To Guess or Not To Guess on the GMAT?


A few weeks back we blogged about the latest news from GMAC as presented during their Test Prep Summit. One of the topics of discussion involved how unanswered questions at the end of the Quant or Verbal Section negatively impacted scores. The bottom line was this: the more questions you left unanswered at the end of the section, the lower your score was.

Turns out, there’s just a little bit more to the story. It’s important to take note of so you can maximize your score by understanding the best approach for you, and how end-of-test outcomes affect overall results.

I recently reviewed a GMAC-published research report (Talento-Miller and Guo) that closely examined the effect of guessing vs. omission on computer adaptive tests. For those of you with an aversion to statistics and tables and charts, this wasn’t a document you’d want to spend much time with. I battled through, so here’s the low-down from the research:

The Data

The GMAC researchers took 135,000 GMAT test results from an 8 month period and assessed them to determine the effects of guessing on or omitting from 1 to 5 questions at the end of the Quant and Verbal sections. They assessed these results along a number of paradigms, the most intriguing being a distinction between effects of guessing and omission according to estimated ability of the test-taker (more on this in a moment).

What They Found

*Verbal- There’s almost no observable difference in scores on the verbal section between those who guessed on some or all of the last five questions and those who simply didn’t enter an answer.

chart-1

(c) GMAC – 2009 “Guess What?” report

* Quantitative – There is a marked advantage in guessing over omission when one is going to guess on 3 or more questions at the end of the test.

chart-2(c) GMAC – 2009 “Guess What?” report

What Does This Tell Us?

1) Generally, as we’ve mentioned before, it’s better to guess than leave things unanswered. Consequently, when determining and practicing your testing approach and pacing you want to always ensure that you’re comfortable leaving yourself enough time at the end of each section to enter an answer for all remaining questions that you won’t have time to work.

2) More importantly, for test-takers who are currently at a pace that will leave them with 5 or more questions unanswered at the end of the Quantitative, it is absolutely imperative that you enter an answer for every question you won’t have time to work. The research shows that for those test-takers who guessed or omitted 5 or more Quant questions, more than 1/3 of those who guessed had average subscores of 3 or more points higher than those who omitted 5 or more questions. 3 points may not seem like much, but depending on where that 3 point difference occurs on the bell curve these 3 points (or more) could mean 10 percentile points or more. Besides, they’re more or less free, so why wouldn’t you make sure you take them?

A Closer Look

The study also revealed some interesting trends on guessing and omission for test takers of differing ability estimates. Basically, a CAT test works by trying to determine or estimate a test-taker’s ability and then consistently asking questions of a difficulty commensurate with that ability. Depending on how a test-taker answers the question (right or wrong) the estimated ability is refined and the subsequent question difficulty generally reflects this revision.

The research examined the effect of guessing vs. omission for low, medium, and high ability estimated test-takers. I won’t bore you here with all that they found, but there were three things of note:

1. For test-takers of lower ability, the numbers indicate that it may be more beneficial to omit than guess, though the benefits are usually pretty small.

2. For medium ability test-takers, the results are mixed so stick to incorporating time to guess at the end of each section.

3. For high ability test takers, the results indicate that under NO CIRCUMSTANCES should you omit 3 or more questions. Higher ability test takers experienced the greatest negative impact to scores for not answering three or more questions in both Quant and Verbal Sections. The bottom line, as if it needed to be said again: Don’t Leave Questions Blank.

chart-3

(c) GMAC – 2009 “Guess What?” report

Ultimately, the thing to remember is that your pacing plan for the real test should be based on how you’ve been pacing through your practice tests. You need to have a good idea of roughly how many questions you’re confortable doing in the 75 minutes allotted for each section and strive for that. Also, you’ll need to incorporate enough time into this plan to enter an answer for all questions you won’t be working on.

  • (646) 414-1586
CONNECT WITH US
COPYRIGHT ©2002 - 2014 BELL CURVES LLC. ALL RIGHTS RESERVED. All tests names are registered trademarks of the respective testing companies, which do not endorse and are not affiliated with Bell Curves.
BELL CURVES - 25 West 36th Street Street, 8th Floor - New York, NY 10018 Bell Curves is an educational services and test preparation company. It delivers high-quality consulting services, test preparation programs, and self-study resources to students throughout the country.
Equal Opportunity Employer - Privacy Policy - Refund Policy