[estevez]

I don't get it. Family income is a very strong (if not the best) predictor of SAT score. In this case, if you order each race by median income you get the exact same order as mean SAT score, Asian-American families with the highest median income and SAT, African-American families with the lowest.

[rottencupcakes]

Correlation does not imply causation.

You have, in fact, provided no proof that family income isn't ordered that way as a result of SAT scores.

[esrauch]

That could be verified by looking at first-generation SAT takers, eg immigrant families or perhaps historical data from when SAT wasn't so common in the US. It seems extremely intuitive to me that families with higher income raise children that have higher SAT scores though.

[estevez]

>Correlation does not imply causation.

Precisely my point. By my reading it appears that the only empirical evidence for race-based discrimination (an extraordinarily grave accusation) is something that could just as easily be explained by factors other than racism.

But this is no defense of those elite institutions, mind you. I'd even go so far as to make an exception and tax the staggering returns on their endowments.

http://harvardmagazine.com/2011/09/harvard-endowment-rises-t...

[fluidcruft]

Causation is irrelevant. The question is the extent to which SAT score and family income contain the same information.

[lurker17]

Causation is measure of the extent to which they contain different information.

Assuming a college wants to admit a student with high "intelligence" or intellectual potential, which cannot be measured directly.

If high income causes high SAT, and high intelligence also causes high SAT, then (low income, high SAT) may well indicate much higher intellectual potential than (high income, high SAT)

If high income and SAT are both caused by high intelligence, then there is less (intelligence-related) reason to prefer (high income, low SAT) or (low income, high SAT).

[microarchitect]

Are you suggesting that affirmative action should be based on family income, then? Seems like a good idea to me, but I might be missing something.

I'd be interested in a break up of admission statistics, SAT scores, race and family income for the Ivies. because I'm not entirely convinced that just because SAT scores follow family income in the general population, the trends are exactly the same among admitted students.

[estevez]

Yes, to an extent. I think the focus on what the ivies do or don't do distorts our view (the majority of US undergraduates attend community colleges, for example) of what truly is broken in higher ed. I've commented before on it here:

http://news.ycombinator.com/item?id=3309894

[michael_miller]

It's not fair to take affirmative action solely on family income, as the cost of living varies vastly across the US. $100K in Manhattan is nothing, but can translate to an extremely comfortable lifestyle in rural Texas.

[_delirium]

What I found most lacking among my own relatively diverse class (relatively diverse for an engineering school, anyway) was actually people from areas with low cost of living, aka "poor areas". For example, the school boasted that the incoming class represented 49 of the 50 states; the one missing was West Virginia. And in general people from places like West Texas or other non-major-metropolitan areas were underrepresented. Also true among minorities; there were a few black students from Los Angeles and Atlanta, but none from Mississippi.

[estevez]

Not solely on family income (but I'd suggest that if 100K is nothing in Manhattan one might be better served by looking a bit farther afield, Brooklyn perhaps...), but as it currently stands low income students receive no advantage at many elite institutions.

http://www.nytimes.com/2011/05/25/business/economy/25leonhar...

[daveed]

Can you provide a source for this?