They are not very knowable. It is expensive to design a study that would work. All too often a real world attempt to figure this out conclude "despite our best efforts we couldn't get people to behave in the needed way". So we have proxy studies that we hope mean something, but might not. Mixed in are lots of people making data fit their conclusion, and then selling it as fact.

Few people have the time to figure things out and so it isn't knowable even though all the steps are easy to lay out.