Hacker News Reader

Comment by xkcdefgh | Hacker News Reader

Why is SAT so ridiculously easy that thousands have perfect score?

It was "recentered" in 1995. The right side of the distribution was essentially truncated. A 1600 on the pre-1995 SAT was quite an accomplishment (I don't have data at my fingertips, but in most years, fewer than 10 students achieved a perfect score of 1600 and only a few dozen scored 1550+). By contrast, several hundred students a year scored 1600 on the recentered test.

The scoring system has been changed several times since 1995 and I have no idea how the current system compares to the pre-1995 system or the recentered test given for some years afterward.

Many, including myself, suspect that one of the motivations for the recentering was to give colleges cover for their admissions decisions. It is pretty hard to turn down students who get 1600s when only 7 or 8 students a year do so. It is easier to "shape" a class without provoking criticism when a large percentage of students score 1500+.

It's hard bordering on impossible to make a single test that measures most of the college-bound high school graduate population and can also make non-arbitrary distinctions among top-1% students.

Lopping off the top of the distribution looks to me like a recognition that pre-recentering scores in the top 100 points or so were essentially measuring the luck of top-performing students on which questions they drew and how well their educated guesses landed.

Why not administer a second test to distinguish between the SAT's high scorers? Students who knew they were very strong beforehand could just take the hard test to begin with.

They have these, they're the SAT subject tests.

It's a metric you can actively improve on by studying. Ala, https://en.wikipedia.org/wiki/Goodhart%27s_law

It's not. A perfect score puts you in the top 0.0348% (2014 test). That about 1 in 2900.

It only seems like a lot because millions take the test.

There's a lot of kids taking it, and it needs to be able to differentiate kids in a country with very wide ranging abilities. If you made it harder there'd probably be a whole load of people bunched up at the bottom.

Okay serious question, which has the larger negative societal impact - making it harder to differentiate between the top tenth of a percent of college applicants, or making it harder to differentiate between the bottom ten percent of college applicants?

I only got a 1500/1600 or 2240/2400. Sad to say I’m probably in the lowest quartile of the HN readerbase.

Guess maybe it’s my IQ holding me back...

Reply on news.ycombinator.com