I took his suggestions and recalculated the variances. (See chart below. Superfan recalculations in bold. Variance of more than 2% highlighted with yellow background.) When I did, some of the variances fell more in line with the 10% theory, but most actually went further away and not in favor of women. The average discrepancy is 12.12% (12.54% for Superfan recalculations), but if you take out the field events and calculate the average only for the running events, the average discrepancy is 10.61% (or 11.37% for the Superfan recalculations).
My gut feeling is that Superfan's suggestions are a more accurate indication of the "true" records than are the IAAF's current official records. One look at Jarmila Kratochvílová's image (click image to enlarge) should leave no doubt that there is every reason to question the validity of her 1983 world record time of 1:53.28 in the 800m. I mean, female 800m runners just don't have that kind of physique, (sorry, I take that back. Some of them do). Unfortunately, there is no way to prove that all of those records from the 1980s were enhanced by steriods.
No matter which records we use, I think it's safe to conclude that the 10% rule (give or take a couple of percentage points) can be applied only to running events. The field events involve more strength, which (there surely cannot be any dispute) is an area in which men have a marked advantage over women.