A Possible Strategy for Improving Your Score? Or Maybe Not....

Question

Good day, folks!
Here is something that crossed my mind. Before I delve into that, let me preface it with my understanding of the GMAT FE examination.
 
 The GMAT FE is a Computer based Adaptive Test. It attempts to ascertain the level of an examinee's skills (be it quant, verbal or data interpretation) by throwing questions at them to which they respond to at a 50% accuracy within the question number limit imposed per section. In other words, it judges you as you go by the questions until you hit a plateau of difficulty and your scaled score is determined at the end. 
 Being an adaptive test, having a big enough streak of questions attempted correctly, increases the difficulty level highly especially for the last few questions, as observed by many people. 
 Also, regardless of how one is performing, the section will consist of a set number of questions with medium and hard difficulty. 
 Getting an easy question wrong, penalizes you heavily as compared to a medium or a hard question. 
 You are not penalised for reviewing a question (changing from incorrect to correct) later after completing all questions. The effect on the score is the same as having answered it correctly the first time around.  
Now my approach is this. Just for the sake of an example, let's say a person has a Quant-85 proficiency, is able to attempt easy and medium level questions with complete precision but fumbles on the harder ones. If this person were to purposely bomb three easy questions, say in the first 10 questions, mark them for review later, complete the section with some time remaining, and would later correctly review the said questions.
Wouldn't doing so significantly reduce the difficulty of that section for that person? The appearance of harder questions would be reduced while all the easy and medium level questions attempted would be correct, possibly raising the score.

Given the hypothesized person has the exam in less than a week, is this a fruitful strategy or is his mind devising methods to avoid further practicing quant?
Would love your views on this Gs,  bb   Bunuel   chetan2u   GMATNinja   ScottTargetTestPrep   GMATCoachBen   KarishmaB   AjiteshArun   MartyTargetTestPrep

GMATCoachBen · Answer

Hoehenheim TLDR: your (hypothesized person's) best bet is to truly master the quant skills, instead of trying to game it this way! It's unlikely to work, unless you actually do get every single question correct once you go back and review. Some people have shown that if you get them all correct, you do score 90, even if you change some earlier questions from incorrect to correct. The major downside would be that if you do miss some questions, the average difficulty will be lower, and you will be penalized much more heavily, especially for missing easier questions. For example, I've seen 3 errors result in a Q78 (-4 points per error) on one test with easier questions, vs. 2 errors resulting in Q88 (-1 point per error) on another test with harder questions. I'm curious to see more data on how much the question difficulty adapts under different scenarios. It would be an interesting experiment for someone to try on a practice exam, and see how much the GMAT Club difficulty data for the questions varied, when getting the first 3 wrong and the rest correct, vs. all 20 correct. bb , do you have any current data/experiments on the adaptivity? I believe you posted here with some data that it wasn't as adaptive as expected, but then later retracted that theory here . The difficulty throughout the test, even when scoring very high, varies more than many people think. Below is an example of GMAT Club difficulty data from a DI90 on practice test #1. You can see I got a wide variety of difficulties throughout the test, including 2 easy, 9 medium, and 9 hard. Of the final 11 questions, I got 1 easy, 5 medium, and 5 hard: https://gmatclub.com/forum/download/file.php?id=81926 https://gmatclub.com/forum/download/file.php?id=81927 GMAT-Club-Forum-1j4oejbs.png GMAT-Club-Forum-jj4m5ork.png

bb · Answer

Hi. When the GMAT went off the rails last year and announced they are doing a Focus thing with 3 answer choices, I have spent quite a bit of time researching various papers and theories. Here is what I discovered:

1. Apparently the adaptive testing and modifying answers idea is from quite a long time ago - at least 80's. There are many papers on the models and benefits/issues but here is one from 1996 that references a bunch of 1982 books: https://files.eric.ed.gov/fulltext/ED401301.pdf

2. The model expects you to do exactly what you are outlining - get 3 first questions or easiest questions wrong so that you can come back and correct them. They have done quite a bit of research and have ways of minimizing the impact. E.g. they can use bins of questions - such as a bin of 6 questions where the worst case you can inflate will be 3 out of 6 are wrong.

3. The strategy you came up with even has a name ... I have spent the last 15 mins but not been able to find it unfortunately (I found it when I was doing the research about a year ago).

P.S. Attached is GMAT Focus patent for anyone looking to cure insomnia. Perhaps someone smarter than I can figure and extract smth out of it.

