Question about MGMAT scoring algorithm

Guest

Hi Stacey -

I just did my first MGMAT CAT. Below are the results. I was hoping you could help me answer 2 questions ... I am a bit confused as to how this should be interpreted.

My verbal score is much lower than my qaunt. However, the Verbal percentile is slightly better than the quant. Inuitively, this does not make sense?

TYPE SCORE ESTIMATED PERCENTILE RANK
Quantitative 42 66%
Verbal 32 67%
Total 600 70%

The second question is concerned with my assessment report. (first column is "avg difficulty right answers" - second column is "avg difficulty wrong answers")

Problem Solving 620 640
Data Sufficiency 630 690
Sentence Correction 710 690
Critical Reasoning 710 710
Reading Comprehension 610 680

Given that these ranges are all above 600 .... how can my total test score be 600? I greatly appreciate your help Stacey.

Stacey Koprince · Posted: Wed Nov 14, 2007 11:12 pm Post subject:

On your first question, math and verbal use the same numbers but not the same scale (who knows why - that's from the makers of the test). So the same numerical 2-digit score will not correspond to the same percentile in math and verbal.

Re: your second question, I'm not actually sure - I'll have to ask our curriculum director how that part of the algorithm works and get back to you.

LSB · Posted: Thu Nov 15, 2007 5:21 am Post subject:

Thanks Stacey - One thing which may be relevant on question 2 is that I did not have enough time to answer the last Q in verbal. It was a 700-800 SC question.

Not sure how big the penalty for this may be .. but I thought since you mentioned you were going to ask somebody this may be relevant.

I really appreciate your help Stacey.

Stacey Koprince · Posted: Thu Nov 15, 2007 12:06 pm Post subject:

ah, okay - the penalty is quite severe actually - about 4 percentile points per blank question. It might be useful too if you PM me your full name and email address in our system so our algorithm guys can go look up your actual test.

Stacey Koprince · Posted: Thu Nov 15, 2007 4:36 pm Post subject:

Okay, I've talked to both of our algorithm gurus and now know how to explain what's going on!

First, the score is what's called "path-dependent" so it actually matters how you're doing towards the end of each section. Your overall averages for the entire section could be higher (or lower) than the averages for the last 5 or 10 questions, when the test is honing in on your score. The overall test score is not reflective of the overall difficulty levels on the whole exam - it's reflective of how you were performing at the end of the test.

Another way to think about it is this: the test sets the score based upon the level at which you can answer about 50-60% of the questions correctly. If you are given a bunch of 600-level questions and you can only answer 30% correctly, your score will be below 600. Conversely, if you can answer 80% correctly, your score will be above 600. (Though this is not actually what happens literally.) The test could, for example, be giving you 650 level questions towards the end based upon your earlier performance, but then you start running out of time (which sounds like this happened, since you had to leave one blank), and so you start getting a lot of those questions wrong towards the end... and then that brings your score down (plus the additional ding in verbal since you left one blank).

night · Posted: Sun Feb 17, 2008 2:09 pm Post subject:

thank you for great explanation! ,Stacey

gmat765 · Posted: Fri Apr 18, 2008 2:43 am Post subject:

Thanks Stacey