First some raw 2005 stats:
(obtained from
https://www.gmac.com/gmac/VirtualLibrary ... 0%9305.htm)
Total Mean GMAT: 525
American Mean GMAT: 519
Indian Mean GMAT: 560
Chinese Mean GMAT: 589
Computer Science Mean GMAT: 571
Engineering Mean GMAT: 582
Now Analysis:
If you've taken statistics, you know how to compare things that have different distributions, example: which is more rare, 20 lb cat or 30 lb dog? you take the z score (difference from mean over std deviation) and compare those.
Lets make an assumption that std dev for gmat score remains somewhat the same between countries (since no data avail here), this means where u rank in ur group is purely based on the mean. So to compare an Indian from an American on how they rank in their own group, one would subtract 41 points. To compare an engineer to the total population, one would subtract 57 points.
If you are both Indian and an engineer, then to be compared to the general population, you would need to subtract 35 + 57 - X points, where X is unknown but less than 35.
Caveats:
The above assumes an Adcom will want to know how you rank within your demographic and compare in such a way. And it also assumes standard deviation is the same in all demographics which can't be true. But it gives an approximation.
By the above logic, my score of 780 (chinese and engineer) is probably comparable to a 700 in the general population. This is like saying, I am ranked among chinese engineers similarly to someone ranked in the general population with a 700 score.
This would have great impact on what someone's target score should be based on their demographic. And really lends a lot of weight to people making the claim 'you are overrepresented, you need to stand out in some other way'
Discuss.