[eitland]

> Your suggestion relies on the assumption that there is some way to change standardized tests such that they do not correlate with wealth.

Good IQ tests exists that are hard to game. They are easy to learn easy to vary and easy to measure.

(I think I have heard someone here talking about IQ tests that include language and geography questions. This is not the kind of IQ tests I'm talking about. Perfect IQ tests should measure your processing speed, not your "software" - i.e. education.)

[learc83]

The SAT is supposed to be a test measuring scholastic aptitude not a pure IQ test.

IQ tests always have a knowledge component anyway. Unless you specifically design a test to only measure something like working memory. And who cares about that. Colleges would only care to the extent that it correlated with performance.

But you can have 2 people who have equal performance in a given field even though one has a higher working memory because the other has better context from more time spent reading.

[eitland]

> Unless you specifically design a test to only measure something like working memory.

The ones I refer to typically test what I'd call pattern matching / pattern synthesis by presenting multiple choice questions showing a number of patterns and asking which out of several patterns comes next.

You can teach how to solve it to a schoolkid and it is still really hard to practice for.

[learc83]

You're talking about Raven's Progressive Matrices and variations thereof.

RPM only measures a very small subset of what we normally thing of as general intelligence, and they correlate with general intelligence less strongly than vocabulary. They definitely aren't nearly as good at predicting future college performance as the current SAT is.

>You can teach how to solve it to a schoolkid and it is still really hard to practice for.

It's actually not, you can definitely practice for them. And they are very sensitive to repeated testing.

Here's a cited Psychology Stack Exchange answer with a good summary: https://psychology.stackexchange.com/questions/20177/does-pr...

RPM was developed to be free from cultural bias, but we now know that this isn't the case, they can actually be more biased than verbal tests depending on the culture.