There is the rub.
An IQ test is advertised as a test of one’s cognition but a pencil and paper written test will never be able to deliver an accurate measure of a man’s true cognition. We didn’t spend most of evolution with those tools. An accurate estimation of a human’s brain power can scarcely be measured by their use of such recent innovations.
It can only deliver an accurate measure of a person’s ability to take a similar test using similar tools in a similar way.
A written IQ test is just a puzzle testing a primate’s ability to decode a lot of little shapes on a piece of paper that we call letters and numbers according to a standard that the testers rely upon existing. It does not test a man’s ability to read another man’s expressions, inflection, the melodic numerical flow of sound, visual acuity, potential for danger, deception, predation, modeling the behavior of others or the ability recreate or recall a thing to our other senses.
A test of a man’s cognition that doesn’t engage the way we evolved to use our senses in the physical world around us, those things for which our brains were evolved to do, is only really just a limited impression made with a skewed aperture into a person’s true intellectual ability.
Furthermore, one slight shortcoming, differences in the common tongue’s interpretation, affects the outcome and can produce a subsequent river of falsely derrived errors. Like say, a southern kid who knows the local rural parlance but is less accustomed to the cadence of inquiry and colloquialism of the East Coast ivy league. Slight differences in trained interpretation can produce variances; the interpretations of those tests must be looked at closely as well to eliminate confirmation bias in test givers and test designers.
There’s so many things that can go wrong.
The typical metric is memory, reasoning and verbal interpretation. Those last two especially seem misnamed and fundamentally flawed… as well, people remember things they are most often accustomed to more easily so it becomes important that tests of memory involve something everyone is highly familiar with or as novel as possible to insure truly accurate results.
For example, I can remember all the words to hundreds of songs, including rythm changes, number and order of notation, pitch, tone, etc. If you were to give me a problem that involved memorizing such things without the context of familiar language of music then I have no doubt I could never replicate such a feat of memory as what my brain does automatically when passively listening to good music.