GMAT Quant 83 with Only 2 Mistakes

Question

Hey everyone. On January 2 I took the GMAT exam and got only 83 in Quant, even though I made just 2 mistakes. Is that normal?
Like, did the system really not give me any hard questions at all?

P.S. The 2nd question (the one with 7 minutes), I skipped it pretty quickly, and those 7 minutes were already added after I answered question 21.

GMATNinja · Answer

Congratulations on being awesome at quant, and I'm sorry that your score didn't quite feel right in the end.

For better or worse, these "weird" scores happen fairly often (particularly on quant), and we asked GMAC about it. Here's a video version of their answer:

https://www.youtube.com/watch?v=oAQuBizfIIQ

I hope that helps a bit!

bb · Answer

Hey! I have added you to the infamous club of horrible scores... congrats I guess?

These things happen sometimes. How were your GMAT Prep scores? https://gmatclub.com/forum/horrible-gma ... 34958.html

AjiteshArun · Answer

Hi ZakMakhonin,

I believe there is a significant mismatch between the skill levels of many test takers and the GMAT's ability to differentiate between those test takers. GMAT quant has  always been like this , and things may be even worse now (no DS, fewer questions). I also believe that the percentile tables are off, as 90 is just too common to be considered "100%".

This is a real problem, but unfortunately, we can't do anything about it.

bb · Answer

Interesting point about the Q90. Maybe for theoretical reasons they put it as 100th percentile. Really depends on how they look at it. However, I don't know that we have enough data to tell for sure. 
Out of about 1,500 scores that people have added and confirmed on GMAT Club in 2025, 227 are Q90. Which is impressive but still a small fraction of the 92K tests taken (approx) in that period. So maybe it is common but it may also not be as common.

AjiteshArun · Answer

Hi bb,

Are you sure your numbers are correct (they show that the situation with the current GMAT is really bad)? I get your concerns about sample size (and possibly how representative your sample is), but  those numbers mean ~15% of the confirmed scores on GMAT club are Q90 .

I know we don't have the actual numbers, but from what I can see, the new GMAT has got all three sections wrong. Quant and (to a lesser extent) verbal are just not in line with the ability levels of test takers at the high end. As for DI, although I also have concerns about question quality, I really do think GMAC has made DI too hard in terms of time management.

If anyone is interested,  this is what GMAC percentiles show about the verbal section . Basically, GMAC expects ~90% of all test takers to get a score of at least V73. For most test takers, this means the verbal scale is effectively V73-90.

bb · Answer

This sampling is not a super-representative one and I would say is instead a self-selected sample where the highest scorers have added their scores while the lower-scored audience refrained.

My point was the opposite - that instead of this being a 15% people are scoring Q90 that even with all the folks who are scoring high and practicing heavy on GMAT Club, only 227 people have scored the Q90 and I feel that represents a much bigger portion of Q90 scorers than Q80 for example as those people are either not on GMAT Club or if they are, they would not be flaunting their scores.

I guess there are many ways to look at one number when there is just one number, but I would not take it as a proof that Q90 is now an 85th percentile like Q50 was. I think it is really hard to get Q90, seems harder than Q51 based on what I am observing but maybe that's me. I never looked at the Q51 rarity.

DI is indeed a much harder section. It went from freebie to an impossible.

P.S. What do you mean quant and verbal are not in line with the ability levels? Meaning people are maxing them out? or the other way?

AjiteshArun · Answer

Got it. It's interesting that you see the number of GMAT club-verified Q90s as "only 227". In India, 227 would be considered too many (here the so-called 100th percentile is used to refer to the top 5 or 10 test takers out of, say, 200,000+ test takers). As for what I meant by quant and (to a lesser extent) verbal not being in line with high-ability test takers, it basically comes down to the number of mistakes we're seeing for the highest scores and the number of people getting those scores. Ideally, what accuracy should we see in an adaptive test like the GMAT? 50%, 60%, 70%? Whatever you feel is a good number, seeing a lot of test takers with accuracy close to 100% can actually be a problem. It means the test is not really adaptive for those test takers, and it also means that there's very little margin for error for other test takers, even those who make only 1-2 minor mistakes. My Q83 may make me among the least-qualified people to say this

but 100% accuracy for Q90 (or V90) doesn't mean the test is hard. It means the algorithm couldn't increase the difficulty level of the test. This is not a major problem if there are a lot of questions or when only a few people get those scores, but the GMAT may have as few as 19 real questions each in the quant and verbal sections (to be clear, these numbers aren't confirmed ). For comparison, if I remember correctly, the old GMAT had 28 real quant questions, and 30 real verbal questions. To be fair, GMAC tells us that the sectional scores are still accurate, but assuming that the reports on GMAT club are correct, it's tough to see scores like these: 1-mistake Q80 (64%) , 1-mistake Q81 (70%) , and 1-mistake Q84 (85%) . Even if these mistakes were on easy questions, surely that's at least partly the test's fault? Finally, I agree we shouldn't read too much into the numbers you mentioned, but do you publish this data anywhere? I think that could be a great resource.

