Hacker News Reader

Comment by _m8fo | Hacker News Reader

The SAT is a pretty silly test. When it was out of 2400 I took it for the first time and got a 1800. I studied for literally a week and increased my score to 2100, which isn't really that good, but a testament to how "studyable" the test is.

If education in this country was worth anything the SATs would be unnecessary.

View on HN

If it is anything like the ACT, it doesn't test knowledge, it tests how well you take tests.

The science section in the ACT is more chart comprehension than anything.

Everyone gets tripped up thinking they need to know more when really they just need to learn to take it better.

Did they change the score again? I thought they just changed it to being out of 2400 but it's not anymore?

Yes, it is back to 1600.

http://www.cnn.com/2014/03/05/living/sat-test-changes-school...

The test will shift from its current score scale of 2400 back to 1600, with a separate score for the essay. No longer will test takers be penalized for choosing incorrect answers.

Is it the writer being confused, or did they add a penalty for incorrect guesses for a few years? I took it in 2010 and am quite confident that AP tests were the only thing doing that at the time.

When I took the SAT in 1982, 1983, and twice in the late 1980s, there was a (statistically fair) penalty for wrong answers.

It was set such that if you could eliminate even one answer as "surely incorrect", it became advantageous to guess among the rest of the choices.

In the 90s, you lost a quarter credit for incorrect answers.

How do you know your ability to do that isn't rare?

Agreed - most research on SAT studying seems to suggest that this is rare. Big gains are most prevalent with either smart-but-unstudied people who get large gains from learning what to expect, or under-educated people who weren't previously taught major test components like algebra. A 300 point jump is extreme.

I don't know if a 300 point jump is extreme or not, but I'd like to point out that most people who do poorly on the SAT probably weren't getting the education they needed, which was my overall point.

Put differently, I assert that all people who don't do "well" on the SAT fall into the two categories you posed.

> I assert that all people who don't do "well" on the SAT fall into the two categories you posed.

I think that's provably wrong. SAT prep courses offer a bump, but even ones with big promises don't deliver all that well. Evidence suggests 300 points is very large.

The "smart but unstudied" pool I mentioned is likely small and tends to get strong gains up front - these are people who get 100+ points from the PSAT or 1-2 practice tests. (This pool is anecdotal. I knew several people in it, but I don't know of studies at the high end of the scale.)

The other pool is real but not the same as you described. It largely represents people who missed core content like 'geometry' and did prep so long and extensive they learned totally new material there. There are lots of kids with low scores from bad education, agreed, but fixing that is not "studied for a week" stuff.

More broadly, I'm saying that in practice "I studied for a week and gained 300 points" is an extreme outlier. Mostly people either see <100 gains from studying or only see big gains from serious, new-content education.

You'd also need good data for test-retest variability here.

Would you expect to get the same exact score taking the test twice if you hadn't studied? How much of the point increase could be ascribed to luck?

I have a vague idea that many-test variability for the SAT is inside 50 points, but it's possible there's a distinct practice gain between the first test and all later ones. Also, this is all hazy unsourced memory.

When I took it, many people got hung up on analogies. Taking a review course would help many students dramatically. Also non native speakers often struggle with the deliberately obtuse instructions.

In my personal case, I went up about 50 points vs the PSAT. I had some trouble with some of the math problems, but once I worked through them a few times it was no big deal.

Reply on news.ycombinator.com