Accuracy of GMATPrep score?

This topic has expert replies
Junior | Next Rank: 30 Posts
Posts: 18
Joined: Wed Sep 30, 2009 6:53 pm
Thanked: 3 times
Followed by:1 members

Accuracy of GMATPrep score?

by mj12g » Sat Aug 20, 2011 12:31 pm
Hi all,

I realized that this question has come up in various forms over and over, but I'm starting to question the accuracy of my GMATPrep score, at least when it comes to verbal. I took the GMAT two years ago after taking a Kaplan course and scored a 720 (40Q,56%/50V,99%). Obviously, I'm much stronger in verbal than in quant. I applied to schools last year and only got into one, so I'm retrying my applications this year. I've been (obviously) focusing my studying solely on quant; I just took GMATPrep #2 and scored a 770 (49Q/48V). I took the first one last weekend and scored a 740 (I forget the exact split, but it was something like 47Q/46V). Now the interesting part (and I know that it doesn't really matter how many you get wrong): on GMATPrep 1, I got 12 quant incorrect but only 4 verbal incorrect. On GMATPrep 2, I got 9 quant incorrect and only 1 verbal incorrect. I have a hard time believing that a single error on verbal should only be a 49 and, furthermore, that I could get so many wrong in quant and still get a BETTER quant score than verbal, since quant has always been my Achilles' heel. I'm starting to think that the GMATPrep has some quirks that make it inappropriate for me to use it to judge my score. Any insight on this?

Junior | Next Rank: 30 Posts
Posts: 29
Joined: Sun Jun 26, 2011 9:19 am
Thanked: 3 times
Followed by:1 members
GMAT Score:710

by dmf5238 » Sat Aug 20, 2011 3:41 pm
The GMATPrep is typically accepted as the most accurate gauge of GMAT performance as compared to other CATs. With that said, the accuracy of it relative to actual GMAT performance is questionable, as it uses retired questions, and you're taking it in a non-official setting, in a non-official mindset.

I posted a question in my own GMAT thread asking for a reasonable score deviation range for the GMATPrep practice exams, and I haven't gotten a response yet. I've heard that +/-30pts is a fair assumption, but I can't recall where I heard that, so take that number with a grain of salt.

Can anybody confirm/deny this?

P.S. I'm in the same boat as you; I'm naturally better with Verbal than Quant. :)

Junior | Next Rank: 30 Posts
Posts: 18
Joined: Wed Sep 30, 2009 6:53 pm
Thanked: 3 times
Followed by:1 members

by mj12g » Sat Aug 20, 2011 6:35 pm
Thanks for the response. I'm not so much talking about the score as a predictor of actual GMAT performance (although that would be nice to know), but the accuracy of the software of transforming your responses into a real GMAT score. In other words, I find it hard to believe I actually got a 49 in quant, since I got 9 wrong (i.e. the score is too high) and also that I got a 48 in verbal with only 1 wrong (i.e. that score is too low). So my basic question is: does the GMATPrep use the same scoring algorithm as the real GMAT, or do they somehow "dumb it down" to prevent people from reverse engineering it?

On an aside, I find the verbal section of the GMAT to be rather unfair. As a native English speaker, I find the questions to be fairly trivial. But I literally could not imagine taking the test if I did not have 25 years of native English under my belt. I realize that business schools are looking for exceptional English ability, but they seem to be testing both English proficiency and critical thinking in the same step, when it would seem better to split them into two sections. In any case, I'm glad I have the "gift" of being an English speaker, as this means I can almost not study for verbal at all.

User avatar
Master | Next Rank: 500 Posts
Posts: 398
Joined: Tue Jul 26, 2011 11:39 pm
Location: India
Thanked: 41 times
Followed by:6 members

by prateek_guy2004 » Sat Aug 20, 2011 11:33 pm
mj12g wrote:Hi all,

I realized that this question has come up in various forms over and over, but I'm starting to question the accuracy of my GMATPrep score, at least when it comes to verbal. I took the GMAT two years ago after taking a Kaplan course and scored a 720 (40Q,56%/50V,99%). Obviously, I'm much stronger in verbal than in quant. I applied to schools last year and only got into one, so I'm retrying my applications this year. I've been (obviously) focusing my studying solely on quant; I just took GMATPrep #2 and scored a 770 (49Q/48V). I took the first one last weekend and scored a 740 (I forget the exact split, but it was something like 47Q/46V). Now the interesting part (and I know that it doesn't really matter how many you get wrong): on GMATPrep 1, I got 12 quant incorrect but only 4 verbal incorrect. On GMATPrep 2, I got 9 quant incorrect and only 1 verbal incorrect. I have a hard time believing that a single error on verbal should only be a 49 and, furthermore, that I could get so many wrong in quant and still get a BETTER quant score than verbal, since quant has always been my Achilles' heel. I'm starting to think that the GMATPrep has some quirks that make it inappropriate for me to use it to judge my score. Any insight on this?
Hi

Gmat prep is the most accurate score reflector indeed, after seeing your scores i am amazed to see ur verbal.

