Factor analysis is more complicated than that. If you look at that psych package details (https://www.rdocumentation.org/packages/psych/versions/2.4.12/topics/fa), you will see that both the loadings and factor scores can be calculated in many different ways. I found that this doesn't generally matter so much, but it might better for this analysis. There is a function in my package that calculates factor scores using every variation of these two methods. You could look at these. For the theory, see e.g. https://openpublishing.library.umass.edu/pare/article/id/1523/
Don't you go dying on us! Fainting is rare, probably indicates an underlying issue. Maybe time to get a complete medical checkup to see if anything stands out.
I wish I picked that as my major. Even learning on my own, it's been hugely eye opening for understanding and explaining social problems, as well understanding as solutions that can, and do, actually work.
Wouldn't we expect a left-tailed distribution? Normal is what you'd expect from many, small, random, independent contributions as from genes. But IQ is also affected by some factors of greater effect which are nearly all negative: (eg, downs syndrome and similar handicaps; nutritionial deprivation; toxin and parasite loads).
I'm not deeply trained in statistics, but I'm not seeing the mystery. Does the skew we see fail to match, in some technical way, the skew that would be expected from [normal distribution] + [some large negative effects]?
(I see that this is part 2 and you mention asymmetric contributions in part 1. Still, I suppose it seems odd to me that cognitive psychologists would expect the distribution to be normal in the first place)
It's interesting that your left-tailed hypothesis is likely true and likely extreme (in that outlier high IQ scores are much rarer than previously thought). I wonder how this affects the correlation between IQ and various outcomes since many analyses just take the normal distribution as a given.
Factor analysis is more complicated than that. If you look at that psych package details (https://www.rdocumentation.org/packages/psych/versions/2.4.12/topics/fa), you will see that both the loadings and factor scores can be calculated in many different ways. I found that this doesn't generally matter so much, but it might better for this analysis. There is a function in my package that calculates factor scores using every variation of these two methods. You could look at these. For the theory, see e.g. https://openpublishing.library.umass.edu/pare/article/id/1523/
Don't you go dying on us! Fainting is rare, probably indicates an underlying issue. Maybe time to get a complete medical checkup to see if anything stands out.
I have naturally low levels of blood pressure and am thin (which correlates with fainting). I'm not too concerned about my health at the moment.
I really am in awe of your ability to do this, Seb. You have such a way with statistics and data. What was your inspiration for specializing in such?
Hated CS, so I pivoted into something else. Tried statistics since I was good at math in HS, and it turns out it was a good fit.
I wish I picked that as my major. Even learning on my own, it's been hugely eye opening for understanding and explaining social problems, as well understanding as solutions that can, and do, actually work.
Wouldn't we expect a left-tailed distribution? Normal is what you'd expect from many, small, random, independent contributions as from genes. But IQ is also affected by some factors of greater effect which are nearly all negative: (eg, downs syndrome and similar handicaps; nutritionial deprivation; toxin and parasite loads).
I'm not deeply trained in statistics, but I'm not seeing the mystery. Does the skew we see fail to match, in some technical way, the skew that would be expected from [normal distribution] + [some large negative effects]?
(I see that this is part 2 and you mention asymmetric contributions in part 1. Still, I suppose it seems odd to me that cognitive psychologists would expect the distribution to be normal in the first place)
A rock has a little bit of intelligence...
It's interesting that your left-tailed hypothesis is likely true and likely extreme (in that outlier high IQ scores are much rarer than previously thought). I wonder how this affects the correlation between IQ and various outcomes since many analyses just take the normal distribution as a given.