kitchen table math, the sequel: fyi

Saturday, June 26, 2010


SAT percentiles


SteveH said...

I've never spent much time analyzing the SAT. They are interesting curves no matter which way you plot them. Is it strictly a bell shape curve applied to the test, or is it a slightly different shape? Also, don't they try to set the 50% percentile to 500? I assume that the 1st percentile is at 200 to account for the (25%) success of guessing. Is each question on the SAT weighted equally and there is straight mapping of % correct to SAT score? I've heard that you can get an 800 even if you don't have a perfect score. If it isn't a linear mapping, how is it done?

Since the curve has been recentered (lower) over the years, does that mean that people are getting dumber?

Also, how many groups of questions on the test are related, meaning that they are based on the same knowledge or skill in math? I've seen many tests where if you get one question wrong, you will get many questions wrong because they are based on the same skill. Is the math SAT like that? If you keep mixing up sine and cosine when you solve a right triangle problem, how many questions will you get wrong?

SteveH said...

OK, I see now that they must define a standard bell curve and then map the raw % correct to that shape. I suppose that % correct above a certain amount all get mapped to 800. What is their mapping formula of % correct to SAT score?

ChemProf said...

I don't think they publish their formula, but you can find tables like this one:

Note that guessing on all questions should get you 0% score (since they subtract 1/4 for each wrong answer) not 25%.