Well to be honest scoring 49 in verbal is almost 99 percentile and its impossible to cross that.

So be confident about it, gmat prep only reflects your master abilities......

Junior | Next Rank: 30 Posts
Posts: 29
Joined: Sun Jun 26, 2011 9:19 am
Thanked: 3 times
Followed by:1 members
GMAT Score:710

by dmf5238 » Sun Aug 21, 2011 4:47 am
mj12g wrote:Thanks for the response. I'm not so much talking about the score as a predictor of actual GMAT performance (although that would be nice to know), but the accuracy of the software of transforming your responses into a real GMAT score. In other words, I find it hard to believe I actually got a 49 in quant, since I got 9 wrong (i.e. the score is too high) and also that I got a 48 in verbal with only 1 wrong (i.e. that score is too low). So my basic question is: does the GMATPrep use the same scoring algorithm as the real GMAT, or do they somehow "dumb it down" to prevent people from reverse engineering it?

On an aside, I find the verbal section of the GMAT to be rather unfair. As a native English speaker, I find the questions to be fairly trivial. But I literally could not imagine taking the test if I did not have 25 years of native English under my belt. I realize that business schools are looking for exceptional English ability, but they seem to be testing both English proficiency and critical thinking in the same step, when it would seem better to split them into two sections. In any case, I'm glad I have the "gift" of being an English speaker, as this means I can almost not study for verbal at all.
I see what you're saying. I suppose that's hard to confirm or deny, because it's logical that GMAC would change the algorithm from the real GMAT so that people can't discover how it actually works, but I don't believe that they say anywhere that it uses the same algorithm. I found the following quotes on the GMATPrep Website:
GMAC wrote:The GMATPrep software uses the same technology used by the official GMAT® exam. With this software, you can simulate the actual test experience.

GMATPrep software features:
  • A 15-question practice section for each type of GMAT question, with answers and explanations for each question
  • 2 full-length computer adaptive practice tests with answers (detailed explanations of answers are not included)
  • Accurate computing of Verbal and Quantitative scores, which you can use to judge how prepared you are for the actual GMAT exam
  • A comprehensive math review
  • Real-time scoring
The only conclusion I can draw from this is to just say that it comes down to how the CAT works. You can't really judge your score by the number of questions you got wrong, but by the difficulty level of those questions that you got wrong. Getting an easy question wrong will hurt you more than getting a hard question wrong.

Best of luck! :)

User avatar
Master | Next Rank: 500 Posts
Posts: 398
Joined: Tue Jul 26, 2011 11:39 pm
Location: India
Thanked: 41 times
Followed by:6 members

by prateek_guy2004 » Sun Aug 21, 2011 8:18 am
dmf5238 wrote:
mj12g wrote:Thanks for the response. I'm not so much talking about the score as a predictor of actual GMAT performance (although that would be nice to know), but the accuracy of the software of transforming your responses into a real GMAT score. In other words, I find it hard to believe I actually got a 49 in quant, since I got 9 wrong (i.e. the score is too high) and also that I got a 48 in verbal with only 1 wrong (i.e. that score is too low). So my basic question is: does the GMATPrep use the same scoring algorithm as the real GMAT, or do they somehow "dumb it down" to prevent people from reverse engineering it?

On an aside, I find the verbal section of the GMAT to be rather unfair. As a native English speaker, I find the questions to be fairly trivial. But I literally could not imagine taking the test if I did not have 25 years of native English under my belt. I realize that business schools are looking for exceptional English ability, but they seem to be testing both English proficiency and critical thinking in the same step, when it would seem better to split them into two sections. In any case, I'm glad I have the "gift" of being an English speaker, as this means I can almost not study for verbal at all.
I see what you're saying. I suppose that's hard to confirm or deny, because it's logical that GMAC would change the algorithm from the real GMAT so that people can't discover how it actually works, but I don't believe that they say anywhere that it uses the same algorithm. I found the following quotes on the GMATPrep Website:
GMAC wrote:The GMATPrep software uses the same technology used by the official GMAT® exam. With this software, you can simulate the actual test experience.

GMATPrep software features:
  • A 15-question practice section for each type of GMAT question, with answers and explanations for each question
  • 2 full-length computer adaptive practice tests with answers (detailed explanations of answers are not included)
  • Accurate computing of Verbal and Quantitative scores, which you can use to judge how prepared you are for the actual GMAT exam
  • A comprehensive math review
  • Real-time scoring
The only conclusion I can draw from this is to just say that it comes down to how the CAT works. You can't really judge your score by the number of questions you got wrong, but by the difficulty level of those questions that you got wrong. Getting an easy question wrong will hurt you more than getting a hard question wrong.

