This effect held in 3/3 of the national standardisation samples I checked, and using three different methods of g-loading estimation.
The g-loading–male advantage associations were all directionally consistent, and statistically significant in most cases.
Italians were an exception, as the gender ability profiles in their sample appear to be less differentiated. Strangely, Italian males even outperformed Italian females on some processing speed tasks.
Moving on, it’s immediately obvious that the cancellation subtest is a uniquely poor g proxy, and there is an intuitive way to demonstrate this: it is 2.79 standard deviations below the subtest mean correlation average.
Thus, I was curious to see how dropping it might affect things. No significant (Pearson) correlations became non-significant, and vice versa.
Inb4 muh measurement invariance: yes, Jensen’s method of correlated vectors is, to quote Emil, the poor man’s MGCFA. But I still think it interesting and worth pointing out that the ‘trivial’ sex difference in WAIS FSIQ is only trivial due to women performing better on the subtests more weakly related to general intelligence.
Sources:
Correlation matrix & reliabilities: WAIS-IV technical manual.
US sex gaps: researchgate.net/profile/Davide…
German sex gaps: sci-hub.africa/https://www.sc…
Italian sex gaps: academia.edu/121922083/Gend…
Share this Scrolly Tale with your friends.
A Scrolly Tale is a new way to read Twitter threads with a more visually immersive experience.
Discover more beautiful Scrolly Tales like this.