SAT is an IQ test[1]. It was always an IQ test. There might be better (i.e. more g-loaded) tests out there, but the SAT still definitely works.
> You might have meant to say that SAT is correlated with IQ. While true, that is by design and is scientifically uninteresting on its own.
The SAT is correlated with g, which makes it an IQ test. You claimed that IQ doesn't correlate with anything other than performance on IQ tests. That is clearly incorrect with regard to academic performance. That you find this correlation 'uninteresting'is frankly beside the point.
> If your model includes scholastic amplitude but you measure it in IQ, you will not only have a less accurate model, but you will also make gathering data way more expensive and difficult.
There's not one 'IQ' test. Anything that correlates strongly with g is an IQ test. Yes, it also tests knowledge, but the ability to learn and recall facts is a heavily g-loaded activity.
I didn’t conclude anything. This is merely a suspicion, I explained my reasoning below.
SAT being an IQ is a stretch. SAT is a scholastic amplitude, a far more narrowly defined construct. In fact IQ (according to psychometricion) should be immutable and somewhat inheritable, while SATs should reflect the work you put in during your school career and should be fair across demographics. IQ is also normally distributed while SAT has a negative skew (raw mean around 1050 with arithmetic mean of 1000).
You might have meant to say that SAT is correlated with IQ. While true, that is by design and is scientifically uninteresting on its own. SATs are far simpler and cheaper to administer. We have way more SAT data on the general population then IQ data. If your model includes scholastic amplitude but you measure it in IQ, you will not only have a less accurate model, but you will also make gathering data way more expensive and difficult.