How is GMAT combined score calculated

I have a few answers to this probably.
1. I will try to go out and talk to ETS about a few things, to try to clarify. After my pleasant experience with one of the GMAC employees (sometimes she is around on this forum), I will probably try to go and ask ETS and see what they give me.

If not, I think I have accumulated enough statistics to derive a formula. I am sure it won't be that easy since thier algorithm is probably more complex than just a standard condition, but an approximation will work just find for our needs. I think the findings will be quite important just seeing how ETS weighs sections. It may be a breakthrough. or it may not That's when large masses of people come handy. Together we can actually pull it off.

Another project I want to undertake is actually to take a PowerPrep several times over and over again. (I would love if somebody would volunteer for this). What I will try to accomplish is taking the math section 5 times: 1) answer all questions correctly. 2)answer first 2 wrong, the rest correctly 3)answer 2 last wrong 4)miss 2 last ones 5) miss 2 middle ones and compare the results. I am sure this analysis would be helpful too should ETS be stingy about information.

Anyway, those are dreams, aka projects
Back to your question, my current blief (not supported by any scintific research) says that a more balanced score will lead to a higher percentile. I got 49 on Math and 42 on Verbal which was 92% and 96% and the total was 750, 99% percentile. Pretty bizzare, but I think few people who scored 96% on verbal scored 92% on math and vice versa, thus though the percentiles are lower than 99, combined they make a 99% percentile. (Well, I guess it is pretty obvious, sorry for boring you).

Anyway, I can try to guess more, but I won't be able to tell until I actually try to take a look at a large sample of raw/final scores. An interesting thing is that I saw a person with the same percentiles as mine, but 740 total score. He had 49/42 and 740 total. That surprized me cause I don't think anybody would intentionally lower their score by 10 points, or maybe the guy just made it up. I saw it on the PR forum.

As to the second question, If understood you correctly, my guess would be that ETS folks accumulate data over time, for the last 3-5 years, say. I know that scores did get inflated in the last few years. It used to be that 700 would get you anywhere, just the score. Not anymore I dont' know actually... I have an old GMAT score guide and I think there is a newer one. I will try to compare them and see if there are any differences. However, I don't think it is due to the flaw in the software of question difficulty, but due to more professional and focused GMAT prep. There are more resources out there, more software, more books that allow people to practice more, thus improving their test taking skills.... According to the ETS bulletin on the test scores, the Standard Deviation (or marginal error) for each individual test taker is 29 points.