>As for the second question, the effect holds even when controlling for all sorts of things, like adoptions, location, socioeconomic status, etc.

Please post a source for your claim.