> I feel like these "sins" are major shots in the dark with no real studies to back up such big claims.

As mentioned in the post, the claims are indeed based on real studies: observing users as they perform prescribed tasks in a usability testing lab.