It’s a tip I hear often and I don’t quite understand.
If the time is nearly up and you have to fill in answers for multiple choice, A,B,C,D for example. Why would you only go with ‘A’?
You should get around 25% of the answers correct but no proper test is ever going to have ‘A’ as correct answer many times in a row.
Would it not be marginally better to just pick at random?
Why not?
Edit: I understand real world is rarely truly random, but is my thinking here correct given answers are randomly distributed?
In: 0
afair one should select “C” in that situation as it is slightly more common than the other options usually.
the idea is that this way you “make sure” that some of your answers will be correct, while if you select answers at random you might statistically get as many or more correct, you can also just miss the correct answer each time. but it is very very unlikely that a test is designed in a way that “C” would never be the correct answer for 15 consecutive questions, for example.
If the answers are truly distributed randomly, then any random selection of letters would produce roughly the same score.
Whether they’re actually random or not would depend on who wrote the test.
Whether you’d get a better score by guessing or not would depend on the way the test is graded – many standardized tests *subtract* 1/4th of a point for wrong answers so that a random guesser will get a zero.
I don’t know the stats specifically, but the answers generally weren’t placed at random. In my head, that would likely make the first and last answer the least likely (outside of the last answer being all or none of the above). Thus, the middle answers are more likely to be selected for the correct answer. “C” then naturally allows you to read more answers. My unsupported two cents anyways.
If you guess randomly (CBADBA or whatever), it’s possible to be unlucky enough to get every guess wrong.
If you guess AAAAA, you basically guarantee that about 20% of them are correct.
Provided the scattering is random, or at least random enough given the sample size, then there’s not actually a statistical advantage to picking one letter vs picking a scatter.
Only *slight* advantage at just filling in a single letter is it saves time guessing if you’re real short on time.
If you ever do this though make sure you understand all the grading rules, some tests deduct points for incorrect answers.
As others have stated, *if* the exam has its answers distributed truly randomly (or at least sufficiently randomly, i.e. by a computer), and *if* all of the answers had their choice selection decided independently, then your guesses will not matter at all. You gain no statistical advantage by any strategy. You are simply rolling an X-sided die Y number of times, where X is how many choices each question has and Y is how many questions the exam has.
The adage that you should select the same letter multiple times in a row to get an edge stems from two things, one of which is completely unrelated (and may not apply) and the other only holds if the assumptions we made aren’t true.
The first is about speed. If you mark every question with *something*, you are statistically expected to get *at least* a score of 1/X on all those questions you marked. So if you anticipate the possibility that you might not even finish your exam, if you have them all pre-marked (switching answers to the correct ones as you read your way through the exam for real), then you might get some extra scoring for your guesses on questions you may have otherwise marked blank. This only works, of course, if the exam you are taking doesn’t penalize incorrect answers. An exam that marks non-answers and wrong answers the same benefits from guesses; an exam that subtracts points for wrong answers and does nothing for non-answers punishes guesses.
The second is that there is some evidence that in multiple-choice exams with answer keys arranged by humans one letter is statistically more likely to be the answer for any given question. In the common four-choice arrangement with A, B, C, and D, that letter tends to be C. So, provided your exam was written by a human, *and* the exam doesn’t penalize guessing, answering all questions with C has a statistical advantage over random guessing.
Okay, why is selecting a straight line of ‘C’s better than filling in with no rhyme or reason like the test would have looked if you actually took it?
I think this is what my question boils down to.
On every particular question, if you’re guessing, you have no reason to prefer any letter over any other. So there is no question where you can expect to do better by guessing B than A. So you gain no benefit, on average, by trying to vary your guesses. Just because you’re trying to imitate something unpredictable doesn’t mean you should act unpredictably. If you’re interested, here’s [an extended essay on the subject](https://www.lesswrong.com/posts/msJA6B9ZjiiZxT6EZ/lawful-uncertainty) that I particularly like.
On the other hand, trying to guess randomly isn’t free. It takes attention, effort, and time. If you pick the letter you’re going to use as a guess in advance, you can guess it without taking time to think about which letter you feel like guessing in order to be the most random today. That could save you a critical few seconds that allow you to answer one more question based on the actual wording of the question instead of a guess.
Latest Answers