8 Comments

I’m not sure if this has any proper answer, but how much are the differences between AA and Sub-Saharan African IQ due to g? I think I’ve heard before that AAs tend to do better on less g-loaded subtests, and people have brought up the issue of African Americans being too much higher in IQ than West Africans for it to be explained by European admixture alone. I assume that the sort of nifty test taking methods people have picked up, like guessing and skipping questions, are not very common in Africa.

Expand full comment

*Relatively* better but not better in absolute terms. The magnitude of the gap is indeed strongly correlated with g-loadings [1 & 2 - pp. 369–379].

[1] https://sci-hub.se/https://doi.org/10.3390/jintelligence4020006

[2] https://emilkirkegaard.dk/en/wp-content/uploads/The-g-factor-the-science-of-mental-ability-Arthur-R.-Jensen.pdf

The thing about the gap in American Blacks and African ones disqualifying a complete genetic origin is true, although Wicherts has argued that some of the gap is due to test bias which would imply that this isn't actually an issue (iirc this is the paper).

https://www.sciencedirect.com/science/article/abs/pii/S1041608009001071

There's also the possibility of Black American IQ being overestimated due to issues with representative sampling, really being something like 82. There's a meta-analysis in the works addressing this but I can't really say more about it.

Expand full comment

I have an unrelated question. Are sex differences on spatial tests on gv or g? Im assuming they are on gv. Thanks.

Expand full comment

Both. Men are smarter than women (indeed on g) by about 4 points, but in addition to the gv difference you'd expect from this, the gap on gv is even larger.

Expand full comment

Thank you. Is the g-loading of verbal tests higher for women than for men?

Expand full comment

>is the g-loading of verbal tests higher for women

metric invariance holds in the project talent data so I wouldn't think so:

https://humanvarieties.org/2023/09/01/g-explains-black-white-but-not-sex-differences-in-cognitive-abilities-in-the-project-talent/

As for whether the post above finds sex differences in g,

1. Gaps in g were shown for the bifactor model but not the hierarchical one:

--"Among the BF models, the no SH constrains the g factor difference to be zero is largely rejected, with RMSEAD=.671."

--see table 5

2. From previous research, our priors should be that bifactor models are more appropriate, even adjusting for model complexity:

https://sci-hub.ru/https://doi.org/10.3390/jintelligence5030027

https://sci-hub.ru/https://doi.org/10.1016/j.intell.2016.01.006

3. Said priors are supported in his analyses here. The model fit is better, and if there were ever a sample with the statistical power to support complex models, it's project talent.

4. The bifactor model was the one with the smallest amount of sex bias. Which, this is the obvious prediction to make given that there are well known sex differences in ability tilts and the bifactor model should be the one which yields the least amount of conflation between g and s

ALSO, this should barely matter, but do keep in mind that this is a sample of ~12th graders. While puberty isn't a concern, sex differences on various psychological variables tend to decline into old age, as dimorphism is most evolutionary necessary at the age of sexual maturity when people are actually breeding. Should mean a decline in the gap in g, but also in the gaps of non-g specificities, which will reduce bias.

Expand full comment

wow thank you for the detailed reply!

Expand full comment

On gv and/or g i meant

Expand full comment