First 10 Questions on the GMAT?

Are you assuming that the test taker answered those first 7-8 questions right? Because I don't understand why else in a CAT 8-10 would be harder if it's based on the previous questions answered.

Thanks bb.

My only question is this:

On my previous two Kaplan CATs I had the same exact number of Quant questions right. I got 16 out of 22 in problem solving and 11 out of 15 in data sufficiency. However, I got the first 11 in a row right for the one I took a test ago but got 7 out of the first 10 right only, on my last test. I ended up getting 88th percentile a test ago, but only 75th percentile last test. Now if the GMAT would give me a similar drop in percentile, wouldn't it stand to reason that the first 10 questions being the most important is in fact NOT a myth?

That's sort of circular logic.

You're using Kaplan practice tests to validate a Kaplan strategy.

Some people here will tell you to use the "focus on early problems" strategy. Others will tell you stick to the ~2 mins/problem on quant and 1.75 mins/problem on verbal.

Check bsd_lover's thread about his analysis of GMATPrep (bb linked it above), the theory doesn't hold much water there.

Yeah that's why I said "if the GMAT would give a similar drop in percentile". But honestly though it seems to make sense doesn't it? If this is a computer adaptive test then my questions would most likely be easier on average on the test that I got the first 7 out of 10 wrong. No? Otherwise what's the point of a computer adaptive test?

If you read the debriefs where people struggled with the real test, you'll find the most common issue is running out of time and struggling to finish. If you give more time to the questions up front, you lose time at the end. With 2/mins or 1.75/mins per question, you're stretched as it is to allocate time around.

The point of the computer adaptive test is it makes it easier to score people on the spot. You don't have to wait for scores to come in to adjust accordingly and ensure your normal distribution is maintained.

If you like the strategy, by all means go for it. It seems you want us to validate your opinion as opposed to take out response as they stand. I've been very clear I am not a fan of that strategy (I tried it on the GMAT on my first try).

I don't use the strategy of spending more time (at least not significantly). I just want to know want to get somewhat of a gauge on the test. Not looking to be validated. In fact if the first 10 questions are overrated that would relief some of the stress I feel when I don't know the answer to a question somewhat early on.

The first ten questions are them "most important" in that they have the largest impact on your score of any questions on the test. However, just because they are the most important doesn't mean you can afford to spend extra time on them. In fact, doing so is actively counterproductive--not only will you have less time for the remaining 27 or 31 questions, but those questions will be harder because you spent that extra time.

Keep practicing, though. Remember, you don't hit the really advanced questions until later on in the test. Once you've mastered the GMAT to the point where you can nail the first 10 questions both reliably and quickly, you'll start to see imporvement!

I am thankful that when I was taking my GMAT I did not know all the possible strategies and ways to "trick" the test.

I strongly believe that GMAT has a way of correcting itself and mitigating any strategy you will use to try to trick it.
For example, if you spend too much time on the first section, it dings you for guessing and leaving questions incomplete. I am pretty sure that if you spend less than 5 or 10 seconds on a question, it won't get counted as a truly correct one. I think it is best not to speculate too much as we'll never have a proof of real gmat logic.

A lot of people make mistakes in the first 10, many make multiple ones and do fine due to experimental questions or simply because it was a "glitch". Many believe that gmat has ability to detect such "glitches" - meaning that if you missed the question it expected you to know, it won't ding you right away and vice versa. (this is speculation of course and I have no way of backing it up). I do think that you need a general "reserve of time" for a question that you will get stuck on because you will and if you don't plan for it, you will panic. I would recommend you practice that way. allow yourself to spend 3-4 mins on one question but only one or two. That was my strategy - I gave myself a few minutes more for the first third of the test. I knew that I would need that time. However, at the end I ended up spending 4 mins on the question and answering it wrong, so all of that time was wasted - lesson for you! Out of 37 math, I planned to spend 30 mins on the first 13 questions, and 23 on the next 12, and then another 23 on the last 12.

I was favoring the first 13 questions but not by much. However, know how much time you can afford to spend. Use these check points to guide you.

Yeah, thats true, I'm assuming that the test-taker has answered most of these first 7-8 questions correctly. Btw, Kaplan CAT tests are far different from actual tests / GMATPrep tests.

I call the first 7-8 questions as "time savers" questions coz these are relatively easier. Double checking the answers would mean time waste. Instead, after 8-9 questions GMAT starts shooting difficult questions and one needs to have extra time to spend on these questions (e.g. in quant sometime one may have to spend 3-4 mins per question on the difficult ones).

I'll call these difficult questions as "score boosters". This is my personal experience (don't know how true it is), nailing down few of these difficult questions would greatly improve the score.

So my suggestion is not to waste time on double checking the answers (contrary to Kaplan books suggestion of spending extra time on first few questions), save this time to hit the difficult ones. Again towards the end one starts getting easier questions, one can solve without spending much time. I'll call these "panic reliever"

Honestly, it's really impossible to "know" how you're doing.

One my first attempt at the GMAT, I *felt* good before I clicked to show my score and not cancel, end result was 610.

I felt TERRIBLE on my 2nd attempt. Sleep deprived, upset stomach, finished with less than 30 seconds on both sections and continuously thought to myself "there will NOT be a third time", clicked to show my score and not cancel, 710.

