The purpose of the market is sort of to elicit opinions and evidence about how to think of the relationship between intelligence and IQ tests. People often seem to make confident statements about the relationship between intelligence and IQ tests, but they rarely seem to back that up very well, so I would be interested to see whether these is something behind it.
The word “intelligence” doesn’t have a universally agreed precise meaning. If one person says “intelligence means being good at solving intricate logical puzzles” and another person says “no, intelligence means good at the cognitive aspects of navigating the practical challenges the world throws at us”, I don’t see any basis for saying one of them is right and the other is wrong.
I think there are ways to at least come with some reasons to favor one approach or another.
For instance, if these two things are completely correlated, then measuring one will also mean measuring the other. This makes the distinction moot for the question of whether IQ tests measure intelligence.
Alternatively, if one can show that one of the meanings is rarely plausibly used, then that is a reason to discount that meaning. Which brings me to your next point...
You could try to undertake some sort of survey of how the word “intelligence” is used, and then declare the meaning to be some sort of average. But what that survey would actually tell you is that different people use the word in very different ways and most of them don’t mean any very specific thing by it.
I think even this sort of survey would be helpful in practice, even if it shows that people mean inconsistent things by it, because it could help map out the domain of applicability of IQ tests.
That is, if there are 5 things people might mean by intelligence, and one of them is g, one of them correlates moderately with g, and the remaining three are uncorrelated with g, then it seems like it would be good to know that IQ tests are a good measure of the first one, a weak proxy of the second one, and unrelated to the other three.
That way, when people have conversations about intelligence, one can notice what notion they are talking about, and adjust accordingly as to whether IQ tests are relevant or not.
However, in the abstract, discussing this possibility is not directly useful, because we would need to know concretely what people are thinking of in order to apply this method.
You could declare that “intelligence means g”, i.e., a “general factor” found when doing something PCA-like to lots of different tests of cognitive capacity. But exactly what you get by doing that depends on what tests you use and exactly what PCA-like thing you do. (I think. I am not an expert on this stuff. I tried some calculations with a toy model which did show some dependence on what tests you use and how you compute the factors, but I can’t guarantee I didn’t do something boneheaded.)
Actually, one of the things that makes general factors interesting is that they can be extremely consistent.
If you average lots of different ability tests, the number you get will be a measure of the things that affect many different abilities, as the things that only affect few abilities average out. And since these things affect many different abilities, they will probably also affect one’s performance on different ability tests.
The purpose of the market is sort of to elicit opinions and evidence about how to think of the relationship between intelligence and IQ tests. People often seem to make confident statements about the relationship between intelligence and IQ tests, but they rarely seem to back that up very well, so I would be interested to see whether these is something behind it.
I think there are ways to at least come with some reasons to favor one approach or another.
For instance, if these two things are completely correlated, then measuring one will also mean measuring the other. This makes the distinction moot for the question of whether IQ tests measure intelligence.
Alternatively, if one can show that one of the meanings is rarely plausibly used, then that is a reason to discount that meaning. Which brings me to your next point...
I think even this sort of survey would be helpful in practice, even if it shows that people mean inconsistent things by it, because it could help map out the domain of applicability of IQ tests.
That is, if there are 5 things people might mean by intelligence, and one of them is g, one of them correlates moderately with g, and the remaining three are uncorrelated with g, then it seems like it would be good to know that IQ tests are a good measure of the first one, a weak proxy of the second one, and unrelated to the other three.
That way, when people have conversations about intelligence, one can notice what notion they are talking about, and adjust accordingly as to whether IQ tests are relevant or not.
However, in the abstract, discussing this possibility is not directly useful, because we would need to know concretely what people are thinking of in order to apply this method.
Actually, one of the things that makes general factors interesting is that they can be extremely consistent.
If you average lots of different ability tests, the number you get will be a measure of the things that affect many different abilities, as the things that only affect few abilities average out. And since these things affect many different abilities, they will probably also affect one’s performance on different ability tests.
So the general factors for different tests can be extremely correlated. To give an example, here’s a study which found a correlation of 0.79 between the general factor underlying video games and the general factor underlying some classical intelligence test tasks.