Best of luck! :)
dmf5238 I agree with you on this.....So better watch out for easy ques 300 - 400 range

User avatar
Master | Next Rank: 500 Posts
Posts: 234
Joined: Fri Jun 10, 2011 8:50 am
Location: Bangalore
Thanked: 47 times
Followed by:60 members

by arun@crackverbal » Fri Aug 26, 2011 6:20 am
Hi,

Some factors to consider about your practice GMAT tests - especially while understanding the discrepancy with the real test:

1. Did you take the test with the AWA? If answer to this question is NO then stop reading. It is pointless as you have defeated the purpose of the mock tests.

2. Did you take it in the same time as the real test? If it was 9am then did you take it then? Or did you take it when you work the best (say late evenings?)

3. Were you checking email/SMS/talking to people/smoking/drinking coffee/anything else you would not do on the real test? If yes this also can be a certain cause for fluctuation.

4. Are you bad at taking standardized tests? This could be because you just are not a big-test player. This is fine - just that you have to accommodate for a fluctuation in scores on the real test day.

5. Did you see any repeats from the OG or any question you have solved elsewhere (even a "similar" question - many test prep companies throw the same questions after giving it a wafer thin makeup). If so not only have you got the question right - you have also shaved 2 minutes of the test time. Unfair.

All these are factors which can sway the scores by a 100 or so points. Here is example - let us say you got 50Q v40V then you might end up with a 740 on one day. Not getting a few questions right on another day may cause it to dip a wee bit - Q48 V38 - a scaled score of 690! Ouch! :)

Arun
Founder of CrackVerbal - India's fastest growing GMAT Prepration and MBA Admissions Consulting Company. https://gmat.crackverbal.com

Free Ebook on GMAT | GMAT Scoring, Study plan, top study mistakes etc
Download here: https://gmat.crackverbal.com/15-minute-gmat-guide

Good enough to get into Harvard? Or would be it ISB? Get a free profile report PDF mailed to you: https://applications.crackverbal.com/fre ... valuation/

Junior | Next Rank: 30 Posts
Posts: 18
Joined: Wed Sep 30, 2009 6:53 pm
Thanked: 3 times
Followed by:1 members

by mj12g » Mon Aug 29, 2011 7:40 am
Thanks for the response arun...again, I'm not so much concerned with how well the GMATPrep predicts the actual score, more with just how the scoring algorithm compares with the real GMAT. But addressing your questions:

1. I didn't take the AWA; I'm going to do a little review of it, but I'm not too concerned about it, and the essays don't frazzle me (bear in mind that I've already taken the GMAT and got a 5.5 on the AWA, which I was fine with).
2. I haven't taken it in the same time as my real test, which is 8AM. I'm going to try to take an early one the day before the test, but I'm a morning person anyway, and I'm not too worried about it.
3. I wasn't doing anything distracting.
4. I'm generally fairly good at standardized tests (the SAT at least), although my 40 in Q last time was positively shocking. Now that I'm prepping only for quant, though, I have yet to see a quant score less than 44, and almost all of them have been 48 or 49.
5. No repeats at all.

In the end, I'm pretty happy with how my prep is going. Took my first two MGMAT practice tests this weekend: 720 on the first one (44Q/45V), 750 on the second (48Q/45V). I found the quant to be very hard on the MGMAT, and I was extremely short on time by the end. If the MGMAT is truly harder than the real GMAT, then I'm happy with how my quant is turning out.

User avatar
Master | Next Rank: 500 Posts
Posts: 351
Joined: Mon Jul 04, 2011 10:25 pm
Thanked: 57 times
Followed by:4 members

by akhilsuhag » Mon Aug 29, 2011 9:00 am
It doesn't matter how many you get wrong.
It is what you get wrong that is the real deal. I got a 49 with 15 incorrect, but what is important is that I did not get the easy ones incorrect. The algo judges what level of questions you get right and what do you get wrong and that is how your scores are calculated.

Also a 47 on Q is not better than a 46 V. A 47 Q is less than 80% I think or smwhr thr and 46 V is close to 99. GMATprep is the closest representation of your scores and it is a very difficult algorithm to judge/understand. The best thing is to let the algo do the scoring and leave the rest. You are scoring great, don't worry you wwill do great!!
Please press "thanks" if you think my post has helped you.. Cheers!!

User avatar
GMAT Instructor
Posts: 613
Joined: Thu Mar 22, 2007 6:17 am
Location: madrid
Thanked: 171 times
Followed by:64 members
GMAT Score:790

by kevincanspain » Mon Aug 29, 2011 1:21 pm
You need to realize that a 46 verbal score is the 99th percentile, as are all higher verbal scores.

