Hacker News new | ask | show | jobs
by mothers 2802 days ago
To those who have a problem with Harvard's alleged discrimination. Do you think female-only schools should exist? In addition, would you agree with an admission policy that only uses SAT and AP scores? Grades inherently are subjective and can unfairly be given on the basis of race on the K-12 level, so that's helpful.

Harvard should just model China and make the SAT the only criterion. If there are too many people with a perfect school, use randomization accordingly. Of course, the typical Harvard admit won't like this since it'll mean their private school boost won't mean anything, nor will the fact that they come from a feeder school.
1 comments
dev_dull 2802 days ago
Goodhart's Law: “When a measure becomes a target, it ceases to be a good measure.”

SATs will quickly become one of the poorest indicators for success.

link
barry-cotter 2802 days ago
There’s an entire academic field designed around making reliable tests with valid results for psychological measurements, called psychometrics. IQ tests and we swear it’s not an IQ test like SATs, the GRE, GMAT, LSAT are all designed to be as close to ungameable as possible and it works. Practice effects exist, you can get better by doing more of them but they’re also limited. Usain Bolt is never going to run 100m in 5s and I’m never going to run a sub 10s 100m no matter how hard I practice. Likewise SATs are not going to stop predicting success in college or in life because they measure g, the general factor of intelligence.

https://en.wikipedia.org/wiki/Psychometrics

https://en.wikipedia.org/wiki/G_factor_(psychometrics)

link
toasterlovin 2802 days ago
Except that SATs measure something real (intelligence), so unless outright cheating and bribery are taking place, it will continue to be a good measure.
link
pgodzin 2802 days ago
Citation needed. The huge amount of time and resources spent preparing for the SATs imply that it is not a direct measure of intelligence.

Is a someone who scored a 2300 after years of test prep necessarily more intelligent than someone who couldn't afford test prep or didn't have access to it and still scored a 2000?

link
bnm123123 2802 days ago
See: https://nepc.colorado.edu/sites/default/files/Briggs_Theeffe...

"When researchers have estimated the effect of commercial test preparation programs on the SAT while taking the above factors into account, the effect of commercial test preparation has appeared relatively small"

link
mlazos 2802 days ago
In my experience, SATs measure nothing - only how well you can take the SAT, after taking a prep course my score went up like 200 points. One of my friends got a perfect score on the ACT after doing alright on the SAT. Some colleges are no longer taking it into account because everyone knows how BS it is.

EDIT: and to make it worse the college board charges exorbitant prices for everything involved in the process - it’s ridiculous

link
dorchadas 2802 days ago
ACT should already be like that. I looked at the last one they sold, and, at least for math, you can easily game your way up to a 22ish on it. Which, granted, isn't needed for higher levels, but that's above what some states set as their benchmark, so...

I'm sure the other sections are the same, though I didn't look at those. But, really, any math exam with multiple choice answers can be gamed, at least to an extent. And then people will start teaching how to game it (especially if funding comes from them, which some does in the state I teach in)

link
chrisco255 2802 days ago
Not sure that holds up under all cases. Grades, as an example, are both measures and targets.
link