GMAT Prep vs MGMAT scores

This topic has expert replies
Senior | Next Rank: 100 Posts
Posts: 36
Joined: Tue Apr 08, 2008 1:36 pm
Location: Texas

GMAT Prep vs MGMAT scores

by czechchamp » Tue Jul 01, 2008 7:48 pm
I've heard (read) that the MGMAT CATs are little harder (especially quant section) than the GMAT Prep CATs. Can someone please confirm this?

Here is little background info:
2+ months ago I scored 570 on the GMAT Prep. I've studied since then a lot and completed MGMAT online 9 week course. Got 600 on my 3rd MGMAT CAT one month ago (half way through the course). I haven't taken CAT since. I feel like my knowledge of the material (quant and verb) has increased tremendously (thanks to the online course) and I'm getting ready to take the remaining MGMAT CATs that I have available and the second GMAT Prep soon. I'll be taking the real test in 3 weeks. Anyways, yesterday and today I went through The GMAT Prep review questions for PS, DS and SC and I did really well on them. It wasn't hard at all. But I find the MGMAT CATs and their quant section much tougher.

So is the GMAT Prep (and therefore the real thing) little easier compared to the MGMAT CATs? In particular the math...?

Thank you

User avatar
Master | Next Rank: 500 Posts
Posts: 418
Joined: Mon Feb 26, 2007 6:41 pm
Thanked: 24 times

by Prasanna » Thu Jul 03, 2008 12:47 pm
I would agree with your observation. During my prep I have found MGMAT quant much difficult than GMAT prep or the actual test. MGMAT tests teach you how to make a eductated guess on difficult questions or even to just chose an option and go ahead. I would also like to add that the actual GMAT questions can get difficult at higher scoring levels.

Senior | Next Rank: 100 Posts
Posts: 36
Joined: Tue Apr 08, 2008 1:36 pm
Location: Texas

by czechchamp » Sat Jul 05, 2008 12:14 pm
Thanks for your reply. I took GMATPrep yesterday and got 640 (Q43, V35). I haven't taken CAT in past 5 weeks and I totally ran out of time on both sections. I had to guess randomly and because of that I missed last 5-6 questions in each section. So with better time management I should definitely improve. But yes I do feel that the MGMAT CATs are little more challenging (quant in particular) compared to GMATPrep. However, this is not a bad thing.
Will see.
Test date: 7/21.
Shooting for high 600s.

GMAT/MBA Expert

User avatar
GMAT Instructor
Posts: 2228
Joined: Wed Dec 27, 2006 3:28 pm
Location: Montreal, Canada
Thanked: 639 times
Followed by:694 members
GMAT Score:780

by Stacey Koprince » Tue Jul 08, 2008 6:17 pm
A decent number of my students tell me they think our quant section is harder (though some say it isn't harder!). I think there are two main reasons:

1) the real test has experimental questions, which can be assigned at any level, and ours doesn't. This means that, the higher you go, the more chance you'll get an experimental that is below your level - so it feels like a little "mental break" during the test.

2) some of our problems are a bit too computation intensive compared to what the real test is like. We're weeding these out as we identify them, but we haven't identified them all yet. (And, when we write new questions, it's kind of a judgment call - we don't know for sure until we get a few hundred people to take the question and look at the data!)

For both of those reasons, we're working on adding experimental capability to our tests - but that takes time (a lot, actually - it's pretty complicated).
Please note: I do not use the Private Messaging system! I will not see any PMs that you send to me!!

Stacey Koprince
GMAT Instructor
Director of Online Community
Manhattan GMAT

Contributor to Beat The GMAT!

Learn more about me

Newbie | Next Rank: 10 Posts
Posts: 8
Joined: Sat Sep 13, 2008 8:15 am

GMATprep versus MGMAT

by goncaljj » Fri Nov 21, 2008 12:48 pm
I don't know what the "adaptation" to scores are in the MGMAT; I just took the online version and I got an array of 700-800 difficulty level (based on final report) questions even though I got them wrong back to back to back to.....to back. Some of these questions need +5 minutes even if you know in advance how to solve them. Needles to say that in 75 minutes I only answered 31 of 37.

GMAT/MBA Expert

User avatar
GMAT Instructor
Posts: 2228
Joined: Wed Dec 27, 2006 3:28 pm
Location: Montreal, Canada
Thanked: 639 times
Followed by:694 members
GMAT Score:780

by Stacey Koprince » Fri Nov 21, 2008 12:54 pm
Although you see a score assigned only in a 100-point range, the questions are actually assigned difficulties in much smaller increments - 10 points. So you could get a 740 wrong and then get a 720 level question and get that one wrong and then get a 700... and they'd all say "700 - 800" on your score report.

It's also the case, by the way, that adaptive tests factor in many other variables besides just the difficulty level. Different question types and sub-types, different content areas, etc.

