I recall from my computer science studies that research had pretty much failed to find that different human-computer interfaces were more efficient for one sex or the other despite much belief that they would be. This was before the web (yes, many many years ago) - but it was still about getting tasks done on a screen, not about punch cards and difference engines 
.
One thing that did come out though was that there were significant differences in preference. That is, while men and women did not show different scores for achieving a goal under one interface rather than another, they did indicate they *liked* different interfaces at different rates. These preferences appeared to be cultural rather than innate, (so for instance, British women might prefer a and British men b, but American women would prefer c and American men d) but highlight the importance of really knowing your audience.
Not sure where those studies came from - I no longer have those lecture notes.
Helen
-----original message-----
>> I would discount any studies that haven't been done recently. As Gavin points out, there is a lot of learned behavior involved in using the web, and older studies could easily be skewed so that different experience levels are interpreted as gender differences.>>
.
wfnet.