GMAT Question - for the math 'impaired'

I am considering a program that requires one to write the GMAT with an acceptable score for admission.

For those of you that have taken the GMAT... and those who are "semi-math/quantification impaired" (like me)- I wonder what (ballpark) score you may have received, as well as your general opinion of the test.

I can normally do the math stuff, but I don't know wabout doing it under the pressure of a stopwatch.

Thanks in advance.

 

I took it in 1981 as a 21-year-old entering graduate school. I scored 590, which was somewhere around the 70th percentile at the time.

The mathematics required to do well don't go beyond simple algebra and a little plane geometry. The test is designed to measure your "aptitude" for problem-solving and thinking in B-school. It does not purport to measure the math skills necessary to get you through an operations management course. Nor should it, since there has been an explosion in the number of MBA programs that do not require much math beyond algebra and statistics.

You're much better off honing your test-taking skills, because that's what the GMAT--and other ETS examinations--really measure. That and an ability to see the world in the same stultifying manner as the question writers for these tests.

Rich Douglas

 

uxu - The only advice I can give is to obtain a good GMAT study guide and prep for the test. A solid study effort can pay off in a higher score. The math part never struck me as being hard - but I'm pretty math literate.

At the risk of getting off track - here is a perspective on the GMAT and other tests like it (LSAT, MCAT, etc.).

The GMAT and other tests like it serve a valuable purpose. They typically aren't used to "admit" students. Quite the opposite, they are used to "eliminate" candidates from a pool.

If you are the admissions director at a top of the line medical school, for example, you may have 7000 applicants for 150 spots. Harvard Business School has a similar problem - thousands of applicants for a limited number of spots (880). How do you say "no" 6850 times? You can't base decisions on someone's hair color. So you turn to a standardized test.

Are such tests any good? Well, there is a statistically significant correlation between scores on the GMAT and first year MBA grades. The correlation (R squared) is about .38. This tells you that about 38% of the variation in first year MBA grades can be explained by the GMAT score. About 62% is explained by a number of other factors - like motivation.

My take is this - at a gross level the GMAT (and MCAT, LSAT, etc.) are useful. If you have a 400 GMAT, you'll almost certainly fail at Harvard, so Harvard is justified in turning you down. Smaller differences are insignificant - two candidates, one with a 550 and the other a 560, are effectively identical.

The biggest problem that the GMAT has is that the test is pretty unfair to non-English speakers and non-whites. For whatever reason they tend to score lower.

Thanks - Andy

------------------
Andy Borchers, DBA
NSU (1996)

 

Uhhhh....no. The stated correlation of .38 is a description of how well test scores and subsequent performance line up. A perfect correlation of 1 would indicate that they correlate perfectly; that in every case, the higher the test score the higher the subsequent grades. A correlation of -1 would indicate a perfect inverse relationship; a correlation of 0 would indiate no relationship. Thus, a correlation of .38 means there is some correlation. The test kinda sorta tells you that if someone does well they might also get good grades. Whether or not .38 is statistically significant (not likely due to random error) is dependent on the number in the sample (n) and population (N). The larger the sample, the more likely the observed value in the sample reflects a real value in the population. But whether or not that statistic has real value is another matter entirely.

Also, a correlation of .38 certainly doesn't mean that "about 38% of the variation in first year MBA grades can be explained by the GMAT score." No way. You could run a number of correlations using a variety of factors, each resulting in a correlation coefficient. (Pick 'em: high school grades, college grades, school ranking, family income, race, gender, whatever.) Sooner or later, they'll add up to more than 1 (which invalidates this notion; you can't have a probability higher than one). There are tests to measure how much one factor contributes compared to another (grades vs. test score, for example), but that's beyond the correlation coefficient. No, this figure simply indicates how well two other numbers move together (up or down).

A better comparison is the correlation coefficient of GMAT scores with the correlation coefficient of other factors. I believe you'll find that college grades are a much better predictor than GMAT scores.

So why do schools use the GMAT? For the exact reasons Andy said: to screen out huge numbers of people. But it is a very blunt instrument. Sure, a GMAT score at the extremes can tell you something, but one in the middle almost surely cannot. The standard error (SE) on the GMAT is somewhere around 60 points. (Near the median; the SE at the extremes of the grading scale are much higher!) That would mean there is no difference between someone with a 550 and someone else with a 610. (And, likely, no difference between the top score and 100 points below it, which could represent as little as one wrong answer on the test.) But who do you think will get to a competitive school? Would you want your chances of getting into a good school based on a few (very few, 2 or 3 even) questions that go one way or another? Or would you want your admission based upon the whole record?

According to the ETS, the GMAT--when used in conjunction with college grades--can add a very small amount to the admissions officers' jobs (predicting who will do well in school and selecting them). This number, unlike the 38% cited above, can be discerned and distinguished from the correlation coefficients of grades or GMAT scores standing alone. But it is small and hardly not worth the use of a flawed test.

Most MBA programs in the U.S. have more openings than applicants, and don't need a test like this to make their admissions decisions. Funny how the rest of the world is able to get by without them, too!

If schools had to carry the burden of paying for the GMAT, GRE, SAT, etc., instead of having them foisted on the applicants and their families, these tests would dry up and blow away.

Rich Douglas

 

Rich - I hate to disagree, but I think I'm right here. The correlation you describes as being -1 to +1 is R, not R squared. R squared does in fact measure the % of variation explained. To quote from McClave "The coefficient of determination (R squared) is the square of the coefficient of correlation (R). It represents the proportion of the total sample variablity around y-bar that is explained by the linear relationship between ya dn x."

On another note, I spent some time at the GMAC (Graduate Management Admissions Council) site. They have some very interesting research. In one set of graphs they show the probability of being in the bottom quarter of one's class based on GMAT scores and undergraduate GPA. For students with undergrad GPAs of 2.75 and GMAT scores under 450, over 70% ended up in the bottom quarter of their MBA classes. For folks with GPA's of 3.25 and GMAT's under 450, 58% ended up in the bottom of the class. Looking at the converse - virtually all of the students in the top quarter of the sample MBA classes had GMAT's over 550.

Bottom line? GMAT and GPA aren't perfect indicators of potential for success in MBA programs - but they are significant and valid indicators. What admisssion directors get from the GMAT is a tool that helps, imperfectly, in picking applicants that have the greatest odds of being successful.

Thanks - Andy

------------------
Andy Borchers, DBA
NSU (1996)

 

Darn it, Andy, you're right about r squared. I just didn't read your first post correctly, and assumed (wrongly) that you were making a common error regarding the meaning of a correlation.

I've not read the site you cite regarding the GMAT. I'm sure the data you report are correct, but they don't necessarily show the efficacy of the GMAT itself, just the GMAT and GPA together. My contention is that the GMAT doesn't add much to the admissions decision; certainly not enought to justify the cost and the qualified students it sometimes excludes. I reiterate: the vast majority of MBA programs don't need it to make effective admissions decisions and almost no non-U.S. schools do.

Seriously, thanks for the stats lesson .

Rich Douglas

 

Thanks for the advice - and the stats lessons...

I dont really intend on getting into 'haaavaaad'... but the local U wants at least a 400 GMAT to be admitted into their program.

Thanks again.