Finally, although the general rule is "get something right, get a harder question; get something wrong, get an easier question," that is not necessarily the linear progression through the test. There are a lot of other things going on in there.
Please note: I do not use the Private Messaging system! I will not see any PMs that you send to me!!

Stacey Koprince
GMAT Instructor
Director of Online Community
Manhattan GMAT

Contributor to Beat The GMAT!

Learn more about me

User avatar
Senior | Next Rank: 100 Posts
Posts: 70
Joined: Wed Jun 25, 2008 8:33 pm
Thanked: 3 times

by axat » Wed Jun 10, 2009 10:15 am
I am bringing this back to life. Let us try and establish a statistic relationship between the scoring model of the MGMAT and GmatPrep and the GMAT.

I haven't written the GMAT yet, but I got a 720(Q49, V39) on the GMAT prep, and a 650(Q 40, V 38) on the MGMAT 1.

I felt the math questions were extremely difficult, even to the extent of being irrelevant as far as the preparation for GMAT is concerned. This unexpected change in the difficulty caused quite a chaos, and I had to guess on the last 6 questions.

I feel that if we have enough people posting their scores for the MGMAT, GMATPrep and the GMAT, we will be able to establish a relativity model.

So please do post your scores. Thanks.

GMAT/MBA Expert

User avatar
GMAT Instructor
Posts: 2228
Joined: Wed Dec 27, 2006 3:28 pm
Location: Montreal, Canada
Thanked: 639 times
Followed by:694 members
GMAT Score:780

by Stacey Koprince » Wed Jun 10, 2009 11:24 am
We actually know this data already.

For the official test, GMAC has published a "within student" standard deviation of 30 points, and GMATPrep is supposed to be about the same. (Though there are intangibles - such as the fact that we know when taking GMATPrep that it's not the real test, which could impact nerves, how we take it, etc.) A "within student" SD means that if the student takes the test a bunch of times in a row (without studying / learning more in between), the standard deviation of that student's scores will be about 30.

We (MGMAT) have done extensive analysis of our large pool of students and their practice and official test scores.

- The "within student" SD is about 50 points on our tests. The "last MGMAT test to official test" SD is also about 50 points.

- Every 2 months, we calibrate our scoring algorithm to ensure that our test, on average (across all students now) is an unbiased estimator of GMAT performance. What this means: some students do better on the official test and some students do worse (compared to last MGMAT test). When we look at performance across all of our students, we attempt to ensure that the average difference balances out to zero. In practice, we ensure that the average difference in last MGMAT test score vs. official test score is between +5 and -5 points across all students. What this means: our tests do not consistently under- or over-estimate scores.
Please note: I do not use the Private Messaging system! I will not see any PMs that you send to me!!

Stacey Koprince
GMAT Instructor
Director of Online Community
Manhattan GMAT

Contributor to Beat The GMAT!

Learn more about me

Master | Next Rank: 500 Posts
Posts: 124
Joined: Wed Apr 30, 2008 10:33 pm
Location: Los Angeles
Thanked: 5 times

by skang357 » Wed Jun 10, 2009 11:36 am
So when somebody buys your MGMAT CATs online, and they have a 1 year access. The questions and the scoring calibrations on the tests are not static and constantly evolving?
Impossible is nothing

Master | Next Rank: 500 Posts
Posts: 124
Joined: Wed Apr 30, 2008 10:33 pm
Location: Los Angeles
Thanked: 5 times

by skang357 » Wed Jun 10, 2009 11:39 am
By the way, I read somewhere that the GMAT contains 18 or 19 experimental questions on each section. Yeah that's right, half the test doesn't count towards your score. Is there any validity to this? Or does this explain the very fact that you hear people saying all the time that they felt like they bombed the Quant section and then receive a 49. I have seen people say they missed 14 or 15 questions on the Q section on GMATPrep and say they got a 49. (They just happened to miss all the ones that don't count maybe?)
Impossible is nothing

GMAT/MBA Expert

User avatar
GMAT Instructor
Posts: 2228
Joined: Wed Dec 27, 2006 3:28 pm
Location: Montreal, Canada
Thanked: 639 times
Followed by:694 members
GMAT Score:780

by Stacey Koprince » Wed Jun 10, 2009 11:50 am
We don't know for sure (because the info isn't published), but the best estimate is that the official test includes about 10 experimentals per section. Still a hefty 25%, but not 50%.

Also, the main reason people think they bombed a section but then get a good score has much more to do with the fact that, the better you do, the harder the questions you're going to get. So if you do an amazing job, you are going to see some serious mind-benders on the test. That makes people feel like they're not doing well - but what people forget is that even the high scorers get a lot of (counted) questions wrong. If I'm going for a 95th percentile score and the test offers me 5 (counted) questions above that level and I get them all wrong, that's not going to prevent me from getting that 95th percentile score. (Unless, of course, I spend way too much time getting those 5 brain-benders wrong, and then cost myself points at the end of the test because I run out of time...)

