Ask the Test Maker

Hi,

Regarding the Quantitative and Verbal scaled scores, I've noticed glaring inconsistencies within GMAC's own publications.

Many of GMAC's materials -- and some official posts here (by Rebecca, username "OfficialGMAT") -- claim a scale of 0-60 for each section. However, it's fairly clear that the actual scale is not 0-60.
Consider:

* Page 830 of the GMAC Official Guide (13th Edition) shows a scale of 6-51 for each of the two sections.

* In this official blog post, GMAC's Larry Rudner states a range of 11-51 for each section.
https://www.gmac.com/why-gmac/gmac-news/ ... cores.aspx

* My own official GMAT score is Q51 V51 800.

* In several years of working with the GMAT and its students, I have never encountered a score of 52+ on either section.

* Getting every single question correct in the GMATPrep software yields a score of Q51 V51 800 (not "Q60 V60").

So, at this point, it seems plain that the top of the scale is not 60 but 51. I still have two questions, though, which are no doubt of interest to almost everyone here.

1/
Why so much internal inconsistency about such a straightforward issue?
I could understand if a certain amount of inconsistency were to surround a more complicated issue, such as the adaptive algorithm.
On the other hand, this striking variation among official, on-the-record statements -- about something as simple as the max and min of the scoring scale -- does tarnish the credibility of an otherwise excellently crafted test.

2/
Could you please tell us, once and for all -- what are the actual max and min values of the Q and V scales?

Thank you for your time.

Ron

Hi, I'm sorry to belabor the point here, but I do find all this rather disturbing. This post will hopefully be my last on this topic, but all of this is still far from clear.

To wit:

OfficialGMAT wrote:a person could theoretically score above a 51

Let's look at the math. According to the percentile scales in the 13th edition Official Guide, a Quant score of 51 lies at the 98th percentile, meaning that approx. 2% of the test population scored at this level.
The given sample size is 787,205, so approximately 2% of that total -- about 15,744 people -- scored Q51 in that recent three-year sample.

In other words, for that three-year sample, we've got about 16,000 test takers with a Quant score of 51 -- and none with a Quant score of 52 or more.
If, indeed, a score of 52+ were possible, this result would be statistically impossible. (Imagine if, say, 16,000 people were to throw a football 51 yards, but no one ever threw it 52 yards. In that case, we could be pretty well sure there was a giant wall 51 yards away from us.)

There's also Mr. Rudner's reference to "the 11-51 scales" in the blog post cited above, unambiguously implying that the scales themselves (and not the scores) have that range. And the fact that GMATPrep gives a score of 51 (on either section) if you get every test item correct.

So, in closing, four questions.

1/
Even though ~16,000 students (in the three-year sample) scored Q51, and no one whatsoever scored Q52+, do you still claim that a score of Q52+ is possible?

2/
If a student were to get every problem on the official GMAT exam correct, what score (Q/V/overall) would he or she earn?

3/
A score of Q51 V51 translates to an overall score of 800. So, certainly, these theoretical 52-60 scores should give composite scores above 800.
If the verbal and quant scores are indeed 0-60, what is the actual upper bound of the overall GMAT score?

4/
Is there any way Mr. Rudner could take a moment to weigh in on the issue?

Thank you for your time.

OfficialGMAT wrote:Consistently, a Quantitative score above 50 and Verbal score above 44 places the test taker well within the 99th percentile.

As David already pointed out above -- and as the Official Guide corroborates -- this isn't true. A 51 is only 98th percentile.

Therefore, as a practical matter, a focus on why the score scale runs to 60 or how a test taker can - or, in fact, if a test taker should - put forward an effort to score above 50 Quantitative or 44 Verbal is not particularly productive or necessarily in the best interest of the test taker.

I agree that this issue has very little utility for the test taker.
What I think is the larger issue here, though, is the trustworthiness of the GMAC organization -- which is definitely tarnished by this ongoing inconsistency and evasiveness.

I don't think it's too much to ask of GMAC to fill in the following two blanks:
The highest Q or V score that a test taker can actually get is ______.
The lowest Q or V score that a test taker can actually get is ______.

These should be very simple and straightforward. If GMAC can't provide solid numbers, the credibility of the entire GMAT enterprise is compromised.

Since so much trust must already be placed in the GMAC organization -- for instance, the GMAT is one of very few tests on which test takers never see their actual test items or specific results -- this issue takes on even greater importance.

What are the numbers?

Thank you.

Seconded! I've always wondered!

-t

P.S. Follow-up question. A higher raw score in verbal seems to result in a better overall score than a higher raw score in quant (i.e. 42Q 46V is slightly better than 46Q 42V). Am I actually right about this, and if so, why is it? Is it fair to extrapolate that the GMAT is weighted slightly heavier on the verbal side?

Tommy Wallach wrote:A higher raw score in verbal seems to result in a better overall score than a higher raw score in quant (i.e. 42Q 46V is slightly better than 46Q 42V). Am I actually right about this, and if so, why is it? Is it fair to extrapolate that the GMAT is weighted slightly heavier on the verbal side?

well, i hope they answer my question, but at least i can tell you this much isn't true.

the best way to collect data for this question is just to search the internet for people's performances (e.g., "gmat q46 v42". if you do that, you'll find that every possible combination of Q and V subscores gives the same range of composite scores. both q46 v42 and q42 v46 are translate to mostly 720's and some 710's.
you'll find that the same is true for any other combination of subscores, too.

there will be some small variation -- 10 points, max 20 -- within any one fixed Q/V performance, because the Q and V scores inside the test software are not integers; all existing adaptive software produces decimal scores, which are then rounded to produce whole numbers. so, if your scores inside the computer are q41.7 v45.6 and someone else's are q42.3 v46.2, then your "q42 v46" will be lower overall than theirs. if you are seeing phantom variation, it may come from there.

Hi, Rebecca.

OfficialGMAT wrote:Ron, as I explained to you in my posts above, the score scales for quant and verbal range from 0-60, in 1-point increments. Quantitative scores below 7 and above 50 and Verbal scores below 9 and above 44 are rare. Quantitative score above 50 and Verbal score above 44 places the test taker well within the 99th percentile.

Rebecca

As already remarked, the boldface part is false. According to the Official Guide, a Quant score of 51 is only at the 98th percentile.

So...

* Can a test taker actually RECEIVE a score above 51 in either section? If "yes", then why has it never happened in millions of attempts?

* If such scores are indeed possible, then what Quant score is actually "well within the 99th percentile"?

Thank you for your time.

R