And incorrect assumptions. As I understand it, "I did a study on this and it turns out there's no connection" generally results in the study not being published (if the study was testing for the validity of the connection)... which is sad, because that's still useful information to have.

Exactly. There should be much greater incentives to (in)validate prior publications. That is what science is about.