GMAT Scoring Algorithm

In my last post I described how Data Sufficiency is, for many, the key to a high Quant score on the GMAT. I decided to continue this theme to discuss a related but, I think, more important point and one that I love to explain to people because it seems that so few people understand this fact: the key to a really high GMAT score is a high Verbal score.

On a general level, most people just tend to neglect the Verbal section of the exam. This is understandable: on the Quant section there is probably more to “learn” for most people, and this requires more time. But another factor is that for most people it is just more enjoyable to practice Quant. Verbal is like the redheaded stepchild of the family. Kicked to the curb, neglected, under-appreciated…you get the point. Oh, and Reading Comprehension? Forget about it. People NEVER practice reading comprehension. I mean, who wants to voluntarily read about a group of Pakistani poets or a new finding about globular star clusters (I hope I am not offending any fans of Pakistani poetry or would-be astronomers out there…I am actually a closet astronomy nerd so I genuinely LIKE those passages!).

GMATers just tend to ignore or neglect the Verbal part of the exam, to their own detriment. But people! The Verbal section makes up half of your score!!! More importantly, most people don’t realize that at the top end of the scoring spectrum, the Verbal score actually weighs in more heavily that the Quantitative score!!! Over the years I have tutored lots and lots of people who have Quant scores of 46-48 and Verbal scores of 35 or so and who want to focus almost exclusively on bringing their Quant score up. Nooooo!!!!!!!

Here’s the deal. Most people will not get above 48 or 49 Quant. Those are great Quant scores (you can’t just look at the percentiles, which are a little bit skewed by international applicants, many of whom do have Quant scores of 50 or in some cases 51). So for most people, it’s good to almost think of 49 as the ceiling for Quant. For reasons I won’t expound upon here, there is not that much of a difference between a 47 and 48 Quant or between a 48 and 49 Quant (in terms of how difficult it is to achieve those scores), but there IS a pretty big barrier between 49 and 50 and an even bigger barrier between 50 and 51, a barrier that most people will not cross. So if you are already at 47 Quant, you are probably near the maximum of where your Quant score will ultimately go. And bringing the Quant score from a 47 to a 49 will not have that big an impact on your overall score.

If you are at a score of, say, 47 Quant and 35 Verbal, there is much, much more room for growth on the Verbal side. And bringing your Verbal score to 40+ or, if possible, to something like 43 or 44 Verbal will do absolute wonders for your overall score. I recently tutored a girl who on her first actual GMAT got a 700 with a 45 Quant and a 41 Verbal. Now, first of all, that is a great score. She wanted higher, however, and was a little disappointed in the Quant score in particular. On her second try she got a 46 Quant and 46 Verbal for a 740!!! Now, this girl is probably capable of a Quant score of 48 or 49, but even if she had gotten a 48 Quant and combined that with her original 41 Verbal, her score would only have been about 720. I mean “only” probably isn’t the right word here, but relative to the 740 that she actually got, obviously 720 is not as impressive.

Part of the issue here is that the percentiles on Quant and Verbal can be misleading. A 49 Quant is 77th percentile whereas a 40 Verbal is 91st percentile!!! So it is really tempting to look at the Quant percentile and feel like you just aren’t doing that well. But at 49 Quant most people are probably at their ceiling! And getting a 50 Quant, while impressive, is not going to do much, if anything, to the overall score. Conversely, it is tempting to look at a Verbal score of 35 (76th percentile) and believe that you have done really, really well on Verbal. Don’t get me wrong, 35 Verbal is a good score, but that percentile is misleading for 2 related reasons. One, it is skewed by the fact that so many non-native English speakers take the GMAT. Two, it hides the fact that there is a ton of room to bring that score higher and have a huge (and I mean huge) impact on your overall score!

Again, at 40 Verbal (which really is an excellent Verbal score) it would be tempting to think that you have kind of topped out since you would be sitting at 91st percentile. Think about it from the perspective of the aforementioned girl with the 45 Quant, 41 Verbal, 700 overall. Her percentiles were 59th percentile Quant and 94th percentile Verbal!!! I almost can’t believe it now that I am writing it – those percentiles are pretty shocking. It would have been very tempting for her to completely ignore Verbal and just focus on Quant. But man, 46 Quant and 46 Verbal for a 740 is amazing and it never would have happened if she had just focused on Quant.

Now some of you out there may be thinking, yeah, but at 46 my Quant score just would not be high enough – schools want to see a higher Quant score than that. I have two things to say to that. First of all 740 is 740. That’s just a ridiculous score. Even if they want to see something more like 47 or 48+, it’s very hard for me to imagine they are going to turn away someone with a 740 on the GMAT, assuming they like everything else about the applicant. Second, and here’s the more important point: if you want a really high GMAT score (like 740 or 750), you almost certainly MUST HAVE a really high Verbal score.

For example, a 50 Quant and 38 Verbal would yield about 720. And a 49 Quant and 41 Verbal would yield about 730. And the fact of the matter is that of all the people I have tutored over the years (many hundreds so far), many more have had Verbal scores of 44 or higher than scores of 50 or 51 Quant. And of the people I have tutored who have had scores of 740 plus, more often they have had scores like 47Q/44V or 46Q/46V or even 48Q/44V than scores like 50Q/40V or 51Q/39V (both of which would “only” yield about 740, by the way!).

