OG 13 contains only 5 Hard Questions [GMAT Club Analysis]

This has nothing to do with whether or not a question is hard for a specific person. This is why I was saying it is subjective for a person. Whether a question is difficult for a person depends on their own abilities and talents. For some people quant 51 problems are easy, but v40 questions are hard.

The percent's quoted above are from GMAT CR Bible page 96. I was slightly off on the 25% it says "...The problem is answered correctly by about 45% of test takers and is classified as difficult (the hardest GMAT questions have success rates under 20%. Fortunately, questions this difficult appear infrequently.)

I wouldn't agree with saying it's not helpful to the top 10%. This reminds me of critical reasoning's survey error? Biased samples? Insufficient samples? Perhaps false premise or needed an alternative premise to support the conclusion?

Posted from my mobile device

I am very open to suggestions or alternative data/input. However, you need to say WHY you do not agree and what specific items that you are referring to. I can't really comment on your post at this time.

You are on your cell phone so take a look at the data when you have time.

BB, quick question... Were OG12 guide categorization of the questions subjected to same analysis I.e user response on timed questions

Posted from my mobile device

We have not done OG 12 yet - I will be happy to do it next week.
I am collecting input about suggestion to improve methodology/approach before I do it, so welcome any critique or improvement recommendations.

It shocking the first time I took a GMAT Prep after having finished OG, so I'm not surprised by this. Here's what I find really interesting about this: ten years ago, this situation was probably the same, so people probably didn't realize that you had to study much more complex problems than what's in the OG to do very well. Now, everyone who uses the internet is going to find this site and its peers and realize that you do have to study those problems, and therefore, this is contributing to the shifting percentiles.

What was your method for arriving at the number in the "Percentiles" column?

WOW, that is one bold statement. Still, it doesn't undermine the fact that OG is the best available material to ace the GMAT.
I don't agree 100% with the analysis. My assumption is, people come to GMATClub for more practice questions only after solving the most of OG questions.
Don't you think the responses are unreflective? At least in my case they are, as all the OG questions I solved in GMAClub are done before.

It depends on how you define top 10%. If you mean quant 50-51 than maybe you're right. But if you just define top 10% as being 700+ I would disagree. Some people will get scores like q45 (66 percentile) V45 (99 percentile) and get a 730 GMAT (96 percentile overall). A lot of people will get Q49 (81 percentile) or lower on quant and still get 700+ overall. And what constitutes a "hard" problem in verbal is fairly subjective. GMAC rarely asks questions where 25% or less of the test takers get it right. I doesn't do any good for them to ask questions that everyone gets wrong. Just my opinion.

I want to add that I think GMAT prep questions are by far the best practice.

Thank you for your input and thoughtful reply - that's a valid take and it is possible that the data is contaminated this way - there is no way to prove it one way or another.
My argument would be... why are they taking questions already attempted? I can remember the answers to pretty much all of the questions I have taken (perhaps that's me) but I would not take questions I have already done. That was my experience and I could see someone taking a question they vaguely remember but I am not sure someone wants to time themselves on that one... it is quite possible but it is not clear to what extent.

Agree Very bold statement (probably too bold for me actually but it is an experiment to see if we can provoke some creative thinking/argumentation to prove me wrong ).
No doubt - OG 13 and GMAT Prep software are the best sources of the official questions.

Thank you for your input. All of it is very straightforward. By the top 10th percentile, I mean anyone with 90th percentile or above on the GMAT.
As to difficulty, it is very easy to tell difficulty of a question once you have a sample of attempts along with the time it has taken to answer a certain question, so it may not be obvious to you but it is pretty clear to someone with data which verbal question is harder than the other.

Now for the stats that you have provided, did you come up with it or that's actually an official quote? 25% of questions? I would be very curious. That's contrary to my GMAT experience where I got hit with a Combinations question as well as contrary to my GMAT Prep experience where I got hit by a few hard ones that I have never seen before anywhere.... GMAC does need hard questions.... it gets pretty tricky with high scorers.

Question for you - do you think the same would apply to GMAT Prep questions? Meaning, GMAT prep questions would be generally having a lower difficulty reading due to many people going through the software before visiting the question on the forum?

It is an algorithm that combines the number of correct/incorrect answers, total number of answers, and time taken for correct/incorrect answer.

I am just trying to get to the bottom of the this issue and see if we can find the same correlation in GMAT Prep (e.g. most questions are rated easy or if there are truly hard questions that are indeed rated hard even though users have seen them). I don't think it will be the same as the OG, since GMAT Prep questions are much harder to fish out/find/discover.

However, I know for sure there is a number of them on the hardest question list here, #4 for example.

Thank you! Thats pretty interesting. I wonder where PowerScore got that stats.
I think they are missing the picture a bit since almost ANYONE can crack even the hardest question with an unlimited amount of time. The trick here is to know how long it takes an average person to answer the question correctly and also incorrectly, which actually I think the incorrect side of the answer, compensates for any contamination of the results by the re-takers. (I will check this with the developers who can look up the algorithm in the code) - thank you for helping me stumble on this!

Wait, are we talking just the Quant section. If yes, I agree that there aren't lot of tough questions in OG and I would agree with your analysis.