GradSchools.com Test Prep with The Princeton Review:
"If you are thinking about graduate school then the words 'GRE CAT' have likely been on your mind. There is an aura of mystery surrounding these words and it is perpetuated by rumors based on partial knowledge and ambiguity from Educational Testing Service (ETS), the writers of the GRE. The Princeton Review has made it a point to study the GRE at a micro level so that we can bring to you the real deal about how to best approach a CAT, computer-adaptive test. So sit back as we discuss the whats, whys, and hows of a CAT."
What does it mean for a test to be adaptive?
This is an excellent question, and as with all aspects of life, it is quite useful to know the lay of the land. A basic part of a CAT is that it does not give equal weight to each question that you answer. As the registration materials from ETS suggest, the difficulty level of the questions you answer is factored into how much they count toward your score. Yet, it is important to clarify what this means in order to optimize your score when you are taking the test. The adaptive part of CAT means that the computer is assessing who you are from the very first question that you answer. Based on your performance on the prior question, the computer will give you a new question that responds to what it considers to be the appropriate difficulty level for you.
Specifically, when you begin a GRE section (quantitative, verbal, or analytical) the first question will be at the middle level of difficulty. As you may know, each section of the GRE has a score range from 200 to 800. An average score is a 500, so you can associate "middle level" with a 500-level question. If you correctly answer a question, the computer "rewards" you with a more difficult question. If your answer does not match what ETS has determined to be the "credited response", you are given an easier question. A CAT doesn't let you go back to a prior question, so do not move on to the next question unless you are sure about your answer (make your best guess if you don't have a clue). Potentially, each test taker could have a somewhat different test in terms of the questions seen.
But, we've only just begun to discuss the peculiarities of this test. The difficulty level is not the only aspect that changes from question to question; concurrently, the computer is adapting your score based on your performance. This is a big part of the "mystery" that we previously referred to. The easiest way to think about how questions are weighted is to divide a GRE section into thirds. At the beginning of the test, the computer knows virtually nothing about you. As a result, the computer puts a lot of emphasis on your success with initial questions in order to determine what your score range should be. The computer is still doing a fairly close assessment of you in the middle of the test, but past the middle of the section your final score has largely been decided. By the final third of the test, it is extremely difficult to change your score substantially from what has been decided after the second third of the questions.
Adaptive tests are similar to people in the way that we sometimes form impressions when we meet someone new. While not universally true, when we are introduced to someone our minds commonly make quick, and sometimes extreme, judgments about the person. We are told that this is a bad thing to do, but this has not stopped the ETS from using this approach in their testing. Certainly, when we meet that person again we compare their actions to our previous character assessment and see if they match up. After meeting people a few times, we have usually formed our opinion about them and it is very hard to change our minds no matter what they do or how differently they act.
Does this seem like a weird way to test people's abilities or knowledge? Well, as you learn more about ETS, you will see that this kind of behavior is actually very typical. And understanding how ETS thinks is a really important part of doing well on the GRE.
The impact of the adaptive style of scoring is that if a test taker does not make a "good impression" at the beginning of the test, it will be very hard to get a score of 500 or better in a section. While people can change their minds, computers cannot truly adapt their way of thinking. The scoring method is very strict in the sense that it is a computer program. The first third of the questions determines the bulk of your GRE score, the second third accounts for much less of your score, and the final third counts for a relatively insignificant amount of your overall score.
In terms of numbers, we can roughly quantify the impact of questions in different parts of the test. In the first third, questions can increase or decrease your score by as much as 80 points. During the middle of the test, questions tend to change your score by about 40 points either way. At the end of the test, questions move your score by as little as 10 points. See how important the beginning of the test is? After only a few questions, your GRE score can become very high or very low. You can also see why at the end of the test, it is very difficult to make any significant changes to your final score.
Why has ETS decided to score the GRE in this way?
Well, remember, ETS is "hired" by graduate schools to distinguish test takers on an arbitrary scale and there must be a reasonable distribution of scaled scores (the scores from 200-800). Most of us have been taught as undergraduates, and even before, that to do well on a test you should finish. And this is largely true; if you don't finish most tests you are at a disadvantage. Using their in-depth knowledge of test taking behaviors, garnered over several decades of administering tests, ETS writers have designed a test that punishes you for rushing at the beginning.
It may be appropriate here to quote logic from the well-known childhood story about the tortoise and the hare, "slow and steady wins the race." The goal of finishing is clearly not the best approach for most people taking the GRE CAT. This is not to say that you should not answer all the questions, you want to answer all the questions, as we will explain below. Rather, we are suggesting that most people will want to work carefully up to a certain point and then guess on any remaining questions. How many questions you do and the time you spend at the beginning of the test require a balance that you must achieve in order to optimize your score.
How can you benefit from this knowledge and what should you do?
Our main advice, be careful. You want to be sure about your accuracy on early questions, and worry less about getting to the last questions in a section. Remember, later questions have a progressively smaller impact on your overall score, but you do need to answer all of the questions. If you are running out of time, just guess. Do not leave any questions unanswered because ETS will lower your score. For example, by leaving 10 questions blank in the verbal section your raw score (a number that is converted to the scaled score of 200-800) would be lowered by one-third (10 out of 30 questions).
As mentioned above, incorrect answers do lower your score, but at the end of the test the penalty is quite low. Unless you are scoring above 750 in a section, leaving blanks will always hurt your score more than guessing. We've known people to take the GRE, guess on the last five questions in the math and analytical sections and still score a 780 in both sections. The best way to guess is to pick one letter and stick with it.
But, don't think that an adaptive test is entirely bad. The amazing thing about a CAT is how much you can improve your score with good pacing and by working the system. You can improve your total score 200 to 300 points by learning how to effectively balance your time while not trying to answer too many questions. Of course, it takes strategies and lots of practice as well. As long as you have full knowledge and understanding of a test, you can learn how to take it effectively and achieve your personal best.
Before wrapping up, let's go back to our primary advice about "being careful." We've worked with thousands of students over the years. One thing we've learned is that weird things can happen to people when they take CATs. Maybe it is their finicky nature, but we recognize that sometimes extreme results can occur given their scoring schemes. Basically, if you do not get at least half of the questions right in the first third of the test, it is incredibly difficult to score above a 500. Be careful about getting stuck on an initial question or two and scoring much lower than what you had expect based on your prior practice.
On the up side, if this happens, you can return to take the test again the next month. If this is the case with you, it is important not to change your test-taking techniques nor do a ton more practice. Just go back and make sure to emphasize the first two thirds of the test. Believe us, we understand how frustrating it can be to have to go through the process again, but remember that a bad performance on a CAT is not uncommon because of the extreme nature of the scoring system. So, don't give up! The tortoise's victory over the hare can be your inspiration.
We wish you all the best of luck in your CAT experience. Remember, a standardized test is only a narrow-minded and tricky stepping stone to the more important experience, graduate studies. Once you are finished thinking like ETS for the purposes of their test, then you can return to your own original and creative approaches to life and apply them to school and your future career. [clap]