|Assuming (C) is the most common answer makes no sense. Just like this picture.|
People say dumb things about the SAT a lot. It’s really common to hear someone say the thing about the most common answer being C. Honestly, I set out to write this post to eviscerate those people, but then I did my own research and saw that they’re wrong, but not as obviously as I had thought. So this’ll be more of a light ribbing than a true evisceration.
Let’s be clear: it’s not true that C is the “most common answer” on a given test. It’s straight-up not, and guessing based on that is tantamount to relying on thaumaturgy to improve your SAT score. It’s a poor excuse for strategy and preparedness.
However, it turns out that if you look at all the tests in the Blue Book in aggregate (my raw data is here, in case you’re curious where I’m getting all this) C is, in fact, more common than most answers, and only less common than D. The least common answer is A. WTF?
Answer choices are ostensibly determined at random, so you’d think with a large enough sample size (the Blue Book is 10 tests x 160 questions per test, so 1600 questions) all the answers should appear pretty close to an equal number of times. Across all 10 tests, here’s how it actually breaks down:
Honestly that’s more variability than I expected, but it’s not outlandish and it does nothing to prove that the answer distribution isn’t random. It’s still superstition to say that the most common answer is C (and don’t get any ideas about assuming it’s D now instead) but it turns out it’s also a dumb rumor to say that every answer appears an equal number of times on every test. So…yeah.
The tests in the Blue Book (besides the first 3) aren’t released tests, so maybe it’s different with the real things? Looking at the data just for the first 3 tests (below), I’m thinking probably not.
Three different tests (all actually administered), three different winners in the “most common answer” contest.
UPDATE: Interesting, if slightly mocking, discussion going on about this at collegeconfidential.com. Make special note of the post about the random integer generator for a pretty convincing argument that any Blue Book variability really is just noise.