How the GMAT CAT Works

The Quantitative and Verbal sections of the GMAT both consist of multiple-choice questions that are presented as a computer-adaptive test (CAT). The CAT is designed to dynamically produce exam questions based on your performance on previous questions. These questions, which range in difficulty from low to high, are pulled from a question bank. The result is a unique exam for nearly every test-taker.

The CAT begins by providing a question of moderate difficulty. About half of test takers are expected to get the question correct, and half are expected to get it wrong. If you get the question correct, then a more difficult question is supplied. If you get the question wrong, then an easier question is supplied. By the end of the test, the CAT finds your score based on the number of correct answers given to questions at different levels of difficulty.

As you might imagine, there is a sophisticated algorithm underlying this process, involving knowing how thousands of other test-takers have done on each of the problems you answered. The algorithm decides the appropriate level of difficulty for your next question, the value of that question, and eventually your score. Because your questions are generated based on previous responses, you may not skip a question and return to it, as is possible on some other standardized tests.

Over the years, many myths have come and gone come with regard to how the exam’s adaptive engine works. Perhaps the most pervasive one is the idea that, because the first few questions can cause a wide swing in the whether you receive harder or easier questions (based on whether you get them right or wrong), these questions are by far the most important ones in determining your final score. Many test-takers therefore spend far too much time on these questions, trying to ensure that they get every one right so that the computer puts them in the “smart” bucket, ensuring they will get a higher overall score at the end of the test.

The reality is that the GMAT holds so many data points on every one of the questions you will see that by the end of the test it will know your true level of ability, whether or not you get an early easy question wrong or an early hard question right. It’s also flexible enough to account for “false negatives” and “false positives” (getting a question wrong when you really should have gotten it right, given your ability level, or vice versa). The people behind the test know, for example, that even if you make a pure guess, there’s a 20% chance that you will get a question right. They therefore know that every question helps in determining your final score, but that every question itself may be “wrong” in making that determination. So, they designed the scoring algorithm to stay flexible enough to withhold judgment on your true level until you’ve been presented with every question on the test.

Therefore, you are far more likely to impact your own score by finishing the exam or not. Spend too much time on these first 10 questions and you may risk running out of time at the end of the section. No matter how well you did on the first 10 questions, not answering every question can knock your score down significantly. (The people at GMAC have stated as much publicly.) So, don’t fool yourself into spending precious extra minutes on the first 10 questions… Instead, you should worry about finishing the test on time!

For more business school admissions advice, take a look at our book, Your MBA Game Plan, now in its 3rd edition. And, be sure to find us on Facebook and follow us on Twitter!

Advertisements

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s