Experimentals can, of course, play into this (the feeling that you bombed when you didn't), but because experimentals are assigned regardless of difficulty level, if you are scoring 95+ percentile, most of your experimentals are likely to be below your level - which means (hopefully!) you're getting most of them right.
Please note: I do not use the Private Messaging system! I will not see any PMs that you send to me!!

Stacey Koprince
GMAT Instructor
Director of Online Community
Manhattan GMAT

Contributor to Beat The GMAT!

Learn more about me

Master | Next Rank: 500 Posts
Posts: 124
Joined: Wed Apr 30, 2008 10:33 pm
Location: Los Angeles
Thanked: 5 times

by skang357 » Wed Jun 10, 2009 3:49 pm
Stacey Koprince wrote:We don't know for sure (because the info isn't published), but the best estimate is that the official test includes about 10 experimentals per section. Still a hefty 25%, but not 50%.

Also, the main reason people think they bombed a section but then get a good score has much more to do with the fact that, the better you do, the harder the questions you're going to get. So if you do an amazing job, you are going to see some serious mind-benders on the test. That makes people feel like they're not doing well - but what people forget is that even the high scorers get a lot of (counted) questions wrong. If I'm going for a 95th percentile score and the test offers me 5 (counted) questions above that level and I get them all wrong, that's not going to prevent me from getting that 95th percentile score. (Unless, of course, I spend way too much time getting those 5 brain-benders wrong, and then cost myself points at the end of the test because I run out of time...)

Experimentals can, of course, play into this (the feeling that you bombed when you didn't), but because experimentals are assigned regardless of difficulty level, if you are scoring 95+ percentile, most of your experimentals are likely to be below your level - which means (hopefully!) you're getting most of them right.
So you're saying they don't give too many experimentals above the 95+% level.

Yeah you were right about the 10 questions being experimental on each section. The actual number is 9. So 28 questions will be counted towards the score. Which explains why people miss 14 or 15 questions and get a high 40s score.

I actually get the feeling that in V section, there are very little experimentals (at least far less than the Q). That's because in order to get a 49 or 50, I know you have to almost get no wrong, which is not the case in the Q.
Impossible is nothing

GMAT/MBA Expert

User avatar
GMAT Instructor
Posts: 2228
Joined: Wed Dec 27, 2006 3:28 pm
Location: Montreal, Canada
Thanked: 639 times
Followed by:694 members
GMAT Score:780

by Stacey Koprince » Wed Jun 10, 2009 3:53 pm
Technically, the experimentals don't have a level - part of the reason they are testing these questions is to determine what level to assign to each question. So, it's not that they don't give many 95+ percentile questions on purpose. It's just that, if the questions can span the entire range, then by definition, you have a smaller chance of getting a question that will eventually be rated 95+ than a question that will eventually be rated 94-. Most of the questions are going to be placed in the latter category.
Please note: I do not use the Private Messaging system! I will not see any PMs that you send to me!!

Stacey Koprince
GMAT Instructor
Director of Online Community
Manhattan GMAT

Contributor to Beat The GMAT!

Learn more about me

Master | Next Rank: 500 Posts
Posts: 124
Joined: Wed Apr 30, 2008 10:33 pm
Location: Los Angeles
Thanked: 5 times

by skang357 » Wed Jun 10, 2009 8:13 pm
Stacey Koprince wrote:Technically, the experimentals don't have a level - part of the reason they are testing these questions is to determine what level to assign to each question. So, it's not that they don't give many 95+ percentile questions on purpose. It's just that, if the questions can span the entire range, then by definition, you have a smaller chance of getting a question that will eventually be rated 95+ than a question that will eventually be rated 94-. Most of the questions are going to be placed in the latter category.
Thanks Stacey, that makes a lot of sense.

I was wondering... how are your 4th edition Strategy Guides different from the 3rd? Are there significant changes?
Impossible is nothing

GMAT/MBA Expert

User avatar
GMAT Instructor
Posts: 2228
Joined: Wed Dec 27, 2006 3:28 pm
Location: Montreal, Canada
Thanked: 639 times
Followed by:694 members
GMAT Score:780

by Stacey Koprince » Thu Jun 11, 2009 6:31 am
Two big differences (but not that significant, ultimately)

(1) 4th edition matches OG12; 3rd edition matches OG11. That is, the problem sets for each are geared toward each respective OG book.

(2) 4th edition splits the "regular" and "advanced" material for each content area into two separate chapters; 3rd edition keeps the "regular" and "advanced" material for each content area in one chapter.

If you've already got the 3rd edition but you bought OG12, look for the OG12 analysis on our web site. It includes a "conversion list" of the problems so that you can get OG12 problem lists based upon the OG11 problem lists in the 3rd edition.
Please note: I do not use the Private Messaging system! I will not see any PMs that you send to me!!

Stacey Koprince
GMAT Instructor
Director of Online Community
Manhattan GMAT

Contributor to Beat The GMAT!

Learn more about me