[email protected] wrote:
The scoring algorithm on the Official GMAT is far more complicated than most people realize. It takes into account a number of different factors, including the relative difficulty of the question, whether you were expected to get it correct or not, the placement of the question, what's going on "around it", the "strings" of correct and incorrect answers, whether the question even counts or not (some questions are "experimental" and are worth 0), if you leave questions unanswered and incur a penalty, etc.
I have an issue with descriptions that make GMAT scoring sound like something other than what it is - that is, descriptions that make it seem that GMAT scoring takes account of factors other than test taker ability. Most of the factors you list simply are not part of the GMAT algorithm. Question position, for example, is not used anywhere in GMAT scoring. If you took two GMATs full of identical 500-level questions, and answered ten of them incorrectly, your score would be the same whether those questions were the first ten or the last ten (or any other set of ten). Questions aren't somehow 'weighted' by where they appear in the test. It's the difficulty level of questions that matters.
I agree with you that test takers do not benefit from understanding how the algorithm works, just as university students don't benefit from knowing how their calculus test will be graded. GMAT scoring is based on decades of academic research into using probability theory to gauge test taker ability efficiently. There are no special 'secrets' to the algorithm, or unexpected 'factors' the algorithm takes into account, that a test taker could learn and use to 'beat the algorithm'.
That said, the test taker should know a few things, for example that adaptive tests will feel challenging no matter your level, and that the test is very forgiving of mistakes on hard questions, and unforgiving of mistakes on easy ones. As a consequence, test takers should be willing to move on quickly from difficult questions, and should answer very carefully those questions they know how to solve. And test takers should avoid guessing at randomly chosen questions, since they then risk answering an easy question incorrectly.
And to reply to the OP's question, since question difficulty is critically important, you cannot begin to guess what score you'll receive only from your number of correct answers. Test takers who score a Q48 normally have roughly the same number of right answers as test takers who score a Q31. The difference is, the Q48 test taker is able to answer harder questions than the Q31 test taker is able to answer.
