GMAT Strategy

Save GMATPrep - don't take one till after you've done some good study (including taking some other tests).

You don't necessarily need to do every last OG question again, but you will probably do a lot of them. Figure out what to do based upon the strengths and weaknesses you identify from your practice tests.

Our company lets you take one free test and I'm sure the other companies do as well. For our company, you can also buy access just to the tests, or you can get access to the tests for a year if you buy one of our strategy guides - see our web site for details. Check out what your options are from other companies and ask around the forums to see what tests people like best. (Obviously, I like ours... but I'm biased )

Don't do old paper tests - paper tests in general are nothing like the real thing; the CAT is a completely different beast. PowerPrep was not truly adaptive, so those also aren't great. I would go with either ours or TPR.

I have had students tell me they think our tests are harder than the real thing (and, as such, that our tests really prepared them well for the pressures of the real test)... but I have also had just as many students tell me that the real test felt harder. So take that however you will - basically, people feel both ways; it's tough to generalize from what just a few people say, and your individual experience may not match what some other person experiences.

I will say that I think a small number of our math problems are too computation-intensive for a 2-min-per-question test. But that also teaches you to move on when necessary, and you will definitely have to do that on the real test, so it's good to get the practice ahead of time or you will spend too much time on problems you really can't do (and then get other questions wrong, too, as a result of not having enough time).

TPR=the princeton review. I don't know how 800tests are, unfortunately.

You are bringing your score down a HUGE amount by running out of time. All of the questions on the test are worth the same amount - so it's the same as getting the first 10-12 wrong (it's a myth that the ones at the end aren't worth as much). If you get, say, the last 10 questions wrong, that will result in a decrease of about 20 to 25 percentile points from whatever you were scoring just before you got those all wrong.

Whenever the test puts a question in front of you, the issue is not just whether you can answer that question, but whether you can answer it IN 2 MINUTES. If you cannot, then even if you get it right, it's costing you another question (or more than 1!) later in the test - so, at best, you're at 50/50 when you spend too much time on a question. That assumes you actually got that question right, though, which is less likely because the reason you're spending extra time is probably because it's a harder question!

What does this mean? First, it means you're going to get a lot of questions wrong, no matter what, so don't worry so much about the percentage you get right - that's not how the test is scored. Worry more about spreading those wrong answers out - the penalties are more severe when you have strings of questions wrong in a row.

Second, you need to move through the test at a steady pace so that you can give yourself a shot on every question. If you can't get something in 2 minutes (1.5 for SC), then pick something and keep going - even though, yes, it means you're more likely to get this question wrong! You'll still have a shot at the later questions though.

Do a little exercise, if you can. If the practice tests you're taking record the amount of time spent on each question, add up how much time you spent getting questions right and how much time you spent getting questions wrong. Most people are amazed at how much time they spend on questions they get wrong - and, lots of times, they knew they didn't know how to do at least some of those questions but they hung on anyway and wasted time. (If you took our tests, you can get the timing data from the problem list that you can access after the test is over.)

In order:

II: you can see OG questions when you take GMATPrep - you just have to hope you don't see too many that you remember. If you see a ton, then that particular test will likely have an inflated score. You may not actually remember all of the ones you've done before though - it has to be something where you remember enough to either get the question right when you otherwise would've gotten it wrong OR you spend far less time on the question to get it right than you otherwise would've spent.

nick_mask and xeqtr: generally, you should assume a 2-min per question average in OG (though 1.5 min for SC), though you should also do at least half of the questions in "blocks" - that is, 10 questions in 20 minutes, where the questions are spread out across topic area and question number (as the numbers get higher in OG, the problems get harder). That last is NOT true, unfortunately, for RC. The questions are not in order of difficulty even within one passage - instead, they are in chronological order of the info presented in the passage. Generally, spend about 2-3 min reading the passage - enough to understand the main information presented and the structural outline of the passage - but not so much that you learn all of the detail. Remember that, although in OG you see ALL of the questions written for the passage, on the test you'll be given only half of the questions written for that passage. No sense in learning a bunch of detail you won't need to know, since you won't actually have to answer all of the questions! Then, take about 1 min for any general questions (the author's primary purpose is...) and 2 min for specific questions.

Also, nick_mask, you've already described very nice improvement in RC. In order to answer the question "how do I improve?" though, I'd basically have to be your private tutor and know precisely where you're doing well vs. where you're struggling, etc. There are a lot of things you could do to improve depending on your individual strengths and weaknesses.

vgmat2: I knew GMAC was going to release the new online test but I didn't know it had finally launched! (The initially slated the math test for release in January.) I haven't had a chance to try the math test yet but it sounded very intriguing when I heard it described last October by some GMAC folks at a conference last year. If it does what they said it would do last fall, it should be a really useful tool. They didn't announce a release date for a verbal test - they said math should be out in January and verbal would be sometime later in the year. So, it's probably going to be a little while before the verbal is out.

Approximately 10 questions per section are experimental - and, no, there's no chance you won't get any at all. Everyone gets experimentals (quite a few, actually).