Attachments

GMAC Patent.pdf [983.63 KiB]
Downloaded 71 times

bb · Answer

GMATCoachBen - I made the same chart you put together (except I have seen many of them based on my GMAT prep iterations of taking tests over and over again). I did not see strong adaptability. It was all over. The funny thing is that GMAC took that statement very seriously and we actually had a call about it where they have assured that the test is indeed adaptive. However, what you posted is accurate, it is adaptive but it is not perfect. You can only have so many hard algebra questions or so many easy probability questions and some of those questions are impossible to write at all, so the difficulty does fluctuate quite a bit.

P.S. It is funny you also calculated average difficulty. I have noticed some patterns by the way that were driven by sections (if you got a low score on one of the sections, e,g. sub 74, the next one, the average difficulty would hover around 30 no matter how I tried to answer, just would not give me hard questions. Not sure if you have had the time on your hands to do that). There was definitely a pattern that I spotted but I don't think there is a way to use that strategy to one's advantage really. I also was doing it quite early on when we did not have sufficient answer quantities and difficulty was not quite settled on some of the GMAT Focus questions.... maybe I re-run the scenarios before I make any loud statements about it and they will be proven to be false because it was early on. Only if we had more time

Hoehenheim · Answer

GMATCoachBen   bb  Thanks a lot for addressing as well as entertaining this post! This has been majorly helpful as after articulating and confronting this thought my mind is put at ease and I can focus more on preparation rather than trying to funnel my way through the exam like this.
The algorithm wins this round then...

bb · Answer

It’s not that the algorithm wins, it’s more like it can’t be easily compromised 😂

And yes, the best strategy is focusing on the material. I would say the area that most people underestimate and overlook is managing your time on the test. The test is very harsh towards people who take the strategy of I will try to answer every question and I’m just gonna hope that time will Stop for me or I’m not really sure what’s in their heads. This is a test of discipline.

You can take control of the test by deciding which questions you guess. This will be the hardest questions which will have the least negative impact on your score and therefore the test will reflect the best of your abilities as long as you don’t do a mad dash at the end of the section.

Posted from my mobile device

chetan2u · Answer

BB, I did scan the attachment you added. It is a complex web of info and graphs that would require very meticulous analysis.

Now, onto the strategy of purposefully answering three easy ones wrong and then changing them. If someone gets three easy ones wrong, the test would most likely rate the test taker around 75-80 depending on how one has answered the other questions.

If now you review and get the three easy ones correct, would 75-80 take you to 87-90, I doubt. It could take you up by 2-3 points at the maximum. After all you never answered any question of the 87-90 level.

So, go with the natural flow of test and I am sure you will get more than what some strategy may get you to.

GMATCoachBen · Answer

bb interesting, thanks! So, do you think this data you posted would be likely to replicate? Average difficulty of 46 getting the first 6 wrong, and about the same (44) getting the final 6 wrong? That did seem surprising when I saw this, since it's a significant difference from what we saw in Classic Edition: https://gmatclub.com/forum/gmat-focus-a ... l#p3263967 Please let me know if you do end up taking the time to do more experiments or updated analysis, thanks!

Hoehenheim · Answer

bb   chetan2u  Duly noted! Always appreciate you guys helping us out.

AjiteshArun · Answer

Hi Hoehenheim, You could tag KarishmaB and IanStewart on this one. Both are great at quant, and IanStewart's posts on the algorithm (for the old GMAT) are by far the best on GMAT club. For what it's worth, I think it's better to take the new GMAT like the old one. That is, keep review and edit only for really tough questions or silly mistakes. Finally, if our hypothesized friend is taking the test soon, all the best!

	Magoosh increases your score, guaranteed - 15% off ALL Magoosh GMAT plans with code GMATCLUB15 + Free GMAT Club Tests.
	Manhattan Prep: $79 Bonus Score Higher on Your GMAT Exam. Guaranteed.
	Target Test Prep - Save up to $325 $200 Discount + $79 Bonus GMAT Club Tests
	e-GMAT: 60% Off on GMAT Focus Get to 735+ from any starting point
	Experts' Global - ~~$1720~~ $960 End-to-End GMAT Prep + App Support: 150+ Videos, 2000+ Questions, 15 FLTs +
	GMATWhiz - $299 ~~$499~~ Experience the course through a free trial.
	GMAT Club Tests - Free Included with every course purchase of $149+ View all deals

A Possible Strategy for Improving Your Score? Or Maybe Not....

Prep Toolkit

Top 5 GMAT Debrief Videos

615 to 715 in 15 Days (Ria's Strategies)

More than Quant/Verbal Abilities: Speed vs. Accuracy

745 with Only Self-Prep and GMAT Club

From 555 to 765: 210-point Improvement

Perfect 805 Score Debrief by Julia

My Rewards