What is Admissions Consulting?
Types of Errors in GMAT Sentence Correction: Part 2
Types of Errors in GMAT Sentence Correction: Part 1
A Glossary of GMAT Terminology
Errors in the GMAT Official Guide, Part 2
Guessing on the GMAT, Part 2
Guessing on the GMAT: Part I
GMAT Exercise: Build Focus and Concentration to Perform Your Best
"How to Solve It" applied to GMAT Quantitative Questions
The Few Mistakes in the Official Guide for GMAT Review 11th Edition.
Inside a GMAT Test Center
Sentence Correction Data: A little analysis of Official GMAT questions.
GMAT Score Inflation
The GMAT Exam: A test of Endurance.
GMAT "Tricks" Part 2: Backsolving Revisited
GMAT Math: What you need to know
Retaking the GMAT
Preparing for the GMAT: Legal Issues, Part 2
Preparing for the GMAT: Legal Issues, Part 1
GMAT "Tricks" Part 1: Backsolving
GMAT Registrations Worldwide
AWA: Is it important?
Guessing on the GMAT: Part I
Posted by ian on Wed 08 Oct 08 at 12:33am
We've been asked more than once: 'If I need to guess randomly at a question, what should I guess?' That is, on the GMAT is one answer choice more commonly correct than the others? Well, the GMAT is trying to test fundamental mathematical and verbal skills. It's not trying to test whether you've learned how often 'E' is the correct answer. So if the test is well-designed, each answer choice should be equally common. A recent research article by Lawrence Rudner, GMAC's Vice-President of Research and Design, suggests that the test designers devote considerable attention to this issue- we'll discuss this article in a future entry. Still, just to confirm that no answer was better than any other, we collected some data.
Before getting into that, we want to point out that you shouldn't need to guess randomly very often. On most questions, it should at least be possible to rule out one or two wrong answers. Because there is a severe penalty for not finishing the test, if you do find yourself with very little time to spare and several questions remaining at the end of a section, you should be sure to finish the section even if you do need to guess randomly. But if you have time to read a question, most often you shouldn't be guessing completely at random.
We went through the quantitative sections of the GMAT Official Guide, 11th edition, and tallied how often each answer choice is the correct answer. Note that, from a small sample of questions, you would not expect each answer choice to occur with exactly 20% frequency; there will be some variation. The 'margin of error' (95% interval) for a sample of this size is about +/- 5%. That is, if each answer choice does indeed occur exactly 20% of the time, you'd expect, from looking at 249 questions, it would be normal for the frequency of each answer choice to range from about 15% to 25%.
In the Problem Solving section, with 249 problems in total, the correct answer is:
A for 42 questions (16.9%)
B for 57 questions (22.9%)
C for 43 questions (17.3%)
D for 49 questions (19.7%)
E for 58 questions (23.3%)
While E appears to be the best choice, by a marginal amount, the result is not statistically significant. So, while it won't likely hurt you to guess E on Problem Solving questions if you need to guess randomly, it probably won't help either.
In the Data Sufficiency section, with 155 problems in total, if each answer were equally common, we'd expect the frequency of each answer choice to range from about 13.5% to 26.5%. In the Data Sufficiency section of the OG, the correct answer is:
A for 24 questions (15.5%)
B for 29 questions (18.7%)
C for 30 questions (19.3%)
D for 43 questions (27.7%)
E for 29 questions (18.7%)
Here, the frequency of answer choice D is a bit unusual statistically, given the size of the sample, and the results here do suggest that D is slightly more likely to be correct on Data Sufficiency questions than the other answer choices. That said, the questions we're looking at are from the Official Guide, and were likely designed before GMAT question design was handed over from ETS to ACT. If you did need to guess randomly on a Data Sufficiency question, there'd be no harm in guessing D, and the above data suggests there's a chance it might be helpful, to a very small degree. That said, on Data Sufficiency questions in particular it is almost always possible to rule out a couple of wrong answers if you have a chance to read the question, and you'll be better off making a guess based on a Data Sufficiency question's content than you will be just guessing D at random.