Through foster children, and measuring whether their IQs are ultimately more correlated with their foster family or genetic family.

> Via what mechanism are we testing? Did we send some kids through a good upbringing and some to a bad one in a randomized controlled way, or did we just take a survey and assume intelligence was passed down through genealogy instead of upbringing?

The most informative studies in this area are usually of twins-separated-at-birth, but the datasets aren't very large.