GMAT Official Practice Test: Verbal (V44, 2 questions wrong)

Hi SynPo800,

The good news is that V44 is a VERY GOOD verbal score. Also, as BB mentioned, it’s difficult to know the specifics of the GMAC algorithm, so you can never know for sure why those two incorrect questions resulted in V44. Have you taken any other official GMAC practice exams?

Need additional insights on the above two points:

1. The assertion about variable quantities of questions in tests 1 through 6. First, is it confirmed (by whom?) that tests 1 and 2 have large pools, while the others barely enough. Would this mean that resetting tests 3-6 would result in many repeated questions?

I understand from some posts here that tests 1 and 2, and perhaps the others - please confirm - can be reset multiple times with some new and repeated questions. Is this correct?

2. Appeal to investigate score of V48 with all correct answers and subsequent revision of score: Did GMAC provide official explanations for how such a situation occurred? Were there errors or something else?

Curious to know these stories in more detail. Yes I have read some of the posts pertaining to these two points, but I remain unclear about them.

Thanks.

Posted from my mobile device

Congratulations on your great score. In my opinion, I think failing 2 questions should have given you a better verbal score. Anyway, all the best.

Posted from my mobile device

The two questions could have been at a lower difficulty level. Congrats on your score!

There have been one or two test takers with no wrong answers who have gotten V50 scores, not V48 scores (or at least not that I've seen, and that seems extremely unlike to me). On appeal, GMAC revised the scores upwards to V51. You can understand why this could happen - if a test taker gets every question right, the algorithm can be very confident the test taker is above a certain level, but can't say exactly how far above. If a test doesn't contain a lot of extremely hard questions, then with entirely correct answers, the test might only be nearly certain the test taker is at least a V50, but might not have enough information to be confident the test taker is a V51. But it's not the test taker's fault her test did not contain enough hard questions, so it makes sense they'd revise those scores to V51. The test isn't really designed to distinguish between test takers at those levels anyway, since anyone at that level is miles above the level of the test in general.

If you have only two wrong answers in Verbal, most of the time you'd expect a V47-V48 score. If you only got a V44, then the two wrong answers must have been on easier questions (and it might also be true that your test was a bit easier than normal overall, so your right answers didn't demonstrate quite as high an ability level as on a normal test). On the real test, you'll likely see more hard questions than on some of the diagnostic tests, and if you only get two questions wrong, you'll probably get at least a V46.

Manual intervention is an extremely unusual case. Usually GMAC (and other test agencies) publish their findings in research papers. I wonder if this was the case here.

Although I understand the direction of the quoted argument above, it may not be entirely accurate.

It is understood that a test which is designed to measure abilities of a large number of test takers, most of whom will fall within a few standard deviations of a mean score, may not be accurate with scoring at the edges - i.e., extremely low or high scores. This is fundamentally due to the fact that the algorithm may not possess sufficient data in those regions to make accurate estimates.

In these cases, researchers usually assign an arbitrary value, say V51 or V(some extreme number) to denote the extreme upper value. This is done with the caveat that the range of confidence of this value is quite large, i.e., the value is less certain. An assignment of an arbitrary value is not generally an issue here because we the test is designed to measure abilities within broader (lower) ranges.

Researchers will not generally go in after-the-fact and manually assign a new value to the upper score. Such post-hoc tweaking of data is discouraged.

I would be interested in seeing an actual GMAC research paper that more fully described the events. Generally, the test agencies will proactively publish such reports to head off any concerns that may arise over their actions.

The situation favored the testtaker who had the time, money and sense to request a review upon which the agency changed their score. Some other testtakers might not be so fortunate.

In my view, this is a highly unusual occurrence.

The details are all here - a test taker who answered every Verbal question correctly received a V50, that was later revised to a V51 after the test taker asked GMAC to look at the scoring:

https://gmatclub.com/forum/q51-v50-253258.html

Internally, when you proceed through the GMAT, your score is not measured on the V6-V51 scale. Instead it's measured in standard deviations above or below average. So initially, a test taker starts with a score of 0, and after answering a few questions correctly, might climb to a score of 1 (one standard deviation above average), and so on. At the end of the test, that standard deviation score is translated to a score on the familiar V6-V51 scale. If the test does not deliver enough hard questions, the test taker might not quite reach a standard deviation score high enough to translate to a V51. That's the only way I can explain what happened to this test taker (and the other one linked to in that thread) - if you look at the difficulty level of their Verbal questions, as reported in their ESRs, their questions weren't as hard as you'd expect, based on their performance.

It would be unsurprising if the test designers hadn't accounted for this possibility before it happened, since it would be rare that the question selection algorithm would fail to deliver a lot of hard questions to a well-performing test taker. They may have since corrected that behaviour, but it would be, as you say, a very rare occurrence, if only because flawless Verbal performances are extremely rare (probably achieved by 1 in 5000 test takers or thereabouts), and rarer still would be the flawless performance that only produces the V50 because of a quirk in the question bank on the day of the test (an insufficient supply of hard questions).

It really depends on the questions, e.g. I had a v46 in one practice test with 3 errors and a v40 with 5 errors.

Its a helpful gauge for your verbal abilities, but it is not useful to assess which amount of errors will result in which final score.

Cheers,
Chris