One thing you can do is look at people of the same sex but with different levels of testosterone. That would effectively control for environment, unless you th

One thing you can do is look at people of the same sex but with different levels of testosterone. That would effectively control for environment, unless you think that higher testosterone has subtly visible effects that cause others to (unconsciously?) treat them differently—which actually seems likely, if e.g. a high-T girl would tend to do more "rough" activities and consequently wear "tough", non-delicate clothing—but that in itself would be a useful takeaway. Another thing you can do is compare countries, or families, or perhaps schools, that are "more gender-role-restrictive" or "less gender-role-restrictive". Surely if the environment has an impact, then varying the strength of that environment would vary the impact. (Obviously there would be other variables to control for.)