Kevin McGrew posted a link to a recent paper by Byington and Felps (2010) proposing an alternative explanation for the relationship between IQ and job performance. Actually it is a supplementary explanation. The authors grant that there is some link between IQ and job performance that is likely to be there whether institutions select on it or not. However, this relationship is intensified by institutional practices that allocate educational and other resources by IQ and IQ-like tests. That is, if you give job skill-enhancing experiences to people who score high on tests, the correlation between job performance and IQ will increase.
I cannot imagine that this self-fulfilling correlation effect is not present. In fact, I can imagine members of committees that hand out merit scholarships thinking, “Of course, this is exactly what we are trying to promote! We have been doing this explicitly for a long time. Did we really need a new paper to ‘reveal’ this to us?”