> The "average" value for women is more like 2 nmol/L than 0.5.
Lie. The 25th -75th percentile range for middle distance runners at the 2011 and 2013 WCs was 0.46 to 0.92 nmol/l. Mean for women who are not athletes is 0.4 nmol/l
"Total testosterone levels greater than 20 ng per dL (0.7 nmol per L) or DHEAS levels greater than 700 ng per mL (1.9 μ mol per L) are suggestive of androgen-secreting tumors; these patients should be referred for gynecologic investigation."
In other words a woman with a testosterone level above 0.7 nmol/l is someone who should be screened for cancer.
A level of 2 nmol/l is found in women with PCOS, who are typically hirsute & obese.
As far as this study goes, it's a bit silly to claim 46 SDs, it's possible (?) in late-stage ovarian cancer that you could get such a level of testosterone, and there are various tumours that can result in >10 nmol/l, so in that sense the outliers aren't particularly meaningful in that they won't follow a normal distribution, in that there could be significant numbers of terminally ill women many S.Ds from the mean, far in excess of the number predicted by the central limit theorem.
Anyway, the point here is that well Caster's T levels are perfectly normal for someone who has testes. So trying to argue about S.Ds is a bit pointless, as you might just as well argue that a horse is 46 SDs faster than Usain Bolt - they aren't in the same category.
(Also your Russians are probably cheating....)