Skipping Hard Questions Really the Optimal Strategy?

[fawwal09]

Hi Team,

I am having an extremely hard time understanding how the GMAT and its scoring works. By this I mean the decision to guess randomly (basically skipping) on seemingly difficult questions (quant and verbal alike). The general consensus from GMAT experts is to guess on these tough problems but it seems I get penalized for it at the end. I'm currently scoring between 570 and 620 on practice exams (both GMAT prep and mgmat). I learned that when I actually try to take the time to answer tough problems (and as a result rushing towards the end) my raw scores tend to be more favorable than when I employ the "guess randomly" strategy on these same tough questions (and as a result finish the exam on time). I don't understand this.

Additionally, the verbal scoring is even more baffling to me. I'm averaging a 30 on verbal but I'm answering lots of 600-700 and 700-800 level questions - these levels make up about 90% of the section. Granted I'm answering as much wrong as I am answering correctly but I mean these are top tier questions after all so I'd expect to answer a substantial amount incorrectly (not answering incorrectly consecutively either). But yet and still why a low score? The fact that I get all these top level questions one would think he's in the top percentile in scoring.

I guess I'm trying to analyze my approach and devise a strategic way to conquer this exam.

[StaceyKoprince]

As a general rule, spending a lot of extra time on really hard questions usually doesn't pay off for two reasons:
(1) People tend to get the really hard questions wrong anyway, even with extra time, because they're really hard!
(2) If you have to rush on more than about 3 questions at the end of the section, then your score is going to go down, because there is a larger penalty for long strings of wrong answers.

Now, there are some definitions to discuss. First of all, "a lot of" extra time is defined as 1+ min above the average time for that question type (so for quant, eg, 3+ min). Spending 2.5m on a quant question is NOT considered spending too much time, because 2 min is an average, not an upper limit. Most quant questions will (should) be in the 1m to 2.5m range, with maybe a few in the 2.5 to 3m range and maybe a few faster than 1m.

So feel free to spend up to an extra 30 seconds, and even up to an extra 45 to 60 seconds on occasion - no problem. Your time should still average out okay in the end.

It's even okay if you have to rush on the last couple of questions in the section - that's not enough to tank your score. If you're rushing on 5 to 10 questions, though (which we often see), then your score is almost certainly going to go down.

The GMAT is a "where you end is what you get" test, so you can answer harder questions earlier in the section but if your score drops by the end, then where you end is what you get (which may partially explain what's going on in your verbal section). And, by the way, the "outlier" questions (the hardest few and easiest few in your section) don't count as much towards your score.

Someone scoring 85th+ percentile on verbal is going to be seeing almost exclusively 700+ questions. A score of 30 is about the 60th percentile, which is consistent with questions in the 600-700 range. (NOTE: don't use the percentiles associated with the overall 3-digit scores to think about this - the overall score is not just an average of your quant and verbal percentiles.)

The short answer to all of this is: the algorithm is an actual computer algorithm and it practically requires a Master's degree to understand. It's pretty much impossible to just look at the questiom mix or score trajectory and understand how it works. :) If you are still interested in learning (a bit) more about the technical details, I'd recommend reading the Scoring section of our free e-book The GMAT Uncovered. (There's already a copy in your student center.) It doesn't get into the crazy technical details, but it gives you more than you know now (and probably more than you ever wanted to know).

Let me know if you have any questions!