They also use personality tests to screen out job applicants, and that is no argument for their usefulness. I mean you could hire only libras and claim that this policy has given you better candidates, and you would not be wrong. But at the same time you couldn’t claim that libras are better at the job then leos. You have other hiring practices and biases that makes sure you are usually happy with the libras you’ve hired.

An IQ of 100 is defined to be the mean intelligence. It is a magic number (albeit not a magic number representing competence). Usually an IQ of 80-90 (around one standard deviation below the average) give you general competence, lower then that and you probably have some mental disabilities (unless the test is badly standardized; which has happened more often then it should have).

IQs above average correlate with loads of desirable things, individuals with higher then average IQ are generally more well spoken, they have higher income, they are quicker to respond to questions, they know more jokes, they are probably more handsome as well (I know; citation needed). These traits are as well desired (if not more) in the criminal world as anywhere else. But this is precisely the problem with considering high IQs. If your measurement is predicting everything, it usually means you are measuring nothing.

The argument for their usefulness is simply the actual experiments preformed. Due to it’s size the military can easily collect a lot of useful information.

> If your measurement is predicting everything, it usually means you are measuring nothing.

The goal of IQ was to predict everything as it's a one dimensional scale. You can always find a better test for any given quality, and frankly should. However, that does not somehow invalidate it's wide ranging utility.

By “wide ranging utility” you mean “detecting and diagnosing mental disabilities” right?

If not: Using IQ for anything else is, at best, quite useless, and at worst, severely problematic as you found out when you heard about scientific racism (such as the eugenics movement).

By useful, I mean saving the US military hundreds of millions of dollars. It's not perfect, and I would argue it's not even very good, but it does work. Which among several reasons is why it's been outlawed for US companies.

you keep claiming that there is some threshold above which it does not predict anything even though this contradicts all available evidence. like taleb your beliefs are based on nothing more than unfounded speculation.

No I don’t claim that (even though Taleb might have; didn’t read him thoroughly enough). I say that there is a (muddy) threshold where these tests become useless. They still hold predictive power, but what they do predict is useless.

The ASVAB can be studied for and grinded, though, like the SAT, it would require more effort with a low IQ. Iirc the breadth is less than the SAT. Is there another test you are referring to in terms of this "IQ test proxy"?

For fair disclosure, there was a study which showed that as IQ decreased, the span of time for a perfect IQ score prep increased almost infinitely, as the low IQ kids forgot the material they already learned. Maybe I will look for it later, would be very relevant here.

You can actually prepare for specific IQ tests and thus inflate your score.

Though presumably that's just gaming a test and has zero impact on IQ as measured by unrelated tests. However as this makes the tests less accurate it's discouraged.

Right. That's a huge problem even when trying to figure out if your IQ changed over time. Even with Raven's progressive matrices, just knowing roughly what type of manipulations can be done to the shapes can help a lot with your score.