No matter what, the early non-experimental problems are always easier. When you get problems right (assuming non-experimental), you get a more difficult problem. Even if you mess up on a few problems, as long as you don't have a streak of crashing and burning on problems, you'll be fine.

The other problem is you have no way of differentiating between experimental problems and real ones. If you spend a lot of time on a problem and it is an experimental one, you've just done two things to hurt you:

Spent a lot of time on one problem.
Right or wrong, experimental problems don't matter and you'll never know either way.

The early problems are not "worth more" in that regards. Hitting a bad patch is fine AS long as you focus and buckle down on the following problems

By reading the Fast Play on Kaplan online, the guy told not to spend too much time on the first 10 questions. Even if you answered them correctly but used most of the time, you might end up with no time for the rest of the test and keep getting incorrect answers or missing answers. No matter how good you score on your first 10 questions, you might end up a score below your initial score. Remember that the GMAT checks how many difficult questions you answer correctly.

So experimental problems are always of the difficult variety?

Experimental questions are questions to be included in future tests, so they are of various difficulties. The computer just throws it to you to see if you get it right or wrong and based on that info, it determines its difficulty level. Does that make sense?

Absolutely. That's what I thought it was, but I misunderstood from the previous post and thought that they were only of the difficult variety.

A few comments on posts above:

-*never* draw conclusions about the GMAT scoring algorithm from a test prep company test. No prep company uses anything remotely similar to the real algorithm;

-the first ten questions are not weighted more heavily in the calculation of your score than later questions. Getting a 500-level question wrong early in the test hurts you as much as getting a 500-level question wrong at the end of the test; the scoring algorithm weights all questions equally. The myth that the first ten questions somehow determine your score, and that the remaining questions are only used for 'finetuning', is based on a profound misunderstanding of how an IRT-based test works. That said, of course it's a great idea to get your first ten questions right, but it's not a great idea to spend 60 minutes doing it;

-Early questions are significant only in that they help to determine which questions you see later - the test is adaptive, after all - and if you want a high score, you need to see some hard questions and answer some of them correctly. To do that, you're going to need to get some questions right at some point, and you can't wait until the end of the test to do that. Still, you can easily recover from an early mistake;

-The GMAT does not have a 'glitch detector', but the algorithm does acknowledge that a 700-level test taker will occasionally get a 500-level question wrong, for example, and that a 300-level test taker will occasionally get a 700-level question right (actually, the probability of guessing correctly is explicitly built into the algorithm). The algorithm is based on probabilities; for each question, the algorithm knows how likely it is a 200-level test taker will answer correctly, a 300-level test taker, and so on (the internal algorithm doesn't use the 200-800 scale, but I'll do so for illustration). After you've finished the test, the algorithm looks back over your performance on each question, and based on the probabilities attached to each question, it determines what your level most likely is. To give a very simplified example, suppose the test consisted only of medium-level questions, all of which were known to be answered correctly 25% of the time by 200-level test takers, 60% of the time by 500-level test takers, 75% of the time by 600-level test takers, and 90% of the time by 700-level test takers. If you answered 65% of these questions correctly, the test would give you a score around 530. Of course, the test is adaptive, so you wouldn't have so many questions with identical properties; adaptivity simply helps to make the test more efficient, and more accurate at the extreme levels (the test would need to be twice as long if it were not adaptive);

-I am 99% sure there is no guessing penalty on the GMAT (i.e. a penalty for answering a question too quickly). It's certainly possible to answer some of the conceptual GMAT Quant questions in a few seconds; I've done it on the real test and haven't been penalized for it. The penalty for guessing is that you have a 1 in 5 chance of answering correctly;

-When you sit down for your test (before you've answered a single question), you are assigned at random a pool of experimental questions, and the test selects in advance, again at random, the question numbers which will be experimental on your test. Experimental questions are organized in batches; I'm sure they do this in order to approximately control the difficulty level of these questions (and topic area), so that experimental questions are fair for each test taker. Experimental questions will range from very easy to very hard. If you are well above-average in Quant, you are likely to find most of your experimental questions to be quite easy for you, though of course you won't know which questions they are - it's best to treat every question as though it counts.

I took a practice test where I got the first 2 Verbal questions wrong and missed numbers 10 and 12 also. Then I was flawless for the last 29 questions. That is only missing 4 on verbal!!! Should have been decent. I scored a 39. Took another test where I got first 12 right and started getting crazy bold questions. Missed 9 and got a 49. If you are shooting for 700+, I would say that you should make sure to get those first 10 correct or it will be an uphill climbed back to the tough questions.

I still do not understand why the first 10 questions must matter more than the rest. I agree with everyone here who says that you will end up hurting yourself more (through wasted time) than helping yourself by spending all of your time on those.

The first few questions will cause your score (if there was such a thing as a score after a few questions) to swing widly because the test has no data on you yet. So, every question will have a big impact on your score. But, if you get the first question wrong, it is not like the test knows for 100% certain that you are in the bottom 50 percent. What if you just made a dumb mistake, and your true level is 760+? After answering enough questions, that should emerge.

By the end of the test, a CAT should still be able to determine what your real level is, regardless of how you did on the first few questions.