The best guess...?

[Number7]

Dear all,

- First of all: Sorry for the many topics I start. These weeks there is no day I do not ask myself anything new -

Question:
I sometimes run out of time and there are two or three questions unanswered. What I would be interested in: Is there an answer-number that appears to be the right answer more often in the official test than other answer numbers (perhaps dependend on the question-type) - or are such thoughts bullshit?

E.g.: Is there a favorite answer-number (for example answer D) at the problem solving section that could raise my chances if I even do not have enough time for an educated guess?

Thank you for your help!

[Ian Stewart]

Is there an answer-number that appears to be the right answer more often in the official test than other answer numbers (perhaps dependend on the question-type) - or are such thoughts bullshit?

I suspect the latter is true, but I looked into this anyway. If we go by the (orange) Official Guide, we discover that of the 249 Problem Solving questions, we have the following distribution of correct answers:

A 42 16.9%
B 57 22.9%
C 43 17.3%
D 49 19.7%
E 58 23.3%

Now, even if the 'true' frequency of each answer is 20%, you wouldn't expect each to show up exactly 20% of the time if you look at a fairly small pool of questions. We can use properties of binomial distributions to figure out whether the above numbers really do indicate that E is, statistically, likely to be a better choice than the rest. And it turns out that the above numbers are entirely consistent with the hypothesis that each answer appears about 20% of the time. Nothing unusual here, and I'd still expect that guessing randomly is just as good as guessing E.

On Data Sufficiency, we discover the following, looking at the 155 problems in the Official Guide:

A 24 15.5%
B 29 18.7%
C 30 19.3%
D 43 27.7%
E 29 18.7%

This is a bit more interesting. Statistically, if the average frequency of each was 20% (and therefore, the expected frequency was 31 of each answer when looking at a pool of 155 questions), then you'd expect a standard deviation of about 5. The frequency with which D is the correct answer is quite high, from a statistical point of view- this is a bit unusual.

So if you had to guess entirely at random, I'd suggest guessing D for DS- it can't do any more harm than guessing randomly, and there's a small chance it might help. On PS, it probably makes no difference what you guess. I'll add that I would still bet a fair bit of money that each answer choice is correct as often as any other, despite the statistical anomaly in the DS data above.

Finally, I'd add that on DS in particular, if you even have time to read the question you can normally eliminate a couple of wrong answers. That's going to help you a lot more than any 'guessing guideline' you arrive at by looking at the stats above.

[Stuart@KaplanGMAT]

Kaplan has reviewed a LOT of GMAT questions looking for guessing trends. We've only found 1 that's reliable:

on regular (i.e. not Roman Numeral) problem solving questions, if the question asks "which of the following" (or some variation thereof), the answer is D or E more often than it should be (i.e. more than 40% of the time).

For all other question types, the correct answer appears to be randomly distributed.

[Number7]

Thank you very much for your input! I really appreciate it.