GMAT quant scores

Inflated GMAT scores... and deflated quant percentiles

  In the most-commented GMAT blog post I’ve ever written, I argued that the so-called “80th percentile rule” – the idea that top-tier MBA applicants need to score above the 80th percentile on both the GMAT quant and the GMAT verbal section – is mostly a myth.

Why? Well, the unfortunate truth is that GMAT quant scores have increased dramatically in recent years, and you’d now need to earn a 50 (!) on the GMAT quant section to score above the 80th percentile, according to the most recent GMAT data. Here, check this out:

From Retrieved October 3, 2016.

Plenty of things are alarming here. When the GMAT score scale was originally designed, the average quant score was supposed to be 30. Now a quant score of 30 is just the 20th percentile (!!), and the mean GMAT quant score is a 39 – a whopping nine points higher than it “should be” if the GMAT score distribution were actually a bell curve, as originally intended.

Even worse: if you want to score above the 80th percentile, you’ll need a 50 on the quant section. Since a 50 is one heck of a high GMAT score (a 51, of course, is a perfect GMAT quant score), I would still argue that no MBA admissions person in their right minds should actually expect it of you. So if anything, the “80th percentile rule” is more of a myth than ever.

But here’s the part of my original GMAT blog post that may not be true anymore, just a few years later:

A 710/47Q/40V is still an outstanding score that eliminates all rational doubt about your academic abilities. For the vast majority of MBA applicants, a GMAT quant score in the 73rd percentile is enough to placate MBA admissions officers, and your odds of admission will depend almost entirely on other elements of your profile.

Let’s pull this apart a little bit. Would a 710/47Q/40V eliminate all rational doubt about your academic abilities? Yes, definitely. If you can get a GMAT score that high, there’s no good reason to question your ability to handle MBA-level coursework. But is that GMAT score enough to “placate MBA admissions officers”? Probably not anymore, depending on which MBA programs you're applying to -- and, of course, on the rest of your MBA profile.

Here’s the problem: GMAT score averages for top-tier MBA programs have continued to shoot through the roof in the past few years. At Kellogg, for example, average GMAT scores have risen from 712 to 728 in just six years. Stanford’s incoming class of 2018 had an average GMAT score of 737.

I’m sorry, but those GMAT scores are ridiculous. Clearly, the GMAT arms race has gotten absolutely out of control – and I say that as a test-prep tutor who profits from that very same GMAT arms race. It’s as if MBA admissions committees have zero understanding of how standardized tests such as the GMAT actually work: the GMAT’s creators would tell you that the difference between a 710 and a 740 may be nothing more than measurement error.

And that’s before we get into the fact that there’s little evidence that the GMAT is predictive of post-MBA success. But that's another topic entirely.

To be fair, I can’t blame MBA admissions committees for allowing GMAT score averages to spiral upward. Frankly, it’s much quicker and easier for admissions committees to whittle down a stack of applications based on GMAT scores than it is to ponder the non-quantitative aspects of each MBA applicant’s profile. Plus, MBA programs are forced to play the rankings game – and average GMAT scores are part of those MBA rankings, whether we like it or not.

Unsurprisingly, MBA admissions consultants are advising their clients to chase GMAT scores in the mid-700s – and applicants to elite MBA programs are working harder than ever to raise their GMAT scores. It’s a vicious cycle, and I’d bet that GMAT scores will continue to creep upward.

In any case, if you’re reading this little GMAT blog, you probably can’t do a damned thing about the way things work: you probably just want to get into a good MBA program. So what should you do if an MBA admissions consultant – or worse, an MBA admissions committee member – tells you that you need a GMAT score in the mid-700s?

Well, take solace in the fact that you’re not alone. And if you think it’s unreasonable that you might need a 750 on the GMAT exam, take solace in the fact that you’re probably right.

And then join the rest of the herd. If you have a profile that will allow you to get away with a lower GMAT score, then consider yourself lucky. But if you’re like most people, all you can do is work as hard – and as smart – as you can for as long as you can.

And if you’re already scoring in the low-to-mid 700s but don’t have time to continue studying, maybe it’s worth taking an extra couple of shots at the GMAT, just to see if dumb luck (or measurement error, if you want to be technical about it) might hand you an extra few points next time around. And if you're a really long way from the GMAT score you want, then maybe it's time to think about the GRE -- but that's another topic entirely.

Then, please graduate from your top MBA program, become a bigshot in the MBA world, and change the way that the GMAT is handled by MBA admissions committees – because I think that the GMAT arms race encourages even more great business talents to slip through the MBA cracks.


GMAT percentile rankings, part II

As a private GMAT tutor, I regularly receive calls from students (or potential students) who are nervous about some sort of "imbalance" between their verbal and quant scores on the GMAT.  In many cases, those worries are absolutely reasonable--if you have, say, a 51Q/30V, you clearly have an issue.

