If you want to do one on gender, you would go in for self-reported gender, and have 3+ selection options (m/f/o).
If you're going to talk about natural selection resulting in career preference, you should talk sex, not gender.
I don't mind recognizing any individuals uniqueness or their socially less-privileged situation for social or political purposes, but it shouldn't have any affect on the science of sexuality which should just reflect nature as it is.
Sexual dimorphism in animals can be dramatic. The male angler fish is a tiny parasite, the dramatically larger female does all the angling work. And we aren't seeing non-binary angler fishes. If none of the other species are coming up with non-binary specimens, why should we assume humans are any different?
Ah, if that's what you're getting from what I'm saying, then I don't think you getting what I'm saying.
If your theory is that biological factors drive differences in behavioral outcomes, then I'm saying you should measure those biological factors, and you should do so on a more detailed level than genitals. I'm not arguing that sexual dimorphism isn't a thing. I'm arguing that a simple m/f measurement of it is insufficient, particularly if you're trying to correlate it to behavioral outcomes. As for your science paper request, I'll just go with this, instead: https://en.wikipedia.org/wiki/Sexual_dimorphism_measures
If your theory is that gender (how society treats you) affects this, then measure /that/; which itself is more complicated than a simple binary. A masculine woman and a feminine man are treated very differently by society than a feminine woman and a masculine man. Although it's a lot more reasonable to go with a binary here, since basically all of society will put people into that binary, I'd definitely want some measure as to how well or poorly that category fits the person in question. It would make for a much more compelling case to see that the outcome isn't affected by that second measure.