undefined | Better HN

0 pointspjscott6y ago0 comments

Imagine a perfect SAT that preparation can't affect; for the purposes of the hypothetical, the test just stares into your soul and measures pure scholastic aptitude.

If (a) scholastic aptitude has a significant effect on your lifetime income, (b) people tend to marry people of similar social and economic class, and (c) scholastic aptitude is fairly heritable (through genes, environment, whatever)...

... then you would expect the Perfect SAT scores to correlate pretty noticeably with family wealth. This wouldn't be a sign that anybody is doing anything wrong; it's just as natural as water flowing downhill. And these are reasonable premises, with strong empirical evidence for each of them.

So, question: just by looking at the correlation of SAT scores with family wealth, how can we possibly tell how broken the test is? How can we know how far the real SAT is from the absolutely un-gameable Perfect SAT?

0 comments

starpilot6y ago

Because we've done studies to address exactly that question. What we've found: The least-gifted children of high-income parents graduate from college at higher rates than the most-gifted children of low-income parents, hence the adversity score to rectify the inequity in life circumstances.

https://www.independent.co.uk/news/science/money-academic-su...

ghufran_syed6y ago

This study has nothing to do with the SAT. Clearly, money is required in order to study full-time in the US, both for tuition and living expenses, so people with more money graduate more often because it's easier to not work during school, and to pay for tuition. But none of that is relevant to the point you replied to, which was that even a perfect test would have results that correlate with wealth

droithomme6y ago

> The least-gifted children of high-income parents graduate from college at higher rates than the most-gifted children of low-income parents, hence the adversity score to rectify the inequity in life circumstances.

Looking at graduation rates instead of admission rates is problematic. Less gifted and underprivileged people also graduate at lower rates despite being admitted using affirmative action procedures as the College Board is now implementing. This means they do not have a degree that would advantage them, but are saddled with undischargable education debt that puts them at a severe disadvantage for the rest of their life. The system not only continues the inequality but makes it much worse. If college was free or education debt was dischargable in bankruptcy then it would not be the oppressive system it is currently where people unable to succeed are admitted, then forced to pay back debt they can not afford after they fail to graduate. Only a sadist and a racist would support such an inequitable abusive system.

TheOperator6y ago

So we need to correct the SAT so that the students most likely to be doomed to fail will be admitted to higher education?

To me it doesn't sound like you're trying to make the SAT reflect say academic potential. It seems like you're just trying to inflate the appearantly already inflated scores of the disadvantaged. Who will proceed to not get scholarships, fail out, and be saddled with debt they can't pay off or be discharged. This of course is the road to an egalitarian society!

Who needs to bother with difficult things like making the disadvantaged do a little better in school? Just add 100 points to their SAT. Problem solved.

Dylan168076y ago

Once you combine all those premises, the "pretty noticeable" correlation might be something like ten percent. So the problem of it correlating too strongly with wealth would be solved.

rmtech6y ago

> combine all those premises, the "pretty noticeable" correlation might be something like ten percent

it might be, but it will probably be significantly higher.

logicx246y ago

I mentioned this below, but if the test is a perfect predictor of future success, then the people that perform well on it are absolutely worthy of college admissions, and there should be no further analysis required. If the test is perfect, then the characteristics of the people that do well on that test don't matter. The reason we need to do implement equalizers like this is because such a test is impossible.

1 more reply

j / k navigate · click thread line to collapse