Are you including the 6 blank questions among your 7 wrong on Quant? Or was it 7 errors + 6 blanks = 13 total?
pinecrest wrote:
1) I was surprised that my quantitative score (Q:43) was higher than my verbal score (V:39), even though I answered more verbal questions and got approximately the same number of questions wrong. Is there an explanation for this?
(1) The 0-60 scale for Quant scores and Verbal scores is different. It’s more useful to compare percentiles. Your 43 in Quant is 67 percentile, but your 39 in Verbal is 87 percentile. You did do better in Verbal, as you suspected.
(current chart here:
https://www.mba.com/mba/thegmat/gmatscor ... gmeans.htm)
pinecrest wrote:
2) I heard that in the GMAT CAT, when you answer a question correctly, the next question becomes harder. If you answer a question incorrectly, the next question becomes easier. I assume questions are weighted differently, correct?
(2) All questions are weighted the same. But missing a very easy question would hurt your score more than missing a very hard question, because it effectively drags your average* down more. So if you must think of questions as weighted, think of them as weighted by how much easier than your goal score/ability they are. As for questions that are harder than your target score, getting them wrong won’t hurt you a bit. (If only we could know during the test which ones are these freebies!)
* Your score is not just a simple average of the question difficulties, but this is a practical way to think of it.
pinecrest wrote:
3) On a similar note, person A who answers his first 10 questions right would in theory need to spend more time with these progressively harder questions than person B who misses half of his first 10 questions (and get progressively easier questions). But that would put person A at a disadvantage in terms of the clock - he would have less time than his counterpart to answer the remaining questions. How does GMAT penalize incorrect questions vs. unanswered questions? In general, suppose someone answers almost all his Quantitative questions right (e.g. only 2 incorrect) but due to progressive difficulty only manages to finish 30 questions (leaving 7 questions unattempted), how will his Q score compare to someone who answers slightly more questions incorrectly (e.g. 6 incorrect) but manages to attempt all 37 questions and therefore has a slightly higher (# Q correct/Total 37 questions) ratio?
(3) The penalty for blanks is roughly twice that for wrong answers. Don’t leave any blanks, whatever you do. Even a wrong random guess is better. In your example, the person with 2 true errors and 7 blanks would definitely score lower than the person with 6 true errors, assuming everything else was the same.
In reality, everything else would not be the same. To finish on time, nearly everyone at every level must guess! The person who made 6 true errors working at an even pace probably made other good decisions along the way: guessing when a problem was too hard to finish in 2 minutes (and therefore one that wouldn’t hurt the score anyway), taking time to be careful on the dangerous easy questions.
pinecrest wrote:
4) In my CAT Practice test, I ran out of time on my 31st Quantitative question; the clock ran out before I could CONFIRM my answer. Similarly, I ran out of time on the 41st Verbal question, before I could confirm my answer. However, in the review column after the test, the computer showed that I attempted 31 questions for Quantitative instead of the 30 whose answers I confirmed, and counted that 31st unconfirmed question as incorrect. Similar for Verbal. In real GMATs, if the tester gets to a question, but fails to confirm an answer before timeout, does that question count as attempted and incorrect?
(4) When the clock stops, the algorithm takes any answer you have selected, even if you do not confirm. bb (GMATClub) recommends selecting an answer bubble the moment you see the question, then change your selection if necessary before confirming. The idea is to prevent blanks.
pinecrest wrote:
5) Can anyone enlighten us on the actual formula/algorithm used to calculate the scores? Seems like scores will vary depending on combination of #attempted/#correct/#incorrect/question difficulty.
(5) Your score will depend on all of that. Every question on the GMAT has not just a single difficulty value, but actually a difficulty function. It is ability (x-axis) vs. probability of getting it right (y-axis). In other words, people with low GMAT ability (score) have a certain probability of getting a given question right (lowest is 20%, because multiple-choice) and people with high ability have a higher probability of getting the problem right.
Your score will be the product of all the upward-sloping curves for each right answer and all the inverse downward-sloping curves (probWrong = 1 – probRight) for each wrong answer. So you have a probability function associated with you. Your score is based on the ability level where your probability peaks.
In practical terms, your score has a lower limit set by the questions you get right and an upper limit set by the questions you get wrong. Thus the importance of the “easy” questions: you want them to be part of your lower limit calculation, not the upper limit calculation.
close
27 Jun 2010, 10:22
