Hacker News new | past | comments | ask | show | jobs | submit login

It's much less wrong than the holistical admission procedure.



FWIW, I've been on graduate admissions committees for many years. Standardized tests work both ways; it really depends on the admissions committee and how they use the test scores. To borrow a phrase, "test scores don't prejudice, people do."

I've seen both sorts of problems.

When test scores are used well, they function the way that some people here see them. They provide another bit of evidence of potential, and can help to provide some signal as to what is going on with someone who might otherwise be underresourced, or facing sociopsychological challenges, or whatever. It's also much harder to score well on a test than to score poorly, so I think that asymmetry is helpful: a mindful admissions committee doesn't hold a less-than-stellar score against someone, but will give credit for an exceptionally high one.

At the same time, I've been in committees where controversies have erupted because of the stereotypical bias situation. There are minorities applying, they have tons of extra-score stuff to attest to their promise, they are coming from ridiculously challenged backgrounds, and they get a middling score, and someone on the committee is arguing that this single score, at a single point in time, should somehow override really high GPAs with challenging courses, volumes of very impressive letters, scores of accomplishments in research and the real world, etc. This is in the context of a program that has lots of experience with disadvantaged students, where you can see that they have this extra "learning the ropes" disadvantage but are equally smart, etc. This is also with committee members who, on paper, seem ultra-woke, but for some reason when it comes to stuff like this become really irrational and ultra-focused on test scores, and come across as highly bigoted.

Anyway, I feel a bit torn in all of this because I've seen both sides of the test score problem and think both sides are right. The problem is that the challenge is really trickier than "are test scores good or bad", the challenge is really about how people in power use them, and that's murkier, and more complex, and fraught with all sorts of obstacles.

The China example is a good one: why have any limits on how often you can take the test? What, really is the basis for that other than some kind of bigotry about human nature? Can people not change? Don't they have rough spots? We're just going to pretend that all those other sources of corruption don't play a role? That tests can't be gamed?

It's not really about the test per se.


Everything has flaws. Everything is game-able. The question is to find the least game-able system. Where is the line between "tests can be gamed" and "one can prepare for tests by educating themselves"?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: