The “Gamble Retake”

Maybe you had a fine day at the test, but you got a 690 and you’re aiming for one of the top 10 or so programs, where the average score is around 720. You hear the standard error is about 30 points, which is almost exactly the size of the increase you’d like. So you wonder whether you should gamble on a retake—simply take the test again and hope to do better. Knowing that you can simply cancel your score if you do the same or worse on the test provides a good measure of insurance against those risks. However, when you are contemplating such an investment of money ($250 for test registration), time, and stress, it’s first worth considering logically what your actual chances are of improving your score without additional dedicated study in the interim.

Here’s a breakdown of the gamble retake. Let’s consider three cases:

1. Say your true score (at your current skill level) is 690. In this case, you are quite likely to get a 690, or close to it, on a gamble retake. Moreover, you’re as likely to get a 660 on your gamble retake as a 720.
2. Say your true score at your current skill level is a 720. This is the possibility you want to believe is the case. In this case, your 690 score was somewhat lower than your true score—not necessarily because you had a bad day but perhaps due to statistical variation in scores. In this case, on your gamble retake, you’re most likely to score at or near 720.
3. Say your true score at your current skill level is a 660. This case is the mirror image of case #2. In this case, your 690 was indeed a fluke, but in your favor. You’re likely to score at or near a 660 on your gamble retake, and you’re just as likely to score under 660 as above 660.

There are other possibilities, of course—your true score at your current skill level could be 680 or 700 or something near this range. It could even be as far away as 600 or 780—but such cases are very unlikely (and if you’d like to gamble on those odds, you may want to dedicate your B-school tuition to playing the lottery instead). But given the statistical reliability of the GMAT, the most likely case of all is #1. Cases #2 and #3 are less likely—and each of those is as likely as the other.

So even if you are only 30, 20, or 10 points shy of your goal, the gamble retake is not a particularly good gamble. The risk posed by case #3 is entirely mitigated by the fact that if you underperform on your retake, you may simply cancel your score. Nevertheless, a straight-up gamble retake will produce the same score more often than any other result—the most likely outcome is a simple waste of time and money. In general, a retake is the right move only if you have a month or two beforehand to devote to serious study.