More and more lately I have had people come to me with the goal of 750 on the GMAT. I used to chuckle a little at that, not because I thought people were not capable of it but because that just seemed like an unnecessarily high score to strive for. Nevertheless, recently I have had many people come to me with that expressed goal. They often think that they need more help with Quant than Verbal (again usually these people have something like 47 Quant and 36 Verbal – something in that range – and are duped by the percentiles). And usually I explain to them that we can and should focus on Quant to try to get them to 48 or 49, but that we NEED to focus more on Verbal if they want to have any shot of getting close to 750.

So as you plan you overall GMAT study strategy and think about how you are going to allocate your study time, remember that at the higher reaches of the scoring scale the Verbal score becomes increasingly important. And more generally, be mindful of how misleading those percentiles can be – its better to focus on the actual scores for the Quantitative and Verbal sections. Scores like 47 Quant and 38 Verbal are both excellent scores. But if that is the range you are in or even close to, understand that there is MUCH more room for growth on the Verbal side and that bringing the Verbal score up is going to be necessary to see a big increase in your overall GMAT score.

Back in the old days the GMATPrep tests did not break down your performance by question type on the Quant and Verbal sections. Nor was there an Enhanced Score Report that did so on the actual GMAT. Now, with the existence of practice tests that indicate your performance on each individual question type and the Enhanced Score Report that does the same for the actual test, it is possible to understand how your overall score is arrived at and what your score is for each specific question type.

With that, I have noticed a trend with my students that I could not have known in the past and that probably has relevance for many GMATers out there: they tend to score higher on Data Sufficiency than on Problem Solving. Now, I don’t mean to imply that everyone scores higher on DS than PS – people who don’t really understand Data Sufficiency tend to underperform there – predictably. But what I would say and what I will explain below is this: for people who are not naturally great at Math, there is an opportunity to really excel at Data Sufficiency and have that score pull up the entire Quantitative score.

First, let me explain a little bit about the scoring. When people take practice tests on the official software or when the look at their Enhanced Score Report (ESR), what they tend to look at are the percentiles. This is especially true on the ESR since the percentiles are what appear on the center of the page, but if you look at the bottom of each page in the “Summary” section you will see hidden down there a breakdown of what your score was for each question type, and this information is what is most telling. Your overall score tends to be roughly the average of your scores for each question type. So if you scored 40 on Problem Solving and 48 on Data Sufficiency, your Quant score will probably be 44.

What I started to notice a while back in reviewing practice tests and ESRs with my students is that they usually do better on DS, sometimes overwhelmingly so. Now, they don’t necessarily come to me in that state – often it is the opposite on the early practice tests that they take and DS is much worse. But as we start to approach their test date and they are ramping up and taking a lot of practice tests, most of my students just score higher on DS. A very common scenario for me is to see something like 49DS/42PS for a Quant score of 45 or 46. And at the end of the day they may end up at 49DS/45PS for a 47 Quant.

Here is what is significant about that. Many of these students are not great at Math and are probably not capable of getting really hard Problem Solving questions right on a consistent basis. But the beauty of Data Sufficiency is that it is much more about logic and reasoning than it is about Math. So if you really understand the question type well you can dominate on DS without really being master of the Math that underlies the questions. And what I have found with my students is that it is much easier to get them to a place where they are killing it on DS than on PS, in part because PS questions, especially at the higher levels, really do require a little bit more Math know-how or even innate Math ability.

It is outside the scope of this post to really get into the specifics of how to dominate on Data Sufficiency (though many of my other posts expound upon some of those strategies). And obviously if you are aiming for a 49 Quant, you need to be equally dominant on PS questions. That said, I would argue that if you are not naturally good at Math, Data Sufficiency really presents an opportunity to far outperform your natural Math ability and end up with a score that puts you above those who are innately better at Math!

I tutor plenty of people who routinely get scores of around 40 on PS and in my opinion are unlikely to ever be hitting scores like 44 and above (they are just not that great at Math and have come a VERY long way to even get to a 40 on PS). Yet they will routinely get scores of 47 to 49 on DS. So they come away with scores of 44 or 45 on Quant. Again, for someone aiming for a 49 or 50 Quant, that might not seem that impressive, but for people starting at 30 Quant who are not very good at Math, a 44 or 45 Quant is pretty damn impressive! And for the people I tutor who are able to get to 44 or 45 or 46 on PS (again often people who are not great at Math), they will often end up with a 47 or even 48 Quant, in part because of the higher DS score.

So if you feel like you are the type of person who is just innately not that good at Math, consider that, in Data Sufficiency, you have an opportunity to really dial into the logic of the question type (along with its common structures, traps, etc.) in a way that would allow you to far exceed what you might expect from your pure Math ability.

People studying for the GMAT are often completely thrown by a correct answer on a sentence correction question that just “doesn’t sound good.” In their minds they think, “there is no way that can be the correct answer – that just sounds bad.” Even if that thinking is true that does not mean that the correct answer won’t sound just downright bad! Making right answers sound bad is a deliberate tactic on the part of the GMAT writers and in this piece I will explain why the do it, how I know they do it, and what you can do about it!

I often ask people, “What do you think is a defining characteristic of hard questions?” Most people often respond with some of the things that tend to happen on harder questions but the simple answer is that from the point of view of the GMAT writers the single defining characteristic common to all hard questions (Quant and Verbal) is that hard questions are ones that most people get wrong! If most people get a particular question right (even if the writers predicted that most would get it wrong), it is, by definition, not a hard question. However, because of the scoring algorithm and the way the GMAT is scored, there need to be very hard questions (in other words, questions that most people would get wrong). But how do the GMAT writers write a sentence correction question that most people will get wrong? Well, they obviously have a variety of tactics, but one of the most common and fundamental is that to make a question hard the correct answer can’t be obviously correct – if it were then most people would recognize it as correct and pick it. So the writers often deliberately write the correct answer on a sentence correction question to make it unfamiliar, awkward, even unpalatable.

Recently, while tutoring someone, I came across a sentence that really proved for me that this is a deliberate act on the part of the test writers. I pretty much knew it already, but I thought it was really interesting to see this deliberate tinkering in action. The sentence in question comes from the GMAT Prep exams. I had seen the sentence many times before, but this time I noticed that the correct answer was a little different from the one I had seen before – it was worse! So I looked into it and discovered that the sentence was in fact different in previous versions of the GMAT Prep software. After digging a little deeper I found what appears to be the original version of the sentence (from the New York Times). Low and behold we have pretty clear evidence of how the GMAT writers take a sentence, tweak it to make it less familiar and appealing, and then, when that isn’t enough, tweak it even further to make it borderline terrible (albeit correct), thereby increasing the level of difficulty even further.

So let’s start with what I believe to be the original sentence (again, from the New York Times):

“Also, unlike many other frogs that metamorphose from tadpoles to adults in one year, the high-elevation frogs take three to four years to reach adulthood, so they are restricted to deeper bodies of water that do not dry up in summer or freeze solid in winter.”

Ok, so standard comparison sentence and one that sounds pretty good in my opinion. Now let’s see how the GMAT writers changed the correct version of this sentence to make it less appealing and therefore harder:

Unlike frogs that metamorphose from tadpoles into adults within a one-year period, mountain yellow-legged frogs of the Sierra Nevada take three to four years to reach adulthood, and so they are restricted to deeper bodies of water that do not dry up in summer or freeze solid in winter.

Now, that sentence may not look that different to you, and it isn’t that different. But there are some subtle changes that make it more likely to be passed over by a test taker. The “mountain yellow-legged frogs” bit is just there because without it we don’t have the context to know what the sentence is talking about. But the less obvious and more important change is from the “so they are restricted” to the “and so they are restricted” version. Having done this question with many people I know that some people do not pick the correct answer because they believe that you can’t have two conjunctions (the “and” and the “so”) right next to each other. Indeed that usage is not that common and is therefore unfamiliar to many people, but it is acceptable and is not wrong. And since it is not wrong, it must be right!

To me that usage is not that unappealing, but what caught my eye with this sentence recently is how it was changed again to make it obviously less appealing and therefore more difficult. The most recent version on the GMAT Prep software changes the underlined part (in the correct answer) to:

“mountain yellow-legged frogs of the Sierra Nevada take three to four years to reach adulthood, with the result that they are”

Wow. I mean, maybe you readers don’t get as excited about this as I do, but my reaction is wow!!! That “with the result that they are” is pretty bad. That type of construction would typically be part of the wrong answer on most GMAT sentence correction questions. It is just not good! Here’s the thing. As unappealing as that construction is, it is not wrong! So I might initially disfavor that answer choice when I see it but I would not outright eliminate it. And on this particular question it is the right answer because the other answer choices just have other things that are definitely wrong with them.

Anyway, this is about as clear an indication as I have ever seen of how the GMAT writers try to ramp up the level of difficulty on Sentence Correction by purposely tweaking correct answers to make them unfamiliar, awkward, or even downright unappealing. So what can you do with this newfound knowledge?

The first thing is that you cannot eliminate answers because you think they sound “bad.” Wrong answers are wrong because there is something concrete wrong with them (either grammatically or on more of a logic/meaning level). So you might want to “disfavor” an answer that sounds awkward or unfamiliar, but I would not outright eliminate it. Answers are not usually wrong just because they are awkward. If an answer is absolutely hopelessly awkward to the point that it is flat out incorrect English grammar than that may be enough but usually there will be other things wrong beyond that.

It may also help you to EXPECT that the correct answer may sound a little bit off. That will be especially true on harder questions. Again there is a difference between unfamiliar or a little awkward on the one hand and hopelessly awkward or completely unidiomatic on the other. As you practice GMAT Sentence Correction questions try to learn where that dividing line sits. And again, worse case scenario, just leave the answer in play and come back to it at the end. If 2 answers are exactly the same and one is much wordier and awkward then it will be wrong. But if one answer is a little awkward but more logical and the other answer sounds better but has a concrete grammatical error that you can identify or even a logic/meaning issue, the more awkward answer will be correct.

It is a pretty well-accepted fact that on the GMAT the early questions are slightly more important than the late ones, especially on the Quantitative section (everything I have seen in all my years as a GMAT tutor has confirmed this and there have been some good experiments done by others on the GMATPrep tests that seem to confirm the theory pretty conclusively…feel free to follow the link to read those articles). In this post I hope to help clear up some of the often misunderstood implications of this idea and help you understand how to manage your time given the slightly disproportionate importance of the early questions.

The first thing to clear up is that there is nothing particularly special about the number 10 that makes it the clear demarcation point. Whenever I think about this concept it makes me think of a conversation I had with my wife about getting pregnant. She was nearing 35 and remarked that she needed to get pregnant before she turned 35 (she was 33 or 34 at the time) because she read that you have a greater risk of having a baby with birth defects when you are over 35. But like the first 10 question concept on the GMAT, it is best to think of it on a spectrum. It’s not like you are not at risk when you are 34 and then all of a sudden you are at a huge risk when you are 35. You are at a slightly greater risk at 35 than at 34 but at 34 you are at a greater risk than at 33 and so on. Why do experts use the number 35 then? The same reason that people say the first 10 questions on the GMAT – it’s a nice round number. So again, the best way to think about it is as a spectrum. The first question is probably a little more important than the second one and the second a little more important than the third and so on. Once you get to AROUND question 10 it probably ceases to be as important, but it is a spectrum – I would be a little more willing to go for it on question 7 than on question 9 and perhaps a little more willing on question 9 than question 11. But at around the question 10 mark it probably ceases to be important.

The second thing to clear up is that this is all a question of degree…again it is a spectrum and is not black and white. It is NOT true that you have to get the first 10 questions right to get a really high score (in fact most people who get really high scores do not get the first 10 right). And no single question really matters that much – not even question 1. I actually just tutored someone last week who got question 1 wrong and then got the next 12 or so right and then proceeded to get a lot more wrong throughout the rest of the test (probably about 13 wrong total) and got a 49. And I have tutored many, many people who have gotten 3 or 4 wrong in the first 10 and gotten 48 or 49. The early questions do seem to matter more and certainly if you dominate on the first 10 or 15 questions you are extremely likely to have a very high score, but again it is a matter of degree – those questions matter a little more and doing well on them will help, but it is not necessary to absolutely dominate on the early questions to get a high score.

Again the way I like to think of it is as a spectrum where the questions go from slightly greater importance to slightly less importance as you move from question 1 forward. People tend to overcompensate and spend WAY too much time on the early questions and then find themselves completely out of time later on the test and are then forced to guess on 10 or more questions. That generally will not lead to a high score. So what should you do? Well it depends a little on the level that you are at…

If you are not at a very high level on Quant then spending loads of extra time on the early questions is particularly ineffective. Here’s why: Assuming that you do get some of those really early questions right (like maybe 1 through 3 or 1 through 4) you are then going to be seeing very hard questions. If you are really not capable of answering those questions correctly then spending extra time on them is a foolish waste of time. There is no sense in getting yourself in front of 800 level questions and spending a lot of time on those questions if you are really at a 600 level. Even if you get some of those questions right you will probably be way behind on time and then you will eventually find yourself getting 600 level questions wrong later on in the test. At that point not only will you not get the equivalent of 800 on Quant but you probably won’t even get 600 Quant because the algorithm knows that even if you were able to get some 800 level questions right, anyone who is really at an 800 level would also get nearly all 600 level questions right.

If on the other hand you are at a pretty high level on Quant (let’s say you are scoring 40 or higher on Quant) then it probably does make sense to spend some extra time on those early questions. But only to a point!!! First of all, if the question is one that you just don’t think you can answer in a reasonable amount of time then it just isn’t worth it – guess and move on. Spending 5 minutes and then getting a question wrong is just about the worst thing you can possibly do on the GMAT. What I recommend is being A LITTLE more willing to spend time on those early questions and that extra willingness should progressively diminish as you approach the 10 question mark.

So for example, let’s say question 3 looks really hard but you think that you can probably answer it in 3 or 4 minutes. On question 3 I would probably go for it. If that same question appeared at question 23 and if I was already 5 minutes behind I would probably choose to guess and move on. That is how the particular placement of a question on the test may affect whether you try to answer it – at question 3 maybe you would go for it and at question 23 maybe you would not.

To take another example, however, if I was at question 7 and was thrown a really difficult question that I wasn’t sure I could answer I might choose to just guess and move on. It would depend on how likely I thought I was to get the question right, how long I thought it would take, and how far ahead or behind I was at that point. If I was 50/50 and thought it was going to take 4 minutes I probably wouldn’t do it (at question 7). If that same calculus presented itself on question 1, however, I might go for it. I might even go for it at question 7 if by some chance I was ahead a little on time, but if I was already behind I probably wouldn’t go for it.

So what we are talking about here is a SLIGHT willingness to spend more time on the early questions. You should NOT feel as though you need to get those first 10 right – that probably won’t happen anyway and trying to make it happen is likely to screw up the entire section. And again this strategy of spending a little more time on the early questions makes more sense for someone who is at a higher level. In either case remember that it never makes sense to spend a lot of time on questions that you are unlikely to get right. Instead find a way to guess cleverly – even if it is question 1.

One more thing I would like to add to round out this discussion. The strategy of spending a little extra time on the early questions depends on counterbalancing that by spending less time on the questions that follow. In my opinion you should probably not spend more than about 5 extra minutes on those first 10 questions. So if you should be at the 55 minute mark by the end of question 10, I would not want to be any further than the 50 minute mark at that point. It’s not that hard to make back those 5 minutes across the course of the rest of the section, but again you have to be aware of the need to do so and plan accordingly. If you “strategically dump” a few questions scattered through the rest of the section (say, for example, questions 13, 17, 25, and 31 just for argument sake) you should be totally fine. And if you were able to do it on what you knew to be very hard questions, questions that you would have been unlikely to get right anyway, all the better!

All of this takes some practice. Some people are able to perfect it pretty easily. Others need to take 4 or 5 practice tests before they understand how to strike the right balance. So be patient and give yourself some time to implement it properly. The key is to be aware that it is a balancing act and that you can’t go too far in spending extra time on those early questions!

In this post I would like to expound upon a theory of mine that gets at the slight differences in the scoring algorithm on the Quant and Verbal sections of the GMAT. In a previous post I commented on the fact that although on the Quant section the number of questions you get right has very little to do with the score you receive, on the Verbal section the number of questions right matters much more. I don’t think there is any official explanation for this phenomenon, but I have long had a theory that I think makes pretty good logical sense. Understanding the nuances of the algorithm can help you learn what exactly you need to do to achieve a high score on each section.

Quantitative questions, by nature, are completely unambiguous. If x equals 10, then x can’t also equal 15 (well ok it can in a quadratic equation and in some other special situations, but you get the point). Verbal questions, on the other hand, are by nature a little less black and white than Math questions. Of course GMAC writes questions that are technically unambiguous, but the problem is that it is tough to make a Verbal question really, really hard without crossing the line a little into ambiguity, so there tends to be a limit to how hard a Verbal question can realistically get. That problem does not exist on Quant. There is almost no limit to how hard the test writers can make a Math question and it will still be completely unambiguous, completely black and white.

This helps explain the differences in the algorithm on each section. On Quant, the questions can pretty much just keep getting harder and harder until nearly everyone would get the questions wrong. Really only at the very top reaches of the Quant section (say scores of 49-51) will people get less than 10 questions wrong. So for most people the differentiating factor in scores on the Quant section is just the level of difficulty of the questions that test takers are able to get right consistently. If your breaking point is 650, your score is 650 (or whatever that would correspond to on the 6-51 scale that the section is scored on). If your breaking point is 720, then your score is 720. The number of questions that you get wrong is not really a factor except, perhaps, in differentiating between people who get a 49 and 50 and 51.

But since there is more of a limit to how hard a Verbal question can get (again before it starts to drift over the line into “ambiguous/unfair question territory”), the number of questions right and wrong starts to become more of a differentiating factor. That’s because many test takers get up into the very hardest Verbal questions so at that point the main way for the algorithm to differentiate between a 38 Verbal and 40 Verbal or 42 Verbal and 45 Verbal is just by number of questions wrong. This is a bit of an oversimplification since the level of difficulty of the questions may vary a little so that would still be a factor in the score, but at the higher reaches of the Verbal scoring spectrum the score really becomes much more about number of questions wrong. And to be at a really high Verbal score (like 40+) you really can’t have many questions wrong. For a 40 you might have 8 or 10 wrong, but for a 44 or 45 you are probably talking about 5 or 6 wrong or even less and then for 47 Verbal or 49 Verbal you are looking at 1 or 2 wrong total!

There are a few implications to be drawn from this. First of all, since you are likely to get a lot of questions wrong on the Quant section, it really does not make sense to spend undue time on questions that you are likely to get wrong anyway. Furthermore, running out of time at the end of the Quant section is not that big a deal (assuming that you actually select answers to all of the questions to not incur the unanswered questions penalty). Again, you are going to get a lot of questions wrong anyway so coming to question 34 with 2 minutes left and having to make some educated guesses on the last 3 or 4 questions will not impact your score that significantly.

On the Verbal section, on the other hand, you need to be much more careful about time management and not put yourself in a position of having to guess on the last 3 or 4 questions. That will impact your score more detrimentally, especially if you were tracking toward a really high score. If you had only 5 questions wrong up until question 35 and then had to guess on 36 through 41 and got 5 of those last 6 wrong for a total of 10 wrong on the Verbal section, that would be a big deal and would lower your score significantly.

Another upshot is that you need to be a little more certain of your answers on Verbal questions than on Quant questions. Most people probably feel like the opposite is what often happens since the nature of Verbal questions is that they feel ambiguous, but of course they are not and you need to be that much more rigorous in the way you choose your answers to ensure that you limit the educated guessing to a bare minimum. You just can’t afford to get Verbal questions wrong with the same frequency that you can Quant questions. What that means, practically speaking, is that you can’t run out of time on the Verbal section and that you need to choose answers because you KNOW they are right, not because you THINK they are right. That is obviously easier said then done, but see my post on The Principle of No Ambiguity for some more insight on that idea.

Knowledge is power and on the GMAT having a better understanding of the scoring algorithm is the kind of knowledge that helps lead to a better score. In a future post I am going to come back to the issue of pacing strategies on the Quant section so stay tuned!

In a couple of previous posts I discussed some of the main reasons that people don’t see much of a score increase when preparing for the GMAT. In this post I will turn to yet another common cause of artificially low GMAT scores: poor time management!

Over the years I have tutored a number of students whose lower than desired GMAT scores have been almost entirely the result of poor time management. One girl in particular stands out. She came to me with a 560 score on the official test, but after I had my first session with her it was clear to me that she should be in the 700 range. Unfortunately she had no idea how to properly manage her time on the test, particularly on the Quant section, and the result was a score that was really well below her potential. I tutored her for about 6 weeks focusing almost entirely on time management and she managed to score a 710, which is probably the score she should have had in the first place had she known how to manage her time better!

First of all, one must have an understanding of how the scoring algorithm works on the GMAT. The exact algorithm is unknown and is undoubtedly far more complicated than I will be able to convey in this post, but there are some generalizations that can be made and an understanding of these general principals will go a long way toward achieving a better score.

Most people understand that the number of questions right is not the main determinant of a person’s score. Actually, at the higher levels of the Verbal section the number of questions right definitely does matter, so you need to be careful on Verbal not to put yourself too far behind because a string of wrong answers at the end of the test will definitely lower your score, especially if you were doing really well and tracking towards a very high score. But on Quant, most people get around half of the questions wrong and even people who score at the top end of the scale still get between 1/3 and ½ of the questions wrong. It is the level of difficulty of the questions that you are able to answer consistently right that really determines your score.

It is important to understand and really accept this fact – people whose score is artificially low because of poor time management often do not fully appreciate this fact and it is often the principal reason that their scores suffer. You can’t think of the GMAT the way you think about most tests (where, in order to score well, you need to get almost every question right). Because the questions will just keep getting harder and harder, it is a simple fact that at a certain point the questions will be too hard for most test takers and at that point it becomes foolish to waste time on questions that you are likely to get wrong anyway. This fact about the GMAT is one of the things that makes the GMAT so great, in my opinion. Just think of it – you know you are going to get about half of the questions wrong anyway, so it takes a lot of the pressure off on a questions-by-question basis. In my opinion this is much better than a test in which you need to get almost every single question right in order to do well (the GRE is actually an example of this and that is one reason that for some people the GMAT is a better test than the GRE).

Even if you believe that you are capable of answering all or most of the questions correctly, you need to be able to do it in a reasonable amount of time. If you can’t and if you try anyway, one of two things will probably happen (both bad): One is that you will spend a ton of time on the question and obviously if you do that too many times you will run out of time before you get to the end of the test. The other is that you will spend 2 minutes on the question, realize that you can’t answer it in a reasonable amount of time, and then make a guess, having now wasted 2 minutes. One of these scenarios is what plays out for most people who have time management issues.

The first step in correcting this is realizing that you will likely get ½ the questions wrong no matter what you do. So it doesn’t make sense to “go for it” on every question, especially if it’s a question that you recognize is hard and/or will take a long time to solve. My next post will cover strategic guessing and the different ways you can guess, both to save time and maximize your score, so stay tuned for that. A second consideration that will help is to realize that 2 minutes per question is just an average (75 minutes divided by 37 questions), but that does not mean that you should aim to spend 2 minutes on EVERY question. I wrote another post about this titled “The Myth of 2 Minutes Per Question” so I don’t want to rehash all that I wrote there, but basically it is fine to spend 3 or 4 minutes on some questions (if you are reasonably confident that you can get them right in that amount of time). If you couple that with some strategic “let-gos” (questions that you quickly dump once you realize that they are really hard or time consuming), you get a much better result than spending 2 minutes on every question since you can spend 3 or 4 minutes on some questions, thereby allowing you to get those right and much less than 2 minutes on other questions that you probably wouldn’t have had right even if you had spent 3 or 4 minutes!

One last thing I will mention and come back to in a later post: the early questions definitely matter a little bit more and warrant slightly greater attention. Again this topic deserves its own blog post, but it is a pretty accepted fact in the GMAT world that the first 5 or 10 questions on the test have a greater impact on your score (and all of my experience over the last 11 or 12 years tutoring people for the GMAT supports that). Now, some people go overboard with this and spend way too much time on the early questions and then completely run out of time way too early in the test. Basically what I advise my students is that you should be A LITTLE MORE WILLING to go for it on the early questions. So if you are close to an answer and you need an additional 30 seconds, I would take it. Or if it’s a question that you are not sure that you can answer but you think maybe you can or you anticipate that it will take a long time but that you can do it, then I would go for it if you are on question 2 or 3 or something like that. BUT, you should not feel like you have to get all of the early questions right – that is almost certainly not going to happen and aiming for that will almost certainly cause you to fall way, way behind. The other thing to throw in here is that the questions decrease in importance the further you move away from question 1, so you should be less and less willing to spend extra time as you move further into the test and by around question 10 it doesn’t really matter anymore.

Remember, time management is an issue for almost everyone who takes the test. I scored a 780 on the GMAT and even I have to push myself on both sections to make sure I finish on time. Even if you are slow in the way that you work, it just means that you need to learn how to be more selective in choosing which questions to really go for it on so that you have time to answer the ones that you really can answer. The flip side of that is that you obviously need to be willing to let go (quickly let go) of questions that are low percentage questions for you or that will clearly take a long time to answer. My next post will expand upon how exactly to judge when to skip a question and the different types of guessing that can be employed in order to save time and maximize your score. Stay tuned!

In my years as a GMAT tutor I have had many students come to me complaining that although they have done a lot of prep for the test, their scores just are not increasing. There are many reasons that this could be the case, but in this post I would like to discuss what, in my experience, is by far the most common cause and also the easiest one to solve: a reliance on non-official resources!

This may seem over-simplistic, but I really can’t even begin to count the number of times that I have been contacted by someone who BELIEVES that they have not been improving when in fact they have. In fact, I just began tutoring someone who came to me by googling the exact title of this blog post (she was the motivation for this blog post). When I found out what she had been doing (essentially using third party resources without using the official questions and tests released by GMAC) I said what I say to all the people in this situation: “you don’t even know if you have been improving or not because you are not using official questions or taking official practice tests.” And so I had her do what I have all people in this situation do: take an official practice test.

What was the result? The apparent lack progress after months of studying and 6 “non-official” practice tests was revealed to be actual progress. She believed that she had gone from a 530 to a 540 (not a very impressive increase after all of the work that she had put in) when in fact she had improved to a 670!!! The truth is that it’s not even clear how much of that was actual “improvement” since her initial diagnostic was from the same prep company that all her subsequent tests were from (I don’t want to name the company but it is one of the major players). It’s quite possible that her initial GMAT diagnostic score might have been a 600 had she taken an official test!

I have written other posts about the need to use official GMAT resources, but I thought that it would be worthwhile to write one devoted specifically to how harmful it can be to rely on 3rd party resources, especially in trying to measure one’s own progress. Let me enumerate the reasons why it can be so detrimental to gauging one’s improvement…

The most obvious and most damaging issue is that 3rd party practice tests do not use the actual scoring algorithm used on the GMAT. Test prep companies certainly aim to approximate the algorithm but they do it with varying degrees of success ranging from ok to not even close! Seriously, not even close!!! So when you rely on these tests you really may have absolutely no idea where you are actually scoring. The tests may be telling you that you are at a 550 when you are really closer to 700 or even the other way around. The aforementioned girl who I just began tutoring received Verbal scores no higher than 34 on the non-official tests she was taking but got a 44 on her first official GMAT Prep test! 44! That is a ridiculous Verbal score (almost 99th percentile) whereas the 34 Verbal was just ok.

Furthermore, the questions themselves are not that accurate. When I do official GMAT practice tests, I usually get 1 Verbal question wrong on the whole section. On 3rd party practice tests, however, I may get 6 or 7 Verbal questions wrong. Why would that be the case? It is not that the questions are just harder – it’s that they are ambiguous or just not well-vetted, well-written questions and so the kind of reasoning than can and should be applied on real GMAT verbal questions often can’t be applied. So not only does one risk having an inaccurate representation of their score when using 3rd party practice tests, one also risks not learning how, exactly, to approach real GMAT questions.

So if your GMAT score SEEMS to not be increasing, the first thing that you need to do is get an accurate understanding of where your GMAT score actually is. It might be that your score IS quite a bit higher than you think (this is often the case for people who come to me complaining that their score is not increasing). Using 3rd party resources is not necessarily bad, but as you get deeper into your GMAT journey you must turn increasingly if not entirely to official resources and of course to get an accurate gauge of where your score really is you need to do an official GMAT Prep test.

If indeed you determine that your score really is not increasing then of course there are reasons why this might be the case. In my next post I will examine some of the other factors that can contribute to the lack of progress and will discuss what can can be done about it.

It never ceases to amaze me how few people use the official GMAT resources that are available. This is perhaps a credit to how good the major test prep companies have gotten at convincing people that their resources are an accurate representation of what is on the test. While third party resources can be very valuable for learning the content that appears on the GMAT or for learning how to more effectively approach questions, when it comes to doing practice questions or especially practice tests, there just is no substitute for using official GMAC-released practice questions and practice tests.

I have written about this in other blog posts, but I am not exaggerating when I say that probably 1 out of every 2 people who contact me for help with the GMAT has either never taken an official GMAT practice test or has not been using official GMAT questions to practice with (and these are people who HAVE been using other resources like those of ManhattanGMAT or Kaplan or one of the other test prep companies). I think what usually surprises me most in these encounters is that people don’t even seem to understand the difference! Often the reaction that I get is something to this effect: “You mean that ManhattanGMAT tests are not official tests?!”

Sometimes just getting people to start using official GMAC resources is enough to get them a significant score increase. In fact, sometimes just even having them take an official practice test is enough. Again a very common occurrence for me is a student who contacts me complaining that they have studied a lot but are still stuck at the 550 mark or 580 mark or something like that. When it becomes clear that all they have done is ManhattanGMAT tests I will often say, “well how do you know you are stuck at a 550 mark when you have never done an official test?” In many of these cases the person then goes and takes an official GMATPrep test and contacts me back to let me know that they scored in the mid-600’s or some score that was significantly higher than what they had been scoring on the ManhattanGMAT or Kaplan tests that they were taking.

Again I am not arguing that third party resources are useless. Even the practice tests can be useful when used for the correct purpose. But trying to give yourself an authentic test-taking experience or trying to gauge exactly what score you are at are not reasons to take a third party test!!! Test prep company materials are particularly good at teaching the content that serves as the foundation on the Quant and Verbal sections of the test and some resources are also good at providing insight into how to approach the questions more effectively. The official GMAC resources are not particularly good in those regards.

But once you have learned the core fundamentals and have begun to understand how to approach the questions you MUST, MUST move on to official questions and then official practice tests to ensure that you are practicing on the types of questions that will appear on the exam.

There are a couple of reasons this is so important. First of all, it is very difficult to write questions that really accurately reflect the questions on the actual exam and test prep companies typically do not have their absolute most experienced and knowledgeable people writing the questions. This is especially true for Verbal questions, which are so hard to write in a way that accurately reflects the logic of the questions on the actual test. If you write a sentence correction question that is too grammar focused and not logic and meaning focused enough then you have a question that really departs from the actual questions in a pretty significant way.

Another issue is the distribution of questions that you see. Often test prep companies focus too much on some question types and not enough on others. So you might be practicing on a lot of combinatorics and probability questions when these questions are not that well represented on the actual test (and even when they are, they are often very difficult questions that may be better to guess on anyway). This is especially true of 3rd party practice tests – if the distribution of different question types is not accurate to the test then the practice test is not a realistic representation of the test. What’s worse is that when you analyze your performance on the test you may walk away with the belief that you need to work on particular topics (like combinatorics and probability) when you are coming to that belief only because you saw an unrealistically disproportionate number of those questions.

And with the practice tests, level of difficulty is important too. If the questions on a non-official practice test are just too hard then you may run completely out of time and draw the conclusion that you have serious time management issues that you don’t actually have. And the score that the tests output may lead you to believe that you are doing much better or much worse than you actually are – test prep companies try their best to approximate the scoring algorithm of the actual GMAT, but no one knows exactly how the algorithm works so the scoring system on third party practice tests is bound to be at least a little bit inaccurate (and in some cases it can be wildly inaccurate).

The official guide books and the official GMATPrep tests, on the other hand, present real questions that appeared on previous GMATs (which means that the questions underwent a rigorous vetting process before they ever made it onto the exam). The distribution of questions, both in the books and on the tests, is a good representation of what appears on the actual GMAT. And the scoring algorithm on the GMATPrep tests is the actual scoring algorithm that is used on the GMAT.

So people, please, please use the official GMAT resources!!! Again it is fine and even helpful to use third party resources to help you master the foundational material and learn how to approach the questions more strategically and effectively. But when it comes to practice you MUST use official questions. And when it comes to taking practice tests and gauging your overall performance, you MUST use the official GMATPrep practice tests!

Practice tests are a crucial component of GMAT preparation, but people often don’t understand how or when to use them or even which tests to use for which purpose. So which tests should you use and when should you use them? Well first you need to understand the differences between the tests that are out there and the fundamental purpose of doing a practice test.

I can’t even begin to count the number of times the following scenario has played out: A prospective GMATer contacts me for GMAT tutoring, often in a state of extreme distress because their “practice test scores” are not as high as the score they want to get on the test. When I ask them WHICH tests they have been taking the answer is often Princeton Review or Manhattan GMAT or one of the other 3rd party tests, yet in many of these cases the person never took an official practice test (a GMATPrep test from GMAC). When I point this out to them the usual response I get is, “you mean tests from Princeton Review or Manhattan GMAT are not real practice tests?”

People who are not in test prep industry often don’t initially understand that non-official tests (those not created by GMAC) are not real tests. This is true for 2 important reasons: the questions are not real questions (they were written by employees of the test prep company) and the algorithm that the test uses is not the actual scoring algorithm that the real GMAT uses (both in terms of the way it spits out questions and the score that it calculates). This means that your score will not be extremely accurate since the questions themselves are not authentic questions, the way they are distributed on the test is not necessarily accurate, and the way the score is calculated does not necessarily reflect the way it is calculated on the actual GMAT.

This does not mean that there is no place for 3rd party practice tests, but the GMATPrep tests are pretty much exact reflections of the actual GMAT (they use retired questions and the actual GMAT scoring algorithm and are therefore clearly superior). Unfortunately there are only 4 GMATPrep tests available, so you have to use them sparingly and for the right purpose. This means that you probably have to use some 3rd party practice tests at some point, but when do you use which test and why?

Well to begin, you need to have an understanding of the different functions that a practice test can serve. I would argue that there are 4 main reasons to take a practice test:

1) To see what your score is
2) To work on time management
3) To get practice putting it all together and taking a full-length exam
4) To get a sense for what else you need to work on as you near your test date

Now, to work on time management or to get some practice putting the whole experience together on the computer in a full-length exam setting, 3rd party practice tests can be very useful. The problem is that you really need to limit them to those objectives and not try to go too far with them, because by and large they are not extremely accurate and are not even that great at indicating what you need to work on further. For example, sometimes the 3rd party tests spit out the wrong distribution of questions, so if you take a Kaplan test and it gives you 4 combinatorics and probability questions, you might get 3 of them wrong and then freak out, believing that you now need to work on those topics. But that may be misleading since the actual GMAT would be extremely unlikely to throw you 4 combinatorics and probability questions and your worries would therefore be misplaced.

It is not easy to take a practice test and then NOT read too much into the results, but for non-official practice tests you really need to try. All too often people freak out because they don’t like what they are seeing with their practice test scores, but if you scored a 680 on a GMATPrep exam and then a 620 on a Manhattan GMAT test, you can’t compare the scores and infer that you are going in the wrong direction. It is not an apples to apples comparison. The Manhattan GMAT tests have different questions and a different scoring algorithm so as difficult as it is you must keep this in mind!

Basically what I advise is this:

Start with a GMATPrep exam in the beginning of the process to get a baseline and see what your weaknesses are. Then spend some time preparing by just studying and doing practice questions but forego the tests for a while. Then, when you think you are around the midway point in the process and have done a lot of studying, it is time to take a practice test again. At this point, since time management might still be an issue, I would take a 3rd party practice test so that you get some practice putting it all together and working on time management, but you don’t burn one of the GMATPrep exams. If that goes reasonably well, you can then take a GMATPrep exam to see what your actual score is. If the experience did not go well, you may want to stick with 3rd party tests until you feel like you have a better handle on time management and the experience of taking a full test on the computer before turning to the GMATPrep tests.

3rd party practice tests generally give you a question-by-question breakdown of how much time you spent on each question, whereas the GMATPrep tests do not, so this is actually one advantage of using the non-official tests. When working on time management is your major objective, I would practice on the 3rd party tests so that you can really do them over and over without worrying about burning the GMATPrep tests.

Nevertheless, once you have ironed out the time issues and especially as you get closer to your test date, you need to turn more and more to the GMATPrep tests so that you have a realistic understanding of your score and of the kinds of questions that you are struggling with (again, remember that the non-official tests can sometimes be misleading in the weaknesses that they point to). And I personally believe really strongly that the very last practice test you do (if not the last 2) should be a GMATPrep test. You just don’t want to be using inauthentic questions on the very last practice test before your actual GMAT.

Again there is nothing wrong with using non-official tests, but you must understand the differences between the 3rd party tests and the ones released by GMAC. Furthermore, you need to have an understanding of what benefit the practice tests can provide and think about what specific purpose or purposes you are trying to achieve in order to determine which test you should use (given the unfortunate limit on official GMATPrep tests). To repeat, if your goal is to have an accurate understanding of what your score is or some realistic feedback about what your specific weaknesses are, you really need to use the GMATPrep tests. If it is really more about working on time management or just getting some additional practice putting it all together with full tests on the computer, then the non-official tests are great for that purpose and in some ways superior to the GMATPrep exams (because of the question-by-question timing breakdown that they offer). Nevertheless, resist the urge to read too much into the scores that they output since they will not necessarily be that accurate and may mislead you as to what your weaknesses are and what more you need to do to prepare for the exam.