One of the things I am most amazed of is the importance of the GMAT to entrance decisions at many doctoral programs. The GMAT is just another version of a standardized IQ test, one that can be gamed significantly if you practice, practice, practice . . . . . . . My proof, I got an 85% percentile hung over and a 97% sober.
I was also the “runt” of the litter in my PhD class at Michigan with that measure in mind. Now you take the publication records of my classmates (indeed consider the two years before and two years after my entry where I continued to be the runt of the GMAT scores in the group) you will find only one of those 25 (at that time Michigan took 4-5 students a year) who research output is comparable to mine.
I am not certain that GMAT tells you a lot about the characteristics of what it takes to be a good researcher. Sure you have to be bright and the closer you are in brightness to the top 1% the better the raw material might be. And the GMAT score, after taking it two or three times will give you a good measure of raw intelligence. But there is also institutional knowledge, drive, ambition, curiosity, the ability to deal with rejection etc that all come into play in carrying out original research. Surely a four year undergraduate degree and any masters degree should tell you a lot more about academic potential than a standardized test.
Yet year after year I hear the same fallacies (i.e. “You need strong math skills and a high GMAT to be a successful PhD accounting student.”) perpetuated in PhD student selection decisions so as to make them almost self-fulfilling. It ain’t so . . . .