The following questions covered so far:

1. What is the purpose of the Stat's database?
2. My GMAT score is less than 700, should I submit my stats?
3. I am afraid to submit my data bacause of privacy considerations!
4. Submission is buggy! I am submitting my stat's but it seems to lag out.
5. What does the rank mean?
6. What is the formulae of the rank?

Q: My GMAT score is less than 700, should I submit my stats?
A: YES! There is no shame if yours is less then 700 or even 600. You are a unique person, no one knows your conditions, and probably for you 600 is as achievement as for some is 700. Your successful or unsuccessful example of application process is very important. We want to know whether people with "deficiencies" in their application like low GMAT or GPA still can make up in the competitive schools. You don't have to give neither your name nor Email address. There is nothing to be afraid of, while you can advance our understanding of the addmission process.

Q: I am afraid to submit my data bacause of privacy considerations!
A: Please have a look at the privacy statement.

Q: Submission is buggy! I am submitting my stat's but it seems to lag out.
A: It is. This is just a first release, fields are not actually validated yet. So, if you put some words instead of numbers in the GMAT field, it may mess-up the thing. Try again if submission didn't work. Technical problems will soon be fixed.

Q: What does the rank mean?
A:  This is an attempt to consolidate all important quantitative properties of every MBA applicant. Frankly saying it doesn't mean a thing because it is widely accepted that at least in the highly selective schools qualitative side such as essays and recommendations play the most important role. However, having a single rank reflecting overall quantitative side of an application might be useful. In particular if we will try to correlate it with the actual results of the applications, ie whether those with high ranks succeed getting into the schools they desired. It is essential to understand that the ranking system is not to find who of us is the best, not at all. The system is an attempt to predict success in the application process (in respect of schools attempted) and should be viewed as such. All admission offiers use some sort of the ranking system in their work. We actually have no idea about their algorithms but we know that they are logical and use quantitative data (at least for screening). So, why not to try to reverse engineer them by creating similiar indicator? All we need is to collect quantitative and preferably qualitative data and comprare against results (rejected/accepted/waitlisted). While the only guarantee you can get is from an insurance cerfiticate, we can try to mastermind something close. Only the loosers don't try, so let's do it!

Q: What is the formulae of the rank?
A: The following ranking system was loosely designed by Alex Chudnovsky. It is by no means complete and comprehensive and I would be pleased to hear suggestions and improvements.

Fundamental assumption is that we should compare everyone against a perfect applicant. Perfect applicant is defined as a person with the maximum possible test scores (where maximum exists) and optimal quantitative indicators (maximum is undefined). Optimal quantitative indicator such as age and work experience are taken from the average admitted student in profiles of the Top10 schools. Optimal in this sense means that the perfect applicant will fit a school needs. For international (non-native English) students we consider TOEFL and TWE (if any).

The following variables are used in calculations:

Tests or indicators with feasible maximum value:

1. GMAT percentile (%)
2. Verbal %
3. Quantitative %
4. AWA%,
5. GPA (standard scale of 4.0)

6. Age (optimal assumption is 27 years)
7. Work experience (optimal assumption is 4 years)

8. TOEFL (maximum of either 300 for computer test of 677 for paper based)
9. TWE (maximum of 6.0)

1. for the GMAT total we may have 88%, which gives us: 0.880
2. for the age variable we may have 23, which gives us: 23/27=0.852

Consequently we will have 9 values ranged from 0.000 to 1.000, now we calculate their average.

However, taking the average as a final answer is way too easy and clearly wrong. We want to award consistently performing people and degrade those who luckily achieved some high scores (like the GMAT). Hence, we get average less standard deviation of the the data we used. This approach nicely degrades those who've had suspiciously "jumping" quantitative indicators. Finally we multiple the result we have by 1000, to make it in range from 0 to 1000 (the bigger is the better).

Some problems in my mind with the formulae so far:

1. Variables bear the same weight. Since we have at least 4 variables related to the GMAT test (GMAT,Verbal,Quantitative and AWA percentiles) in a typical US application this will account for 2/3 of the overall score (since TOEFL is not required).
2. While there is a penalty for underachieving optimal scores such as age and work experience, there is no penalty for overachieving them. Clearly, a 45 years old person has little chances to get in for a full-time MBA (executive MBA is still an option though), hence we should penalize for having age over some threshold. Such a threshold is yet to be defined, and probably should be at psychological barrier of 30 years. Also, a person with 6 years of work experience has probably more chances than an average 4 years.
3. It is not yet clear whether International applicants are fairly represented in this ranking. Since their GMAT scores are likely to be less because of language proficiency. Although, we can compare ranks among English natives and non-natives.
4. Another problem with internationals of TOEFL/TWE existing in paper and computer based incarnations. Problem is that we have maximums of 677 and 300 respectively. Under current formulae we have to divide actual score on it's maximum and we will have considerably different results in similiar scores. Say, I have 263 in CBT, which is claimed to be in 623-627 range of the old paper-based version. Hence we have the following: 263/300=0.877, and either 623/677=0.920 or 627/677=0.926, a considerable difference of at least 5.6% between marginal values. While this might seem find and fall within margin of error, we want to lower that margin of error! For this reason we may use TOEFL percentiles! You may be surprised but they do exist in some of the ETS published research papers I've got hold of! Percentiles are probably the best way to do that, at least it fits the GMAT counter-part used in this ranking system.
5. Male vs female thing. Females should clearly get a bonus as an admission officer may wish to improve schools statistics. Maybe 7-10% to a calculate score? Minorities should get a similiar bonus. Problem is how big the bonus should be.
6. We should either creat Qualitative rank or reflect somehow otherwise qualitative side of an applicant, since this plays important role. Problem is how can we evaluate essays? A terrific margin of error should be expected.
7. pretty much everything else I didn't think of yet.

Ideally, after collecting enough statistics in the main database we could find some neat coefficients, to adjust each and every case, thus giving somehow precise indication of a possible success in the application process. Therefore, we need as much as details submissions as possible! Act right now, submit yours and ask your friends to do so too! The value of the database might be really high!

Umm, I should have applied Wharton or MIT, well, it's never late. Wonder where did I apply? Check it out here.

Any comments will be appreciated via E-mail: Alex Chudnovsky (although I prefer either English or Russian languages in my correspondence you can use Spanish, German, French, Japanese, Hindu, or any other language, but do not expect I will understand you!) :-/