I think that, much like with Thinking, Fast and Slow, anyone recommending it needs to add a caveat that some of the results didn't hold up.

Ideally, there were would be revised editions with mistakes corrected.