[fsloth]

"Grit" by Angela Duckworth contains excellent examples of the unfair outcomes false negatives cause. She's also published research dealing with this topic:

http://www.pnas.org/content/108/19/7716.abstract

People can get bad scores just because they are really nervous, etc. Now you have a score, that does not display your innate abilities at all, but, it blocks you from several activities you might excel and enjoy.

To inject some qualitative opinions of my own, totally lacking of any refrences:

IQ tests can be used to sieve through populations - with false negatives. Their only utility in career context is as an arbitrary tool to reduce candidate population. The only sane motivation for their use would be a political or economic pressure to restrict number of candidates to make their evaluation and processing cheaper downstream. The downsides are: arbitrary unfair blocking of individual careers, potentially removing candidates that would excel. If the point is to reduce the population, then generally, if the cost function of candidate quality cannot be evaluated precisely, a completely random process would likely lead to a better outcome than some arbitrary numeric metric. (I'll need to dig through my algorithm resources to formulate a precise reference if someone wants for this last statement).