This survey is (basically) a subset of the census, and there are ~100 female responders to that one. So, unless there is a big surge of female responses to this survey, the results you described here are likely to be more representative/reliable (since absolute numbers, not relative numbers, are important).

However, I think this post is still useful for the discussion it has generated.

Although, that said, the breakdown of the referred-by-friend category (into platonic vs. romantic) could be interesting. I predict that more women than men will be referred by a date, given at least 85% of the LW population is attracted to women, but at most 20% are attracted to men. I would lean towards more romantic then platonic referrals for both genders, but most people have more platonic friends than romantic ones (only 13% of LW is definitely polyamorous), so I'm not sure.