[yetanotherphd]

No matter what the control for, these studies are very limited.

If they say that intelligence is correlated with X, if we like X we will say that this proves X is a good thing, and if we don't like X we will say that this proves intelligent people are not so "intelligent" after all. Put another way, it's so hard to do proper Bayesian updating on such loaded statements, they probably make people worse informed on average.

I'll end with a quote from Mohammad Ali being interviewed by Michael Parkinson:

Ali: [Animals don't race-mix so human's shouldn't either]

Parkinson: But we have intelligence

Ali: They don't have intelligence, but yet they stay together [with their own species], we should have more intelligence than them, right?