> In conclusion, our work focuses on the predictive power of admissions tests and shows that this power is not an artifact of SES. There is a substantial SES–test relationship, though it is important to note that test scores reflect far more than SES. About a quarter of the variance in test scores is shared with SES, and thus there is large variability in test scores at any given level of SES. Thus, claims that tests are merely proxies for SES are unfounded

I've never understood the model of the world that people who don't agree with this. If you believe that

a) genetics exist, and traits such as intelligence are therefore hereditary; and b) socioeconomic success is correlated to intelligence

Then the inevitable conclusion is that wealthy people will be more likely to have higher SAT scores. Of course this is obviously just a correlation; lottery winners are rich but with no correlation to intelligence, children of two intelligent parents tend to regress to the mean, two average parents can produce intelligent children, families might be poor simply through bad luck or some particular emotional/mental block they have, and a million other factors. But nevertheless we'd expect test scores to be our best available wealth-blind predictor of college success and later success. Certainly it's hard to imagine children with money-poor - or maybe mildly abusive - parents having a shot without test scores. They'd be less likely to have a deep extracurricular involvement or anything else requiring parental involvement.