GMATPrep Test-2

So, you ask a really good question regarding GMAT scoring. Here are some thoughts:

People often wonder how it could be that two people who achieve the same scores on the Quant and Verbal sections of the GMAT may get different total (200-800) GMAT scores. Others wonder how it could be that the percentile rankings of their total scores (200-800) are significantly higher or lower than the percentile rankings of both of their Quant and Verbal section scores. The answers to these questions lie in the way the GMAT is scored.

To get to an understanding of how the GMAT is scored, let’s first counter a common misconception, which is that the GMAT total score (200-800) is calculated using the two section scores (the Quant score and the Verbal score). For simplicity, let’s define performance as the number of questions a test-taker answers correctly and the difficulty of those questions. The total GMAT score (200-800), while based on your performance in answering questions in the Quant and Verbal sections, is not calculated by using the Quant and Verbal scores. Once a test-taker has completed the GMAT, the system separately calculates a total (200-800) score, a Quant score, and a Verbal score.

The next thing to understand is that the three scores are not exact representations of a test-taker’s performance, but rather each score level represents a range of test performance. For example, a test-taker could get 28 Quant questions correct or get 30 Quant questions correct and score Q51 in either scenario. Likewise, that two people both score V38 or both score 670 means that their levels of performance were similar but not necessarily exactly the same. Thus, a test-taker who scored, for instance, Q38, could have performed at the low, middle, or top end of the range of performance that generates Q38. Likewise, V36 could be a higher V36 or a lower V36, and 720 represents a range of performance levels.

To understand how these aspects of the GMAT scoring method come together to generate a set of scores, consider the following example: A score of Q51 is based on the test-taker’s performance in the Quant section, and a score of V50 is based on the test-taker’s performance in the Verbal section. To get the total score (200-800), the computer wouldn't use the Q51 and V50, which, as discussed above, could mean a variety of things in terms of performance; the computer would use the test-taker’s actual performance on the 67 questions that appeared in Quant and Verbal sections. So, for example, you could score Q51/V50 with 28 questions correct in Quant and 34 correct in Verbal. The total score (200-800) would be based on 28 Quant questions correct and 34 Verbal questions correct and would likely come to 790. Alternatively, you could get the same section scores of Q51/V50 with 30 questions correct in Quant and 35 correct in Verbal. Now the total score (200-800) is based on 30 questions correct in Quant and 35 questions correct in Verbal and you score 800.

So, to get the Quant score, they use your performance on the Quant section, to get the Verbal score, they use your performance on the Verbal section, and to get the total score (200-800), they look at your entire performance on the 67 questions that appeared in the Quant and Verbal sections.

With this knowledge, it’s easy to understand how two people could get the same section scores but different total scores. If a person gets a certain set of section scores, say Q47 and V33, that person may have performed at the high ends of the performance ranges represented by those scores, and thus receive a relatively high total score, such as 650. Another person might perform lower in the ranges that are represented by those same section scores, Q47 and V33, and receive a lower total score of 640. So, the two people got the same section scores via slightly different performance levels, and thus received different total scores.

Also, since a test-taker’s total score (200-800) percentile ranking is based on the person’s total score, two people with the same section scores will have different total score percentile rankings if their total scores are different.

https://www.reddit.com/r/GMAT/comments/ ... mat_score/

The GMAT is not scored on any kind of curve. So, how a particular test-taker's GMAT score is calculated does not depend on how other test-takers have scored. Rather the GMAT is designed to generate the same score for any level of performance that the test would have generated for that level of performance in past years.

This consistency is achieved through testing of new questions by including them on the test on an unscored basis. Only after the creators of the test have determined the difficulty level of a question does that question appear on the test. Theoretically at least, this process enables the test-makers to keep the difficulty level of the test the same year after year and to score the test in the same way year after year.
The score percentiles, on the other hand, are based on the performance of all test-takers who have taken in the test in the past three years. So, the percentiles fluctuate depending on how test-takers perform.

You ask an interesting question but unfortunately, it’s nearly impossible to provide an accurate answer, though I can give you a rough estimate. A very rough estimate is that, to score Q49, you have to correctly answer 25 to 28 quant questions, and that, to score V38, you have to correctly answer 26 to 31 verbal questions. I'm not taking experimental questions into account, by the way. The estimates are the numbers out of the 31 total credited and experimental quant questions and out of the 36 total credited and experimental verbal questions. So, you can see that you can achieve the scores you mentioned even if you miss a handful of questions in each section. All the same, you have to remember that your gmat quant and verbal scores are based on not only the number of questions you answer correctly/incorrectly, but also the difficulty of those questions, among other things. Thus, there is no way to guarantee a particular quant or verbal score by correctly answering a particular number of quant or verbal questions. Thus, my advice is to focus on the things you can control, right? Let’s not worry about the GMAC algorithm but put your energy into just getting better at the GMAT. If you get to a point at which you can dominate the GMAT, the specifics of the scoring algorithm won't matter because you will have the skills you need to get an amazing score, right?