CHAPTER 2

Understanding the CAT

The GMAT is a computer-adaptive test, or CAT. The test is called “adaptive” because, in the course of a section, the test notices whether you answered the previous question correctly or incorrectly and “adapts” in its selection of the next question.

A few basic rules make the adaptive format possible.

THE CAT EXPLAINED

Here’s how the adapting works. You start the section (Quantitative or Verbal) with a medium-difficulty question; about half of test takers get it right, and half get it wrong. Those who answer correctly get a harder question for the second item, and those who answer incorrectly get an easier item. This pattern repeats: Throughout the section, if you got the previous question right, generally you’ll get a harder question next. Conversely, if you got the previous question wrong, generally you’ll get an easier one next. This pattern reiterates so that you follow a generally upward, downward, or flat trajectory through the questions. The test homes in on the difficulty level that is best matched to your performance; at that difficulty level, generally, you’ll get about half the questions correct and half incorrect. How high on the difficulty scale you end up is one of the criteria that determine your score, along with how many questions you answer and other factors.

image

A Rough Schematic of How Adaptive Scoring Works on a CAT

The adaptive design of the test has two purposes:

  1. Accuracy: A CAT is allegedly more accurate than a “linear” (i.e., nonadaptive) test because it zeros in on a test taker’s ability level. Lucky guesses cause the GMAT to give lucky testers harder questions that they cannot answer correctly, thus eliminating any gains resulting from chance. Conversely, unlucky arithmetic errors on tough problems give unlucky testers easier problems, and these unlucky testers should be able to get the easier questions right, thus correcting the nonrepresentative drop in score.

  2. Time: CATs can be made shorter than comparable linear tests, and the shorter duration is a benefit both to you and the testmaker. The reason for this efficiency is that a CAT does not waste questions. If you get most of the questions right, you pretty much never see an easy one, and if you get most of the questions wrong, you pretty much never see a hard one. On a linear test such as the SAT, on the other hand, everyone gets the same mix of easy and hard questions. On such a test, students struggling on the easy questions will do little better than chance on the challenging problems, while high-scoring students will get close to 100 percent of the easy questions correct. Thus, giving low-scoring questions to high-scoring students (and vice versa) doesn’t actually provide much useful statistical data. In this respect, many questions are “wasted,” whereas the CAT can afford to be a much shorter test at equal accuracy.

Those points define the basic pattern of the CAT, but there are additional bells and whistles in the algorithm. One of the most important details to be aware of is that the test does not always adjust difficulty level question by question. Therefore, avoid the temptation to assess the difficulty level of a question you’re on or to infer whether you got the previous question correct. Even if you could precisely assess a question’s difficulty level (and you can’t, in practice, for reasons we discuss partly later), you wouldn’t be able to draw any conclusions, since the test doesn’t always adapt immediately.

The experimental questions are another refinement to the CAT formula. Some of the questions in each section do not count toward your score. The testmaker must try future questions out on people who do not know that they are experimental in order to determine the validity and difficulty of the questions. We’ll talk more about this topic later, but we’ll give away one headline early: do not try to guess which questions are experimental.

ARE THE FIRST QUESTIONS MORE IMPORTANT?

One of the most frequently asked questions about GMAT scoring is “Are the first 10 or so questions more important?”

As we’ve discussed, the GMAT adaptive algorithm starts with a medium-difficulty question. If you get it right, your next question is harder, and if you get it wrong, your next question is easier. The swings are relatively large at the beginning but then zero in on an estimate of your performance. For that reason, you may find it tempting to spend lots of extra time at the beginning of the test.

The short word on that idea: don’t.

The testmaker concedes that the computer-adaptive testing algorithm uses the first 10 questions to obtain an initial estimate of your ability. The key word, though, is initial. As you continue to answer questions, the algorithm self-corrects by computing an updated estimate on the basis of all the questions you have answered, and then it administers items that are closely matched to this new estimate of your ability. Your final score is based on your responses, the difficulty of all the questions you answered, and the number of questions left unanswered. Taking additional time on the first 10 questions will not “game the system” and can hurt your ability to finish the test.

The testmaker insists that, despite persistent rumors to the contrary, you can’t outsmart the GMAT by spending extra time at the beginning. The reason for this is timing: if you answer more questions correctly than you should in more time than you should, then you will face much harder questions, under more time pressure, in the remaining three quarters of the section. Your short-term gains will be erased.

However, you still want to adjust your test-prep strategy to account for those early swings. Specifically, remember that even when your test-taking skills have become so strong that most of your test will be made up of challenging, high-reward problems, you’ll still have to go through some simpler problems to get there—don’t rush or become overconfident just because those first few questions are easier.

A good comparison is to a sporting event. Are the first innings or the first quarter of a game more important than the following ones? Perhaps, since the early part of the game sets the tone for the game and gives the leading team options. But doing well during the first part of a game does not guarantee a win; you need to start strong and finish strong.

The cost of not finishing strong on the GMAT is substantial. If you don’t answer all the questions, a penalty is assessed that will precipitously lower your score. In fact, this effect is more exaggerated in the case of high scorers. As an example, provided by GMAC, if you are at the 91st percentile but then fail to answer five questions, your score could drop to the 77th percentile. A score difference of that magnitude is substantial.

THE IMPORTANCE OF PACING

The GMAT is a test of both accuracy and speed. There is a substantial penalty for not finishing a section, as we’ve seen. But there is no need to think of the GMAT as a race. In fact, according to the testmaker, the GMAT is created to be optimally timed so that most test takers finish the first time they sit for the test. Those who don’t finish the GMAT the first time often retake the test, and almost all finish the second time.

You want to be in the group that finishes the test on the first try. Also, while you don’t want to rush or make sloppy guesses, you do need to finish the test on time in order to maximize your score.

The graph below is an illustration of the penalty incurred by test takers who leave a string of unanswered questions at the end of a CAT section. Even if you had previously been performing well on questions at a high level of difficulty, running out of time will lead to a severe drop in your score. Fortunately, pacing can be improved through practicing some key principles of time management.

image

Spending Extra Time at the Beginning of a Section Can Lead to Failure at the End of the Section

You can pace yourself on both the Quantitative and Verbal sections, broadly speaking, by dividing each section into three parts:

Each part has its own strategy.

Now you’re done. You’ve maximized your payoff. It can be exciting to set a pace and stick to it, and guessing on the trickiest questions can reduce your anxiety and frustration.

OTHER CAT STRATEGIES

In addition to the strategies mentioned earlier about pacing, keep in mind other CAT-specific strategies that will have a direct, positive impact on your score: