How accurate is this score? Please help.

Completely agree with you ajitesh and mike.

But I was wondering if the verbal score really shows the accuracy ? I mean to say that I got good number of questions correct but still managed to score only 27.

Also what advise would you give to take this score to a 680+?


Thanks

Dear Shiv2016,

I'm happy to respond.

First of all, unlike every other test you have taken in your life, the GMAT uses Computer Adaptive Testing. On the CAT, the # of question you get correct is entirely meaningless. Basically, the person who gets a 460 and the person who gets a 780 probably get about the same number of questions correct--what differs is the difficulty of the questions. The logic of the CAT is very subtle: I urge you to read that linked article. You see, the CAT is trying to home in on your level. Say it gives you a question of the 45th percentile and you get it right, then it gives you a question of the 48th percentile and you get it wrong. After this, the computer will keep giving you questions in narrow difficulty range, so that it can give you a very focused result. In the course of this process, it will give you a large number of questions in the mid-40-something percentile, and, of course, you will get many of those correct, but that simply serves to narrow your score into that percentile range. Reporting the number you got right is about as useless as reporting the color of the shirt you were wearing--it is a number that contains zero information about your performance.

In my understanding, the GMAT Prep program follows roughly the same CAT algorithm as the real GMAT. The GMAT is psychometrically one of the most well-designed and well-calibrated tests in the world. The GMAT Prep tests are very well calibrated. Now, the companies that have been in this business a while, such as Veritas, MGMAT, and Magoosh--we do our best to mimic the official scores, and because we have a lot of data from the performance of our own students, the score predictions of these companies are reasonably good. There is absolutely no a priori reason to doubt the validity of your score.

Now, what can we do about this? The performance was 600 (Q46 and V27). Let's say, hypothetically, that you simply keep your strong Q score at the same level. Getting to the 680 level would involve boosting your verbal score up to the high 30's. That's a HUGE change. This can be done, but you need to be willing to work as hard as, and perhaps harder than, you have ever worked before. You are going to have to take your diligence, your dedication, and your commitment up a few notches. You are going to have to push yourself to levels you didn't know you had. Yes, you can do this, but this challenge is not for the feint of heart.

First of all, I encourage you to embrace ardently all the habits of excellence. You want to achieve a truly excellent increase, so you need to live, sleep, and breathe excellence 24/7. You need to raise your expectations of yourself in every area of your life so that the habits of excellence take root most firmly.

Of course, practice GMAT preparation materials. Magoosh has a lot that can help you. But above and beyond any GMAT-specific work, you need to immerse yourself in the English language. In particular, you need to develop a serious habit of reading. See this blog article:
How to Improve Your GMAT Verbal Score
You have to throw yourself with full enthusiasm into whatever is most difficult for you.

My friend, the reason that many people don't achieve excellent results is because they don't have the moxie. Folks want convenience. They want to be able to mail it in. They want to follow a small handful of tricks and tips, hoping this will produce spectacular results. Of course, that lily-livered level of commitment produces little more than mediocrity. Here, I will remind you of the great law of mediocrity: if you put in about the same effort as other people put in, you will wind up with results that are about the same as the results that most other people get. If you want an outstanding increase, an increase from 600 to 680, you need to put in an effort that is outstanding in every way.

Finally, I will say: do not underestimate how common it is for business schools to see applications from non-native speakers from India and China who are really good at Q and who struggle at V. Improving your GMAT score would help you stand out, but you also need to stand out in every aspect of your application. You need to know who you are in a profound way. You need to have passion and drive about what you want to accomplish in your life, and you need to convey that effectively. Then, you will stand out.

Does all this make sense?
Mike

That's roughly true if you compare two scores closer to the middle of the scoring scale. But the 780 scorer will have far fewer wrong answers than the 460 scorer. It's just not possible to get a high enough Verbal score to get a 780 with a typical number of mistakes. I've never seen a V47 or higher score with more than two mistakes in total on GMATPrep tests, for example. The reason your hit rate has to be extremely high to get a 99th percentile Verbal score is because there just aren't any questions in the Verbal question database that are 'above your level' if you're a V47+ scorer. And to get a Q51 score you also need to be perfect, or nearly so, though on a test with a lot of hard questions, you can get a Q50 with a lot of mistakes.

Thanks a ton Mike and Ian. Your words are truly amazing.

I will keep you updated with my CAT scores.

Hi Experts,


Even i faced the same issue.

I gave my CAT today(23rd July 2017) and i got a 520(Quant 30,Verbal 31). It is really surprising because i got more wrong questions last time but i still scored 610 and now i have a score of 520.

I have attached the screenshot of my today's CAT. I saw a few repeated question from my CAT 1(GMAT EXAMPREP 1).

Can someone confirm is the scoring accurate?

GMATNinja ,mikemcgarry , IanStewart

Unfortunately, it's not about how many questions you get right! The GMAT's scoring algorithm doesn't even look at how many you got right or wrong. Here's some reading material:

https://www.manhattanprep.com/gmat/blog ... mat-quant/

Now, if you looked at all test takers, people with higher scores would get more questions right (usually), and people with lower scores would get more questions wrong (usually). (However, the difference is small, unless you look at the very high end and very low end.) But think of that as a symptom of their high or low score, not the cause of their high or low score. The same behaviors and practices that give you a high score on the GMAT will cause you to get more questions right. That doesn't mean that trying to maximize the number of questions you get right will increase your score (in fact, taking the test with that mindset is more likely to decrease your score.)

In short - it's impossible for us to say whether you can get a particular score with a particular number of questions missed, since the GMAC won't release that data - and it wouldn't be very interesting data anyways. But I've seen a very wide range of scores with an identical number of missed questions.

I like studies like this one, and it's especially interesting to see how many wrong answers you can have and still get a Q49 or Q50 on some tests. That's important for test takers to understand, so they don't become overly concerned in the middle of the test when they encounter hard questions they might need to guess at.

But I think you'll find that if you repeat the same experiment, but using a different GMATPrep test (one of the GMATPrep tests you can buy, for example), that the number of wrong answers for each test taker will generally be lower. A 600-level test taker will answer 60% of 600-level questions correctly - that's the definition of a 600-level test taker. So if a GMAT were to deliver a test taker questions exactly at her level throughout the test, that test taker would normally answer 40% of all questions incorrectly, or about 15 questions incorrectly.

But in practice, the test doesn't deliver questions at a test taker's estimated level throughout the test. The test would like to do that, because those questions give the test the best information about the test taker, but there are other criteria besides ability that the algorithm uses to choose the next question - the test needs to ensure the right number of DS questions, and of Geometry questions, and so on, and it also tries not to overuse any one question. Because there's a limited supply of questions in the question database (and that supply seems especially limited in the paid-for GMATPrep tests), then the test is often delivering questions that are not very close to its estimate of the test taker's level. And just by 'luck' (or really by the characteristics of the underlying question bank), the test can sometimes generally be 'hard' for the test taker, because the test is more often than not choosing questions above the test taker's level, or it can be 'easy' for the test taker. That doesn't affect test scores in any way, but it will affect the number of wrong answers a given test taker will have.

From the data in your experiment, it sounds like the particular GMATPrep test the students took was a hard test overall. On a test that skews to the easy side, you can find test takers with only 9 or 10 wrong answers, but still with scores no higher than the mid Q40s. I'd guess the average number of wrong answers on a normal test, for test takers in the Q30-Q50 range, is about 14 or 15 (and that includes Q50-level scorers - they just only answer the hardest questions incorrectly).

And for test takers trying to understand the relationship between right answers and your score, this analogy maybe will be helpful (there's a different analogy I prefer to use, because it's more precise, but it also would take longer to explain).

Say you were on a college baseball (or whatever sport or game you like) team, and you wanted to know how good your team was before the college baseball season started. If you played 10 games against the New York Yankees (one of the best teams in the world) and you won 60% of those games, you'd (correctly) think your team was absolutely amazing. If instead you played 10 games against the five year olds on a local school's t-ball team, and you won 60% of those games, you'd (correctly) think your team was not very good.

So just knowing that you won 60% of your games is not enough information to tell you if your team is good or average or bad. You need to know who you were playing. And that's the information the scoring algorithm is using. The algorithm knows, for each question, how often a 300-level test taker, and 500-level test taker, and 700-level test taker (etc) will answer correctly. And using some slightly advanced probability theory, when it has your responses to a batch of questions, it can use those probabilities to give you a reliable score. But when you're facing a very tough question (much like the college baseball team playing against the Yankees) you aren't expected to 'win' very often, even if you're very good.

Hello IanStewart, mikemcgarry, GMATNinja, and bb

This post is an update. I gave another Veritas test and scored a 610 : Q45 and V29. I did not take GMAT Prep because it is not working in Dell notebook.

Would you like to suggest how can I take this score to 680 or may be 700 in less than a month?


Thanks

One more thing I would like to add is that all the questions were of 500-550 level.Not a single 600-650 level question asked even when I answered questions correctly in a row.


Why is that?

I had that issue too where I had an amazing run in verbal of 1 wrong question in a 24 question string with 9 wrong total and it didnt seem to make any difference, 9 wrong in quantitative and a 610 result.......I understand that the test evaluates you at a very high level but it seem ridiculous. In particular when the tested scenarios that ive seen on this site i.e get 6-8 in the middle wrong or guess c for last 10 etc are still getting higher results.

So did you take any other test to compare your score?

ive taken several practice tests, bought a couple of new tests ranging from month ago 550 to more recently 600-660. my # of wrong answers keep improving, my question sets out of the book keep improving but im stuck in this band. 10/11 Reading comp = RC 73rd percentile? 3 wrong DS questions=DS 55percentile? I understand if I get multiple questions in a row wrong or get a really bad RC topic but I dont get the razor thin margins on some of these grades

Yes that can happen. But as explained by the experts in above posts, scoring depends on how difficult question you answer. That's what creating confusion.

In Veritas test, I got all easy questions in quant. I am wondering if this (45) even true or is my score much lower. Also why do people call Veritas test tougher than the real test when such easy questions are asked in the test?

difficulty depends on which questions you get, some catrgories maybe you find easier than someone else or vice versa. The getting harder questions right or easier ones wrong is understandable and the score will find your score or the Entire Test, it the one or two or three wrong answers which is scary in that these simple little misakes can wipe out your entire test. Why a 10/11 1 wrong answer even with a 500L weight puts your RC into 73rd percentile seems extremely harsh.

Dear Shiv2016,

I'm happy to respond. :-)

First of all, I want to point out an assumption in your question. You are assuming that a 610 on a Veritas test, taken in the privacy and comfort of your own home, is in all ways equivalent to a 610 on the real GMAT. I want you to read this blog article very carefully:
Lower on the Real GMAT than on Practice Tests
Read not only that article, but all the linked articles. This is a very important issue.

The CAT operates by tuning into your range, so after any GMAT, most of the questions you saw on Q &V would have been approximately in the range of your score. This is not unusual at all.

Let's assume you really are at 610 level, about the 62nd percentile, and you want to go to 700, the 90th percentile, in just a month. Do you understand that, in terms of percentiles, this is an extraordinary leap? It's not impossible, but it's not as if you can simply follow five quick & easy tips to get there. You can't mail this in. You really have to show up as you have never shown up before.

Here are my questions to you. How hard are you willing to work, above and beyond what you have been doing so far? How diligent can you be? How dedicated can you be? How much focus can you sustain? How much determination can you generate? What non-essential parts of your life are you willing to sacrifice for this? How much can you do all of this at a much more intense level than you have been doing?

It will help you to embrace the habits of excellence. It's not just enough to read those and acknowledge them as good ideas. You must make each part and parcel of how you live over the next month. Remember: how you do anything is how you do everything. If you can generate breakthrough levels of commitment and engagement over the next month, you can have a breakthrough performance. By contrast, if you simply continue to do more or less what you have been doing, you will get more or less the same results.

Does all this make sense?
Mike :-)