MGMAT CAT Questions

[beatthegmat]

Moved to Manhattan GMAT section of this forum. I'm hoping that Stacey might have some comments to enlighten you...

[Stacey Koprince]

The test is adaptive and is built on the same algorithm as the real test. The fact that you noticed such differences actually indicates that it IS adaptive, not that it isn't.

Difficulty level is only one factor that determines what question you are given next. The algorithm also has to factor in:

(1) question type (so that you get an appropriate number of DS vs PS questions)

(2) content area (so that you get an appropriate number of, say, geometry vs. algebra questions and, within geometry, an appropriate spread between the different sub-categories of geometry)

(3) questions you've already had (you can't get them again)

The real test factors all of those things in, too. So if your next question is supposed to be at the 710 level according to the algorithm, but the test also needs to give you a circles-based geometry DS question, the test might select a 680 problem or a 720 problem or something like that - it doesn't just go strictly by difficulty level.

Because you have been scoring at such a high level, you likely "used up" many of the 700-800 questions by the time you got to test 6; that test might then have had to go a little lower to make sure that items 1 and 2 (above) were covered properly. Also, FYI, though you only see difficulty levels based on 100-point increments, the questions are actually rated by 10-point increments. So a 600-700 level question might actually be a 690 question which, for purposes of the algorithm, is not significantly different from a 700 or 710 question.

Also, one difference between our test and the real test is that we don't have experimental questions built in. If you are a very high scoring test taker in math, the real test can feel easier at times b/c you will suddenly get a question that is much easier than your ability level. That's part of the reason why people feel our math test is harder than the real thing - but that sentiment is more common among those who score well on math. Those who have more average scores don't notice this difference as much.

(Though I agree with you that some of our problems are too computation-intensive. We are working on weeding those out as we identify them.)

Last thing to remember. The official test reduces your score by about 3 percentile points for every question you leave blank at the end. When you say you ran out of time, I hope you guessed randomly - just make sure you don't leave anything blank. (Ideally, though, make sure that you move steadily through the test and don't have to guess randomly at the end - a string of wrong answers in a row is the worst thing you can do for your score.)