I suspect there is a self-selecting / pre-selected thing going on here, where no one asks the same question to adults who are straight / gender conforming. So the data is biased before the research is actually done.
The research I’ve seen referenced to back up this claim is based on follow ups of kids who were referred to gender clinics (back in the days when that group was mostly boys.) They mostly grew up gay.
I think the key might be that the kids in those studies actually had to have expressed unhappiness with their sex - that’s presumably more likely for a future gay kid in a homophobic society than for a boy who just happens to have some stereotypically “feminine” traits or preferences.