Hacker News new | ask | show | jobs
by lordgrenville 2439 days ago
One suggestion I like is to totally change the test format every year. This would basically invalidate test prep and put everyone on the same footing.
2 comments

This would give the wealthy even more advantage because they have the time and resources to throw at the test compared to the poor.

The test changes every year? No problem, just prep for the meta test like how people prep for the FAANG interviews. Multiple formats? No problem, just spend more time preparing for each format. Changing topics? No problem, just cover as many topics as necessary.

I think the key is to see that the poor, by definition, have much less time and money to tackle the problem. Therefore, to have a level playing field we should design the test such that prep may help, but there is a ceiling to it, and I think the SAT is somewhat decent in this respect.

The optimal test strategy can be covered in one afternoon. The format is well known and cheap study guides/sample tests are available pretty much everywhere. If the test changes every year, imagine how much more difficult it would be for a poor student to prep for it.

Edit: typo

If is it just resources how do the poor from certain cultures manage to do well on the standardised tests while other cultures with upper middle class incomes underperform? The problem is a lot more complex than just money.
Agree with you that resources is just one factor and probably not the dominating factor. I am arguing from the point of all things being equal, a test that doesn’t change from year to year requires less time and resources to prep for compared to a test that does change year to year. And for a poor student, the former is probably more equitable than the latter.

As for standardized test scores and ethnicity/culture, that is a sensitive and contentious topic that I prefer not get into.

I've definitely like that as a thought experiment. Whether these tests assess your capabilities is already questionable. I wonder if there would be any objective way to measure outcomes if the tests were entirely different from one year to the next.