Note that spending that extra time on one question prevented you from getting to 3 questions at the end. Was that a good trade-off? Obviously not!

You basically have to convince yourself that (a) on average, going way over does not improve your chances of getting the question right, and (b) going way over does improve your chances of running out of time at the end (which leads to more wrong answers).

Do this by going and looking at the data from your last few CATs. How many times did you go over 3 min? How many did you get right vs. wrong? How many sub-1-min questions did you get wrong on which you either had to rush or thought you got it right but made a careless mistake? Do the analysis to show yourself just how much this is bringing you down.

By the way - on ones on which you randomly guess, you have to assume worst case scenario: that you'd get them all wrong on test day. Don't want to trust to luck here!

Then, go look at the characteristics of the ones on which you tend to spend too much time. PS vs DS? Different content areas? Find patterns. Next time you see problems that fit those patterns (eg, wordy rate problems that use algebra instead of real numbers), be extra conscious about the time.

Also, make sure this is your mindset: I'm going to have to guess at some point on this test - I don't get a choice about that. My only choice is: do I want to be in charge of selecting the specific problems on which I guess, or do I want to be forced to guess on x questions in a row at the end?

There's only one good answer to that question: you want to be in charge. You want to select the hardest questions as you see them throughout the section. This way, you're guessing on the hardest ones, not some mix of difficulty levels, and you're spreading those guesses out, rather than having them clustered together. (A string of wrong answers in a row is far more detrimental than the same number of wrong answers spread out, with correct answers interspersed.)

Most people have to guess on 5-7 questions per section. Note: this does not mean people have to guess randomly on that many. Some of those should be educated guesses - you actually take time to eliminate some wrong answers before you make a guess.

That's within the standard deviation of the test - nothing has changed about your inherent ability. You're not getting worse.

Remember that this test is not actually precise to within 10 points, although the scoring scale is in increments of 10 points. The SD on the official test is about 30 points and the SD on our tests is about 50 points.

How's your timing during that last bit? Are you still spending about the same time as you are earlier in the section, or are you working more quickly (for whatever reason: you're tired, you're low on time, etc)?

Any patterns within those? Do you tend to get certain types or sub-types wrong? (Maybe you're mostly falling down on those questions that are your weaknesses.) Or is it across the board, no real pattern? (In which case general brain fatigue might be a bigger factor.)

As you take future tests or sit for long study sessions, you can also experiment with things that can help to keep your energy up: what you eat and drink on the breaks, what physical movement you do on the breaks. You can even do things in the test center, such as stretching and flexing your legs, sitting up really straight in the chair, taking deep breaths - anything to get your body loose and energized.

As you continue to study, here are some things to try (including some of the things I mentioned in my last post):

- study as though it's the test: eg, if you sit for a 2-hour study session, take one 10-min break about halfway through; otherwise, work hard the entire time without stopping. Figure out what you want to do for the entire 2 hours before you start and make yourself stay on track / concentrate the entire time (except for your 10 min break)
- during the breaks, try eating and drinking different things and notice how your body and mind react (literally keep notes, track the data, and see what gives you energy and helps you to maintain focus)
- same as above for physical movement during the breaks (stretching, flexing, touching your toes, walking around, etc.)
- same as above for mental stimulation during the breaks. Should you: read a magazine or do something totally unrelated to the test? give yourself a mental pep-talk? review notes related to the upcoming session? Different things will work for different people.
- Also do things while you're sitting there studying, such as stretching and flexing your legs, sitting up really straight in the chair, taking deep breaths - anything to get your body loose and energized. Remember that you can't leave the chair and you can't make audible noise (pretend it's the test center!).

Oh, and I forgot to add for others who want to take tests on which they might see repeated questions:

You can still take both GMATPrep and MGMAT CATs as long as you follow a few guidelines to minimize the chance of artificially inflating your score via question repeats. First, anytime you see a problem that you remember (and this means: I know the answer or I'm pretty sure I remember the answer, not just "hmm, this looks sort of familiar..."), immediately look at the timer and make yourself sit there for the full length of time for that question type. This way, you don't artificially give yourself more time than you should have. Second, think about whether you got this problem right the last time. If you did, get it right again this time. If you didn't, get it wrong again. If you *completely honestly* think that you would get it right this time around if it were a new question (even though you got it wrong last time) because you've studied that area and improved, then get it right this time.

You don't actually know what you would need to know in order to make the comparison. You don't know the difficulty levels of the ones you got right vs. the ones you got wrong. You don't know which were "experimental." I put that in quotes because they aren't technically testing the questions via GMATPrep - they're already retired - but the algorithm is the same as that on the official test. Only about 27 in quant and about 31 in verbal actually count.

It's just not possible to look at # right/wrong (and even placement of # right/wrong) and come to good conclusions about the score - there's so much more going on than that.