Given the very low incidence of transsexualism in the general population and the fact that we're talking about mental work, using people's self-reported gender will either not hurt, or might even help.
Additionally, since gender refers to behaviour and societal roles, I don't think "gender at birth" makes sense, given that all babies act pretty much the same for the first few months. You would be better to phrase your question as "gene-determined sex", which is for now unchangeable.
There is a low incidence of transsexualism in the general population, but we are talking about the tech scene. And particularly among programmers, it seems to me that a lot of female programmers have a Y chromosome. Considering that there aren't that many female programmers to begin with, I wouldn't be too sure that effect is negligible.
Well, it's just my personal impression. I'm not saying it's a majority, but there are a lot of trans people in the tech community. So if you want to make statistics, I think you should take that into account, and not pretend that they don't matter.
