Also, just because one can game the test, or make its result less accurate, it doesn't mean the population level data is much less accurate. I'm reminded of the Body Mass Index - men, by strength training, can put on lots of muscle mass and become classified as "overweight" or even (rarely) "obese" without being fat. But the fact is that most "overweight" men are in fact fat, and not muscular. The fact that a test for an individual can have reduced or zero validity, does not necessarily threaten the validity for the overall population.
If everyone were to start gaming the tests, things might be different. Then again, being able to game the test might be a highly IQ dependent task, so the IQ test remains an IQ test.
it's spectacularly unstable for comparisons between cohorts though, most notably in the case of the Flynn effect. I find it highly unconvincing the average person in the 1930s had the same fundamental level of intelligence as those people scoring 80 on the same tests today.
Also, just because one can game the test, or make its result less accurate, it doesn't mean the population level data is much less accurate. I'm reminded of the Body Mass Index - men, by strength training, can put on lots of muscle mass and become classified as "overweight" or even (rarely) "obese" without being fat. But the fact is that most "overweight" men are in fact fat, and not muscular. The fact that a test for an individual can have reduced or zero validity, does not necessarily threaten the validity for the overall population.
If everyone were to start gaming the tests, things might be different. Then again, being able to game the test might be a highly IQ dependent task, so the IQ test remains an IQ test.