I wasn’t planning on doing a third norming of the KAMIKAZE but after a commenter enquired about it, I discovered there was a lot more data to work with.
I now have 20 cases of KAMIKAZE takers reporting Wechsler IQs and the correlation between the two tests is 0.54 which is surprisingly low given how g loaded the WAIS-IV arithmetic subtest is.
It’s likely not the case that the correlation is depressed by range restriction, as the KAMIKAZE takers (mean 131; SD 16.5) were slightly more variable in their Wechsler IQs than the U.S. population (mean 100; SD 15) though this could be because these are self-reported Wechsler scores taken at different years which introduces new sources of variance.
One possibility is that Arithmetic is highly g loaded but math isn’t and the reason for that could be that math is much more dependent on education and practice, while Arithmetic is something few people practice, especially since the advent of calculators.
Another possibility is that the correlation is dragged down by the unreliability of self-reported Wechsler scores, my not correcting KAMIKAZE scores for age, and the fact that Wechsler and KAMIKAZE were not taken at the same time.
Nonetheless, when KAMIKAZE and Wechsler scores are both listed from highest to lowest, we get the following equivalencies: