups. ↩︎