ZakMakhonin · Answer

Thank you all for your responses. After reading other stories, I’m honestly shocked, someone got an 80 after making just one mistake, that’s incredible.

In my case, my GMAT Prep scores were consistently 81 - 82, so I’m happy with the result and, overall, I really have nothing to complain about.

Thanks again for the replies.

AjiteshArun · Answer

It's good to know that you're happy with the result. All the best for the applications!

cumfugiat · Answer

Thoughtful take. I think you’re drawing an important distinction that often gets lost:  100% accuracy at the very top isn’t evidence of extreme difficulty—it’s evidence of a ceiling problem.  In a truly adaptive test, high-ability test takers should still be challenged enough to miss  some  items, otherwise the algorithm simply runs out of headroom.

bb · Answer

Yes, I agree that Q83 with 1 incorrect is a test's fault to a large extent unless of course that was the real level for the user, in which case the test was spot on but the irregularity is absolutely the test's fault. However, having learned a lot over the last 6 months about the IRT testing algorithm and how it works (we have just completed its adoption for the GMAT Club tests but have not released it in the wild yet), a lot of the question picking is based on constraints and probabilities - there is a randomness element baked into the test not so much to make it unpredictable and wild for the test-takers but the theory's point is that randomness creates opportunities for a greater extent of testing than what the test creator even had in mind so to speak (I am heavily paraphrasing here; if this makes no sense, you may want to look the theory up on Wikipedia or elsewhere). Anyway, there are 10+ constraints for each question you are served - e.g. topic (did you get this topic already or not), ability level (is this accurate level for you), standard error, guessing likelyhood, recency, question overuse, and even more of things I don't know about. All of these depend on some kind of a probability level and the system decides on the fly if you get a hard or an easy question based on the bank of available items and somtimes (very rarely) it happens that even with 20 out of 21 correct the system still does not think you are that awesome (I think there is actually an override at 100% accuracy to give the person Q90 as otherwise it would be a major problem but that's in part why we are not seeing Q87 with 100% correct or something like that).

Anyway, the randomness makes it possible that a small number of tests do indeed come out weird but that's a feature of the test (side effect of the feature) and not necessarily a problem with the overall system. Of course the fact that there are much fewer questions does not help (there has to be at least 4 experimental questions on the Verbal as RC's have 4 questions for some passages and at least 3 on the DI as MSR's have 3 questions; not sure about Quant but I would guess also at least 3 as they need a lot of data points for scoring experimental items - at least 500 and ideally 2,000 or more, which means it can be a bottleneck if the test volume is low).

Anyway, just a long way to agree I guess.

P.S. I have not published this data anywhere and did not think anyone cared enough about it as it is a limited sample and thus most conclusions would always run into the "limited sample" challenge. The main reason I was looking this data up was the scoring research we were doing for new test algorithm.

PPS. I think it is not impossible to think about Q90 or V90 or DI 90 as about the Q51 that was 93rd percentile if I am not mistaken. I don't think the test has gotten that much harder so there is no reason 7% of the population could be scoring Q90 like they were scoring Q51? I think you could get Q51 with 1 incorrect in the past but not sure what you think about looking at it this way?

AjiteshArun · Answer

I completely agree that there will be multiple issues with the sample, but maybe GMAT club could publish a limited number of data points every year, like a year-in-review post? You could let readers know that this is from a sample (only GMAT club-verified scores).

GMAT club may be the only source of this kind of data. I certainly don't expect large test prep companies to publish their actual numbers (even if they do, it'd be hard to trust those numbers).

	Manhattan Prep: $99 Bonus Score Higher on Your GMAT Exam. Guaranteed.
	Target Test Prep - Save up to $599 $400 Discount + $99 Bonus GMAT Club Tests
	e-GMAT: 60% Off on GMAT Focus Get to 735+ from any starting point
	Magoosh increases your score, guaranteed - 15% off ALL Magoosh GMAT plans with code GMATCLUB15 + Free GMAT Club Tests.
	Experts' Global - ~~$1720~~ $960 End-to-End GMAT Prep + App Support: 150+ Videos, 2000+ Questions, 15 FLTs +
	GMATWhiz - $299 ~~$499~~ Experience the course through a free trial.
	GMAT Club Tests - Free Included with every course purchase of $149+ View all deals

GMAT Quant 83 with Only 2 Mistakes

Prep Toolkit

Top 5 GMAT Debrief Videos

615 to 715 in 15 Days (Ria's Strategies)

More than Quant/Verbal Abilities: Speed vs. Accuracy

745 with Only Self-Prep and GMAT Club

From 555 to 765: 210-point Improvement

Perfect 805 Score Debrief by Julia

My Rewards