Steve Sailer wants a measure of average college admissions scores by state, similar to what NAEP provides but with the benefit of the test takers having some extrinsic motivation to perform well. The existing duopoly in the testing market currently precludes easy apples-to-apples comparisons from being made across states. Once achieved, standardized college admissions scores may provide better IQ estimates than NAEP tests do.
Parenthetically, though, a hypothetical formula like the one Steve is after would have to be careful not to artificially boost the apparent average intelligence of states with wider IQ distributions at the expense of states with more cognitive parity--ie, California's college admissions test takers look better vis-a-vis North Dakota's college admissions test takers than California's entire 8th grade student body does vis-a-vis North Dakota's entire 8th grade student body because diversity = disparity.
My guess is that a carefully designed synthesized measure won't deviate much from what NAEP reveals. Using a slapdash index calculated by taking a state's ACT composite average, multiplying it by 71.333 (to nominally scale it to the SAT mean), multiplying that by a state's ACT participation rate divided by the sum of its ACT and SAT participation rates and then adding that to a state's SAT composite average multiplied by its SAT participation rate divided by the sum of its ACT and SAT participation rates, we get a correlation of .64 (p-value = 0.00000004) with NAEP 8th grade math and reading assessment scores. Clean the formula up and address potential confounding factors--there are several states, for example, where the ACT is compulsory for high school juniors--and the correlation is virtually certain to become even stronger.