GRE > GRE Computer Adaptive Testing (CAT)
How a Computer Adaptive Test Differs from a Paper-and-Pencil Test
Computer Adaptive Tests, or CATs, are quite different from the paper-and-pencil standardized tests you probably have seen in the past. Aside from being taken on computer at a special test center, the main difference between Cats and paper-and-pencil tests is that Cats "adapt" to your performance.
Each test taker is given a different mix of questions depending on how well he or she is doing on the test. This means the questions get harder or easier depending on whether you answer them correctly or not. Your score is not only determined by how many questions you get right, but by the difficulty level of these questions.
How a Computer Adaptive Test finds your score
When you start a section, the computer:
- Assumes you are of average ability (about 500 on the GRE).
- Gives you a medium difficulty-level question. About half the people who take the test would get this question right, and about half would get it wrong.
What happens next depends on whether you answer the question correctly or not:
- If you answer the question correctly, your score goes up and you are given a slightly harder question.
- If you answer a question incorrectly, your score goes down and you are given a slightly easier question.
This continues for the rest of the test. Every time you get the question right, the computer raises your score, then gives you a slightly harder question. Every time you get a question wrong, the computer lowers your score, then gives you a slightly easier question. In this way the computer tries to "home in" on your score. Theoretically, as you get to the end of a section, you will reach a point where every time the computer raises the difficulty level of a question, you will get it wrong, but every time it lowers the difficulty level of a question, you will get it right.
To do your best on a CAT, you must have a grasp of the mechanics by which the CAT finds your score. The best way to learn how a CAT works is to actually take one. But the way the test changes from question to question on the real GRE CAT is quite subtle.
So here are two short (4 question) adaptive tests where the changes are more obvious than on the real test to demonstrate how a CAT finds your score. As you're working on these tests, pay close attention to how your score goes up and down and to how the difficulty level of the questions changes. (Note that general knowledge questions such as those found on the World Capitals test are not found on the real GRE. We include them here to illustrate how a CAT works.)
CAT Strategies
A CAT differs from a traditional paper & pencil test both in how it works and how it determines your score. Plan your test-taking strategy with this in mind.
On the CAT...
the computer shows you only one question at a time and does not allow you to see the next question until you've responded to the one at hand. You can't go back to change an answer once you've responded and gone onto the next question. So you should...
Double-check your answers before moving on. If a question is taking a long time to complete, cut your losses. Eliminate answer choices you know are wrong, guess one of the remaining choices and move on. Don't get bogged down.
On the CAT...
it is easier to change the computer's estimate of your ability at the start of the test than at the end of the test. (Click here for a more detailed explanation of why this is so.) So you should...
Spend more time on the early questions in order to make sure you get as many of them right as possible. On the first third of each section you should double check your work before moving on.
On the CAT...
there is a penalty for unanswered questions: any questions at the end of a section that you do not reach will hurt your score. So you should...
Make sure you answer every question on the test. If you have any questions left as time runs out, guess the answer to these questions rather than leave them unanswered. Getting them wrong will hurt your score less than not answering them at all.