How Accurate Are GMAT Practice Tests?

You should follow me on Twitter. While you're at it, take a moment to subscribe to GMAT Hacks via RSS or Email.


In a perfect world, every GMAT test-taker wants to reach their target score on a practice test before going in and taking the real thing. It's an admirable goal, and sometimes it's achievable.

There are some drawbacks, though. I've described a few of them before in an article on practice tests in general. They aren't the best use of your study time, especially once you've taken more than one or two of them. The other criticism I mentioned is that they aren't that reliable.

We can separate all of the practice tests on the market into four categories:

  • GMAT Prep practice tests (2), available from
  • Tests from Manhattan GMAT, Princeton Review, and Kaplan
  • Adaptive tests from all other sources
  • Non-adaptive (paper) tests.
The last two of these categories are so worthless that I'm not going to discuss them further. If you're using them, you are almost certainly wasting your time.

As I mentioned in that earlier article, the best tests out there are the ones distributed by the test-maker itself. For the most part, they are the most reliable for the same reason that the Official Guides are the best books to use to prepare for the test: the makers of the test made these resources. If that isn't enough to convince you to use them, I'm not sure I'm going to make much headway with you.

To be more specific, the accuracy and reliability of a test rests on two factors:

  1. The realism of the adaptive algorithm
  2. The realism of the practice questions

Apart from some general characteristics the test-maker has shared, the adaptive algorithm is a closely-held secret. Try as they might, Kaplan and the other test-prep companies are unlikely to come very close to designing a practice test that approximates the real experience. In fact, their best efforts aren't (in my opinion, anyway) anywhere close.

Because of that, easy questions are hard questions are going to "cluster" more on those practice tests than they do on the real thing. It's hard to describe this in detail, but the experience you get taking a commercial practice test and the experience of the real GMAT is very, very different.

The other consideration is the nature of the practice questions. This is another hobby horse of mine; I've written about it, as well. Again, the commercial test-prep companies try to create accurate, realistic test items, but in large part they are unsuccessful. Even if they are exactly right 90% of the time, that leaves 3 or 4 questions per section that don't ring true. That's bad practice, and it leads to an inaccurate score.

What about the scores?

Before we start talking about the different types of practice tests, a note is in order. As I wrote in this article, how you score on any given day is at least a little bit random. Your mood, the quality of your sleep, the exact questions you're given; all of those things influence the precise score you get. Someone who gets a 650 on Tuesday may well get a 690 on Wednesday, and not because they stayed up all Tuesday night cramming.

So, accept the fact that any test, no matter how accurate, is going to have a substantial margin of error. Let's say that margin is +/- 40 points. In other words, if you got a 600 last weekend, that could mean anything from a 560 to a 640, and that's without taking into account the quality of the practice test you used.

The GMAT Prep tests are by far the most accurate. They are nearly as accurate as the real thing except, perhaps, on the high end. While a 550 GMAT Prep score is reliable, I might not get quite so excited about a 750. It's impressive, yes, but your likelihood of replicating that under test conditions is much lower than it would be for the 550 scorer.

If you're using commercial tests, such as those from Kaplan, Princeton Review, or Manhattan GMAT, I have bad news. Because of the flaws in their practice questions and the adaptive algorithms, the margin of error is probably more like 80-100 points. I've heard of many experiences where students repeatedly get 680s, then get a 580 on test day. Or, believe it or not, the other way around.

The moral of the story is...

To repeat myself: you don't need to take a lot of practice tests. If you haven't taken any so far, plan on taking the GMAT Prep tests and nothing else. Most of all, don't take 8 or 10 tests just because you have them and you think you have nothing better to do.

If you have already taken commercial tests, use this article as a guideline for evaluating your progress. As you can tell, the tests themselves are very unreliable for that purpose; stick with the best practice tests you can, and spend the majority of your time doing and reviewing quality practice questions.



About the author: Jeff Sackmann has written many GMAT preparation books, including the popular Total GMAT Math, Total GMAT Verbal, and GMAT 111. He has also created explanations for problems in The Official Guide, as well as 1,800 practice GMAT math questions.

Total GMAT Verbal

The comprehensive guide to the GMAT Verbal section. Recognize, dissect, and master every question type you'll face on the test. Everything you need, all in one place, including 100+ realistic practice questions.
Click to read more.