[caitp]

IQ is not "cognitive ability", it's a fallible (and arguably biased) quantification/quotient of a (not exhaustive) set of measures of (some form of) cognitive ability.

People generally don't have their IQs "measured", and even when they are, the things they are kind of good at predicting are only a small subset of important qualities, so it would hardly be a good measure by which to compare applicants.

I think the problem you have with this is that you're assuming "high IQ" and "smart" are automatically the same thing. A high IQ is only one slice of the "smart" pie.

[fiatmoney]

IQ is defined as a point estimate of the value of a general factor of intelligence, which is in turn defined as the statistical construct that most strongly correlates with performance on disparate cognitively-loaded tasks. To the extent that there can be such a thing as "general cognitive ability", which they are apparently looking for, the only question is whether whatever specific test they're using correlates strongly with that g-factor. To the extent that it does, it is axiomatically an IQ test.

[lukasb]

While you're correct, the fact that someone can define an IQ test as the test that most strongly correlates with the g-factor (itself merely a composite factor from a number of different measures) doesn't mean that IQ, or the g-factor, or "general cognitive ability", corresponds to any physical causative factor: http://vserver1.cscs.lsa.umich.edu/~crshalizi/weblog/523.htm...