GMAT Strategy

Hello all. I have seen some talk about this here before, but would be curious to see how you all have fared in Knewton's practice tests versus other practice or the real GMAT. I am finding the scoring to vary hugely from one day to another. For example in a span of 2 weeks, I took 4 Knewton CATs and received the following scores (in this order):

730 Q45 V46 (Only 2 wrong in Quant!)
710 Q46 V41
680 Q40 V45
640 Q45 V33 (Only 6 verbal incorrect - got first 9 in a row correct and last 18 correct too)

That's right, my score has actually decreased drastically- However these scores look INCREDIBLY inconsistent and irregular. Moreover, I also took the GMAT Prep test last week and scored a 710 (Q47 V41).

I felt that I had gained significant knowledge since my first real GMAT (670- Q43 V39). Basically, this has been a HUGE confidence killer and makes me question my readiness to re-take the GMAT. Does any one put any weight in Knewton's CATs?

Well everyone will have their own opinion on this, and you may even hear from a Knewton rep in this thread. But, personally, I would put near zero weight on third party CAT scores. The GMATPrep scores are, in my opinion, the only scores that can be used as a predictor. Now, of course, this isn't to say that third party CATs have no value. Certainly they do. But their scoring algorithm isn't something I'd trust.

Based on your GMATPrep score vs your GMAT score, it looks like you've indeed improved in both Quant and Verbal. But to have the most impact on your overall score, it looks like bringing up your Verbal even further is the best action plan.

Thanks for your thoughts and suggestions, Chieftang. I would like to get a 46/47 in Quant, which I think is doable, but I agree that stealing a few extra points in verbal will be easier than in Quant.

I also agree that what I saw with MGMAT CATs really bears little resemblance to the real test.

Knewton's questions seem far more test-like, but it seems that they don't have enough challenging questions and you get handicapped to a lower score- for example, 2 wrong on a Quant section should never only be a Q45, although Knewton reps always say "it depends on which questions you get wrong", it seems to me that If I got all but 2 correct, the questions should be getting much harder, but they didn't.

Also, in the last Knewton CAT, it seems almost as if they ran out of 700-800 questions, as the entire test the questions were really easy. Thanks to anyone else who has any opinions about this!

Diagnostic GMAT 43 40 680
Practice GMAT #1 34 42 610
Practice GMAT #2 42 38 650
Practice GMAT #3 42 37 640
Practice GMAT #4 41 41 660

I experienced a steady linear progression with Knewton's CATs, with nearly identical scores on Manhattan GMAT and GMAT Prep.

MGMAT: 750
Knewton Diagnostic: 750
Knewton 1: 760
Knewton 2: 770
GMATPrep 1: 770
Knewton 3: 780


I believe that Knewton's CATs have a theoretically accurate algorithm for computing a GMAT score, but in my experience they don't do a good job at scaling question difficulty to your level. My impression is that making a 'silly' mistake on one easy question on the Knewton could have a much bigger effect on your score than it would on the real GMAT.


I've heard a number of complaints about Knewton's CATs, but rather than dismiss the exams entirely, I think you should use them for what they are (a tool to improve your performance) and not for what they aren't (an accurate gage of what you might score on Test day).

My advice:
1) Look at your Knewton CAT's - how many questions did you get wrong, that you could easily have gotten right?

2) Focus on the reasoning that made you choose the wrong answer. Silly mistake? Missed a piece of information in the prompt? It doesn't matter if you knew the material, it matters that you made a mistake - so focus on why you made a mistake.

3) Knewton CATs rightly penalize you for making mistakes on problems at or below your level. You should be penalized for that. Don't make mistakes on problems that you know how to do.

4) Cover your bases. Make sure you do every almost every problem at or below your level correctly. If you want to do well (700+), it's more important to not make mistakes on easier questions than it is to do well on harder questions.