Neither of the above opinions you have referenced is exactly right. Yes, it's true that every test will have some wrong answers and that the difficulty level is far more important than the number of right answers -- however, half is a little high. Most practice tests I've seen that resulted in 700+ scores had around 20 total wrong answers combined.

However, the idea that there is some magic number that you must hit, in terms of right answers, to get a 700 is not backed up by data OR by how the algorithm works. The algorithm gives you questions based on its estimate of your ability level. Sometimes -- often, in fact -- it gives you questions that it thinks are above your ability level. It expects you to get them wrong. If you get them wrong, it doesn't hurt you at all! It just limits you in the sense that the algorithm gets more confident that its estimate is correct. Don't worry, though, you will get another chance at a harder question later on, a chance to prove yourself better than the algorithm initially thought.

It's very important not to play these games while taking the test, of course; you try to answer every question that the test gives you and you try not to figure out if it's easy/medium/hard (because this determination itself takes in a panoply of factors and requires a very advanced computer program, your intuitive sense of the question's difficulty is, frankly, misleading and irrelevant, so don't bother making one, not even in practice).

In terms of the question about a correlation between your "raw score" and your scaled score -- there is none. None at all. In fact, there's no such thing as a "raw score" on the GMAT the way there is on the LSAT (that is, the number of questions you get right in each section). Part of this is the adaptivity issue, in that the difficulty of the questions matters, but a big part of it is the existence of experimental questions. There are 6-8 experimental questions per section on the GMAT, and sometimes even more than that on paid test-prep companies' tests (since they are trying out new questions on a smaller sample of test-takers). So, if you get 12 questions wrong in a section but 2 of them were experimental, that's very different than getting 12 questions wrong while 8 of them were experimental. You can't know anymore on those tests than on the GMAT, so again, there's no point in trying.

To improve, you need to get more questions right. Period. Perhaps the "raw score" of total number right won't go up, but some of the questions you got wrong on your last CAT, if you'd gotten them right, you would have scored higher. It's always as simple as that. The less we worry about the algorithm and how to trick it, and the more we focus on how to improve our math and verbal skills to the point where we're getting more questions right faster, the better we'll do. For all this complexity ... it's as simple as that.

Thanks for the post. Well I have been taking the MGMAT CAt exams and even on sections that I am doing really well in I am getting between 40% to 50% incorrect. Is this a normal phenomenon?

GHong14 wrote:Thanks for the post. Well I have been taking the MGMAT CAt exams and even on sections that I am doing really well in I am getting between 40% to 50% incorrect. Is this a normal phenomenon?

Normal? Yes. But you're not going to get a 700 while still getting 15+ wrong answers on each section. It means you still have some concept-building and strategy focus to do before you're fully ready to ace the GMAT.