When the Damore memo came out, people have calculated that to explain the only 20% women working for google, the average IQ of google engineers needs to be 160. That is laughable.
Interesting. Do you have a link to that calculation? Maybe tptacek was impliticly referencing that calculation. I took what he said at face value but maybe real world numbers make my claim moot.
By way of calculation, apparently 30k software engineers work for Google[1] so there are probably around 15k in the USA. There are 600k software engineers in the USA[2]. I estimate that Google employs about 10% of the engineers that could pass an interview with them if they tried, so the Google-level cohort in the USA is about 150k or 25% of all engineers.
The top 25% of engineers is 0.7 standard deviations above the mean of a normal distribution. Assuming equal means, female engineers would have to have a variance in "software engineering quotient"[3] 45% that of the male engineer variance.
So it seems that if men and women have the same mean propensity to be good software engineers (not just IQ, or even IQ at all, but general propensity) but the variance of the female propensity is 45% of that of the male, then that would entirely explain a 20%/80% ratio at Google.
(Caveat: I've calculated the SWEQ level based on male engineers only but since there is such a small proportion of female engineers I don't think this skews the calculation very much.)
Thanks! A variance ratio of 1.15 is nowhere near the 2.2 that my calculation predicts would be necessary! I think yorwba and I have done the same calculation from opposite directions and reached the same conclusion. A huge difference in variance would be needed to explain an 80/20 split at Google level.