A 49 in the quant section is 'only' the 89th percentile.
Kevin Armstrong
GMAT Instructor
Gmatclasses
Madrid

Junior | Next Rank: 30 Posts
Posts: 18
Joined: Wed Sep 30, 2009 6:53 pm
Thanked: 3 times
Followed by:1 members

by mj12g » Mon Aug 29, 2011 1:30 pm
kevincanspain wrote:You need to realize that a 46 verbal score is the 99th percentile, as are all higher verbal scores.

A 49 in the quant section is 'only' the 89th percentile.
True, although since I got a 50 in verbal on my first taking of the GMAT, I'm shooting for a 48-50 on my second try. If I can also pull off a 48 in quant, I'd be quite happy.

GMAT/MBA Expert

User avatar
GMAT Instructor
Posts: 2621
Joined: Mon Jun 02, 2008 3:17 am
Location: Montreal
Thanked: 1090 times
Followed by:355 members
GMAT Score:780

by Ian Stewart » Fri Sep 02, 2011 7:18 am
Your GMATPrep scores sound entirely normal. When comparing Quant and Verbal scores, you should be comparing percentiles (which essentially are what are used to generate your score out of 800), and not the raw scores out of 51. The raw scores were comparable when the GMAT was first developed (long before it was even called the 'GMAT'), but they're far from comparable now. If you score 47V, 47Q, while that's certainly an excellent result in both sections, you're actually miles ahead in Verbal - a 47V score puts you well into the top 1% of test takers.

If you are at the very extreme high end of the Verbal score range, each mistake will lower your scaled score (but not your percentile rank) significantly. On GMATPrep, each mistake (if you're in the 44-51 score range) on Verbal will lower your score by either 1 or 2 points. The same is not true on the Quant section. Most test takers will get about 1/3 of their Quant questions wrong, regardless of the level (except those either at the 51 level, or at the lower levels of the score range). The difference between someone scoring a 30Q and someone scoring a 49Q is the difficulty level of the questions he or she answers incorrectly. The 30Q test taker is answering a lot of medium level questions incorrectly (and is thus never seeing any hard questions, because of how the test adapts), whereas the 49Q test taker is answering almost all of the medium and medium-hard questions correctly, but not the crazy-difficult 800-level questions. Because the 49Q test taker answers all of the medium questions correctly, the majority of his or her test will consist of very hard questions, and the test is very forgiving, in Quant anyway, of mistakes on those questions.
For online GMAT math tutoring, or to buy my higher-level Quant books and problem sets, contact me at ianstewartgmat at gmail.com

ianstewartgmat.com

Junior | Next Rank: 30 Posts
Posts: 18
Joined: Wed Sep 30, 2009 6:53 pm
Thanked: 3 times
Followed by:1 members

by mj12g » Fri Sep 02, 2011 2:04 pm
Thanks ian...i actually took the gmat today and was shocked by how low my scores were: 44q/41v/700. I guess the gmat prep is not the best indicator of performance. Perhaps the best indicator is a random walk. I posted my disappointing results in the "I just beat the gmat" forum, but i thought I'd update this thread with my actual performance. I'm lost and confused at this point: I KNOW that I did well in verbal but somehow came away with a 41...my lowest score ever. C'est la vie

Newbie | Next Rank: 10 Posts
Posts: 1
Joined: Fri Jan 03, 2014 6:07 am

by [email protected] » Mon Jan 06, 2014 7:46 am
none of gmatprep are correct just they are for practice that's it

Newbie | Next Rank: 10 Posts
Posts: 1
Joined: Tue Jul 05, 2016 9:25 am

by zey783 » Tue Jul 05, 2016 9:34 am
This thread was helpful for me as I was preparing (I was a nervous wreck before the exam) so I thought I'd share my experience as well.

I scored a 750 on the real thing today (50 quant, 41 verbal). Here's how I did on GMAC practice tests:

GMAT Prep Exam 6: 750
GMAT Prep Exam 5: 740
GMAT Prep Exam 4: 740
GMAT Prep Exam 3: 740
GMAT Prep Exam 2: 720
GMAT Prep Exam 1: 720

I also took all six of the Manhattan Prep CAT exams and never broke 700. Their quant sections are way more complicated, with each question involving a lot more steps that the actual exam. If quant is your weaker section, don't be discouraged if you're doing poorly on the Manhattan Prep exams. As everyone always reiterates, GMAT Prep is the most accurate measure.

I felt like I was doing terrible during the actual exam. My quant section started off ok but then I felt like I completely bombed, especially because my last few questions were very, very easy. I honestly thought I had blown any chance of a 700+ score. Don't be discouraged! Even if you see easy questions towards the end.

Good luck and remember that hard work will pay off!