It's probably an overestimate too, because if you look at the dataset a lot of the 44 women are fake (i.e. have improbable ages such as 407,407,99,95 and are pr

It's probably an overestimate too, because if you look at the dataset a lot of the 44 women are fake (i.e. have improbable ages such as 407,407,99,95 and are pre-high schoolers). People filling out the survey for the lulz would choose as different a profile from the typical HN-er as they could, and that includes making a female respondent.