In many cases, however, the imbalance might not be quite as bad as it seems, especially if you're (overly) focused on percentile rankings.  Over the years, I've met quite a few people with wonderful GMAT scores (48Q/48V, 44Q/42V, 44Q/49V) who worried that they have an imbalance because their quant percentile rankings are much, much lower than their verbal percentile score.  In many of these cases, I don't think that the test-taker has much to worry about.

In a previous post about percentile rankings, I mentioned that a large percentage of GMAT test-takers do extremely well (raw score of 47 and above) on the quantitative section, but not so well on the verbal.  I admiringly call this the "Asian effect," since I'm convinced that the bulk of these GMAT quant studs come from math-intensive education systems.  (The United States, of course, is one of the world's worst wealthy nations when it comes to teaching K-12 mathematics.  You should never hire an American GMAT tutor... crap, wait a minute... scratch that last part.)

Anyway, I clumsily punched some GMAT data into an excel spreadsheet, and made a little chart out of it.  The chart shows the rough shape of the GMAT verbal score distribution (approximately normal or bell-curved), the composite GMAT score distribution (also approximately normal or bell-curved), and the GMAT math score distribution (not so normal).

Please keep the following disclaimers in mind:

Disclaimer #1:  This data is extrapolated from a GMAT score report.  It is definitely NOT very precise data.  I also tinkered with a few numbers to smooth out the curves, so take all of this with a grain of salt.

Disclaimer #2:  GMAT section scores and GMAT composite scores don't really belong on the same axis, since you can't easily convert from raw scores (0-60 scale) to a composite scale (200-800) unless you know the GMAT algorithm... and even then, they still wouldn't really belong on the same axis.  Again, this is just a rough visual representation.

Okay, enough prefaces--I think you'll get my point as soon as you see the graph.  Here it is.  Enjoy.

why you shouldn't worry (too much) about percentile rankings

I’ve already met some brilliant MBA candidates in my first few weeks here in New York, but the most gifted GMAT student I ever tutored was probably a young woman I met in DC, before my Manhattan days.  She was a cum laude graduate of Duke University’s economics program, and she had already earned a ridiculously high score on the GRE.  She was solidly scoring in the 700s on her GMAT practice tests, and just wanted help getting her math score “as close to perfect as possible.” I’ll be honest:  there’s a pretty good chance that this particular GMAT student (let’s call her Susan) was smarter than I am, and that’s a beautiful and intimidating situation to be in as a GMAT tutor.  I suspected (correctly) from the start that she didn’t have any fundamental GMAT math weaknesses; all I could offer was some practice questions designed to help her make connections on the hardest GMAT math questions.  (Basically, math formulas are of limited value on hard GMAT questions, and you need to be able to "see" a connection before you'll have any prayer of getting the right answer.  I'll expand on this in a future blog post.)

Basically, tutoring Susan meant spending two hours inventing the hardest GMAT-style questions I could possibly come up with.  I loved it, and I probably learned more from the experience than she did.  Occasionally, I’d manage to stump her, and that was great.  After about eight or ten GMAT tutoring sessions, we decided that she had probably “worked her GMAT math muscles” as much as was reasonable, and she went to take the test.

And she was bitterly disappointed when she looked at her percentile ranking on the GMAT.  She scored in the 99th percentile on the verbal, and her composite score was also in the 99th percentile.  Her quantitative score, however, was only in the 84th percentile.  How could this be true?  She scored a 760 on the GMAT, but somehow wasn't even in the top 10% in math? WTF?

If you have any experience with the GMAT, you surely realize that a score of 48 on the quant section is pretty darned impressive.  (Frankly, anything above a 45 should, in theory, be good enough to allay any legitimate fears that an MBA admissions committee might have about your quant skills, but that's another topic entirely.)  For whatever reason, however, there are lots of impressive math whizzes taking the GMAT.  A full 20% of GMAT test-takers earn a score between 47 and 51.  The score distribution, then, is not a bell curve at all for the quantitative section--it basically looks like a slowly increasing function, with an extra little jump at the end.

I admiringly call this the "Asian effect."  Based on anecdotal evidence, I'm convinced that most of the people who achieve quantitative scores in this range are not products of the United States education system.  (News flash:  by global standards, we Americans are generally pretty lousy at math.)  Basically, there are a ton of people in Asia who demolish the quantitative section of the GMAT, and they make a whole lot of GMAT-takers feel inadequate when they see their percentile score.

Interestingly, the GMAT verbal score distribution is almost a perfect bell curve.  Similarly, the GMAT composite score distribution is also a fairly non-skewed bell curve.  This suggests that most of the GMAT "quant ninjas" are not particularly great at verbal.

The bottom line?  If you're trying to get into top MBA programs, it's probably OK to have a math score "only" in, say, the 70th or 75th percentile.  Don't let the percentile score mess with you:  if your raw GMAT quant score is well into the 40s, it probably won't, by itself, sabotage your MBA dreams.