I think it means that if you created a model that used IQ, quantitative skills, and working hours as inputs and tried to predict pay across all jobs in the US, it would predict that science jobs should pay higher than they actually do. Which leaves open the question of whether those variables are actually the most predictive of pay levels in non-science jobs; I suspect not.

It means normalizing a certain data set for the IQ that the subjects from whom these data points were taken have. In other words, adjusting one property of something be an amount relative to some other property of that something. This is done in cases where it is suspected that this one property influences the other. In this case, the assumption (quite reasonably, imo) is that (in the aggregate) people who are smarter make more money.

It means comparing people in different jobs who have the same IQ.

Haven't we reached the point yet where the whole idea of IQ testing is pretty discredited? And adjusting for IQ, meaning that someone is smart therefore they are expected to earn more money? It sounds like "taking into account these variables, and introducing this ridiculous one to allow me to adjust the outcomes as needed."

Well if you don't agree that somebody with an IQ of 120 can be reasonably expected to be better at making money, and in the aggregate people with that IQ do actually make more, than somebody with an IQ of 80, then you have such a radically different view from the mainstream that any discussion is pointless.

FYI, adjusting for certain variables is a fundamental aspect of statistics. All studies in social sciences and economics of real-world data do so, because no real-world effect can be isolated to the point that it can be measured independently.

"Haven't we reached the point yet where the whole idea of IQ testing is pretty discredited?"

What? No, of course not. Are you saying that there are no people who are smarter than others?

IQ is not synonymous with intelligence, it is married to the 20th century philosophy of psychology and history of testing methods.

For IQ to simply mean the intelligence scale normalized so that average intelligence = 100 would take a big marketing effort amongst the education and psychology communities.

Even if it were divorced from the twists and turns of its historical development, it seems pretty clear that at it's best IQ can aggregate the values of creativity, lateral thinking, calculation, memory-retrieval, memory-storage, memory-organization, (even, despite tester's best efforts) domain knowledge together and replace them with one number.

I think that some of the "everybody learns differently!" stuff has jumped the fence and become an old wives' tale, but there has to be a happy medium between assigning someone a 40 column printout to summarize their intelligence and slapping one number on it.

[This is not to mention all of the shift in emphasis away from intelligence towards results and output based partially on Outliers, and partially on the idea that if you praise kids for an inherent trait that they have no control over that they will stop playing to win and start playing not-to-lose.]

I'm not sure if you're agreeing or disagreeing with me? Yes of course there are various definitions and measurement methods of 'IQ' and 'intelligence' and one can define all of them in various ways. Exact definitions aren't interesting for the current purposes. What I said was, some people are smarter than others, even when considering orthogonal traits. If we hypothetically consider 'intelligence' as a combination of trait A, B and C, and we choose A, B and C carefully enough so that we can score or normalize each of them to a scale of 0-10, then Alice with a score of 8 on each of them is more intelligent than Bob who scores 4 on each of them. Now in the margin you can argue who is more intelligent when the scores are 5-8-4 and 5-4-8 but that doesn't take away from the point.

It's not like we're talking about one specific methodology for measuring IQ. The whole argument is in the context of the OP arguing that correcting for intelligence is necessary for making a meaningful comparison between wages earned (basically, it's discounting for opportunity cost). Which is totally reasonable and obvious.

I think we were agreeing

I don't think I made the argument that being functionally disabled is meaningless, but straw men live in all discussions at some point.

My argument is that above average and even extraordinary intelligence have no correlation to money making. Smart people live in poverty all the time. To say that expectations of earned income should be adjusted for IQ is meaningless in that context because the adjustment would be zero. Making money doesn't derive from general intelligence, but from how it is applied and luck. There are many smart technical people on this message board who are clearly lucky that the world is in the middle of a massive expansion of technically complicated economic areas like apps and programming in general. It allows them to achieve wealth that otherwise is not a predetermined given.

When I refer to IQ testing I refer to the concept of a universalized IQ test that can, without cultural bias, give an objective measure of intelligence.

"My argument is that above average and even extraordinary intelligence have no correlation to money making."

Well you're objectively wrong. Go to your national data office and look at income vs education (using education as a proxy for intelligence is not perfect but works OK enough for this purpose). You will see strong correlations between the two. (not perfectly linear, and not perfectly correlated, but strong enough to be not random). Look at any data set of reasonably stable and free countries that have these two data points, and you will find the same.

"Smart people live in poverty all the time."

Anecdote != data.

> Haven't we reached the point yet where the whole idea of IQ testing is pretty discredited?

No, the opposite! Throw together any list of questions requiring intellectual ability, on any subjects you choose (making sure to have a wide range of difficulty). Use the list to test a few thousand people selected at random from the same society. Perform principal components analysis (a type of statistical cross correlation) on the answers. The answers will turn out to correlate with a single characteristic of the individual test takers. This common factor is labeled g, general intelligence. There is virtually no sign of multiple intelligences or other factors, just one honking big signal for the g factor.

> And adjusting for IQ, meaning that someone is smart therefore they are expected to earn more money?

There have been large studies of people from the general population, with the scientists measuring every data point they can lay their hands on. It turns out the only factor that significantly affects adult income is IQ. Earned income is almost totally uncorrelated with race, skin color, culture, family wealth, family social rank, location of residence, school system, characteristics of siblings, and so forth. IQ also strongly predicts criminality.

Studies of twins separated at birth show huge IQ correlations between identical twins, but IQ correlations between fraternal twins are no greater than for sibling pairs from different pregnancies. So IQ is mostly inherited, and mostly fixed by the time of conception. In other words, genetic.