I dislike this emphasis on gender identity. I haven't seen enough non-anecdotal evidence of this to be >0.8 confident, but my model predicts that this strategy wouldn't achieve all that much, and has much more risk of being damaging (due to biases and two-steps-removed complications) than a strategy of behaving as non-sexist as possible (and 'teaching' this to the child, but that is most effective by example during childhood AFAIK).