Expected Number of very hard question on the real test 70%+ difficulty

Hello, adexit. One thing to keep in mind is that GMAC™ does not officially place questions into any category other than Easy, Medium, and Hard. I am sure that within, say, Hard, there is a spectrum of question difficulty, based on the percent of test-takers who had chosen the correct answer. Keep in mind, though, that any other scoring system from a third party, such as 700+ or Very Hard, is speculation. Someone can design a question that may prove quite difficult for many test-takers, but until data are in hand and the question is normed, there is no way to gauge how difficult that question may actually be. Furthermore, I have seen plenty of discrepancies between GMAT Club categorization of a question and its official rating. The reason is simple: as bb himself has noted, the GMAT Club rating of a question is based on site users. The more who answer correctly, the lower the difficulty will indicate. Since I speculate that many GMAT Club members are well-versed when it comes to official questions, the data for those questions may be tilted toward an easier categorization.

With all of that said, however, you should also keep in mind that the exam is adaptive. To hit a 700 or above, you will want to string together streaks of correct answers. If you get to question two, a Medium question, and miss it, for example, you might not see a Hard question until you can prove to the system that that incorrect response was an aberration. You can only see those devilishly difficult questions if you prove adept enough at answering perhaps the lower-level Hard questions in succession. From ESRs I have examined through this site and just as a tutor in general, I can tell you that you will want to answer more accurately in the earlier part of the test to see more Hard questions. Miss too many Hard questions early on, and the ceiling might have been set. Could you still walk away with a 700 in such a circumstance? Yes. That is why, in general, I urge students to play the probabilities and consider their performance across different types of questions within each difficulty level. My 700 gauge is a little conservative:

Easy - 95% accurate
Medium - 80% accurate
Hard - 65% accurate

Provided you could achieve such accuracy across sets of official Verbal and Quant questions under timed conditions, I see no reason why you would not achieve your aim. You might still feel as if I have not answered your original question, but that ties into the adaptive nature of the test. Consider two scenarios:

1) √ √ √ X √ X √ √ √ √
2) X √ √ √ √ X √ √ √ X

In both sets of ten questions--ignoring, for the sake of simplicity, where those ten questions might be in the set of thirty--the test-taker was able to put together a string of correct responses. Maybe in case 1), the student saw a Hard question in slot 2, 3, 4, 5, 9, and 10; in case 2), the student saw a Hard question in slot 4, 5, 6, 8, 9, and 10. Both students could have seen the same number of Hard questions through different paths. This is just a scenario, but the point I am trying to make is that you simply do not know when a "Very Hard" question might come at you. It is all based on test-taker data, and you might breeze through a question at, say, 85% difficulty, while you might flounder on one that is 65%. Place more stock in your practice data and, most certainly, your official practice tests. You cannot game the system. Just work on improving your accuracy at all levels, and then watch the results follow.

Good luck with your studies.

- Andrew