This is the fourth year of us using the CommunityRank methodology for ranking India's Most Preferred B-schools. The methodology is explained in detail below. Before you dive into the methodology (which we'll appreciate if you read completely --- from top right till bottom before forming an opinion), here are some salient features of what was under the hood in the 2011 PaGaLGuY B-school Rankings.

1. The underlying belief – "Perception captures everything"

"XYZ Business School placed all of its 120 students in top notch companies within 7 hours"

"ABC School of Management was fined by AICTE for irregularities in fee structure"

"Five people from PQR Institute of Business have started their entrepreneurial venture even before graduating"

"A severe ragging case on campus has put MNO Management Institute under fire"

"123 School of Business closed its CAT cut-off at 97 percentile, thus elevating quality of students over previous year."

These are only some of the news snippets that stakeholders in the MBA value chain today constantly read and see in the media, share in hundreds of online communities and message boards and hear from their friends, relatives and those in their professional networks. There is such an abundance of avenues for information and knowledge exchange that today's MBA students, MBA aspirants, recruiters, academicians and other stakeholders are a lot more informed about which business school they would prefer for their purpose than they could have been 10 years ago.

All this information adds together in an abstract manner to create Perception, and it is sharpened perceptions that eventually leads to decisions — whether of taking an admission, or making a job offer. Today's MBA stakeholders have ever-strengthening perceptions about the best known business schools. And they actively use all available information resources to evolve that perception.

In short, perception captures everything.

As a result of sharpened perceptions, here's a question most MBA stakeholders always have an opinion on: "Which of the following b-schools do you think is better — X or Y?"

Taking any two B-schools at a time from a long list, they mostly always have an opinion.

It is this hidden treasure of perceptions among a large population of MBA stakeholders that the PaGaLGuY B-school Rankings tap. Consequently, these are the rankings of India's Most Preferred B-schools, and NOT India's Best B-schools.

2. A rankings of the 'Most Preferred' B-schools

"Why not an objective data-based rankings?" might be an obvious question in your mind. We have covered this in detail in the previous years, so the best way to read the answer to this question is to read this article.

That said, a rankings based on perception reflects what people prefer rather than what is the best. The PaGaLGuY B-school Rankings is hence, a rankings of the Most Preferred B-schools in India.

Before you start reading through the methodology, we would like to mention a few features of the rankings and explain where they fit into the scheme of things for an aspirant, so that it helps you understand the survey better.

• The PaGaLGuY B-school Rankings are not a rankings of the Best B-schools, but of the Most Preferred B-schools.
• They serve to inform aspirants of what their peers, current MBA students, Alumni, Women, Freshers, Experienced people individually think about rankings in order for them to make a better decision about which b-school to apply to or join.

3. A step by step description of the process

In Stage 1 of the ranking process, we published a list of 100 best-known b-schools on the home page of PaGaLGuY as the 'B-schools in consideration for the PaGaLGuY B-school Rankings' on October 17, 2010 and invited representation from stakeholders to make additions or deletions from this list. Two questions were asked in the representation,

1. Which B-school would they like to remove from the list and why. They also had to compulsorily suggest an equal number of b-schools to replace the ones they wanted removed.
2. Which B-schools would they like to add to the list.

We publicized this list extensively through our channels, feeds and email newsletters, which have a combined reach of over 5 lakh people per month. And we did get a representation of over two hundred emails. Many representations were copy-pasted emails sent by entire batches of lesser-known B-schools. These we ignored. Several others were genuine requests and suggestions from other MBA stakeholders to add or subtract B-schools, or inform us of name changes of their B-schools.

Although we had set November 5 as the last date to accept representations for changes in the b-school list, we continued taking representations until as late as December 1, 2010.

In the end of Stage 1, the list of b-schools in the rankings increased by four to 104 b-schools. These are the b-schools which stood to be ranked in the PaGaLGuY B-school Rankings 2011.

In Stage 2, we shall crowdsource the comparison of these 104 b-schools through a perception-based survey.

Brief explanation of Stage 2

To explain the methodology in a single sentence — The survey makes respondents compare each of the 104 b-schools with every other multiple number of times using a simple question: "Which one of the two b-schools do you think is better?". Each time a b-school is voted as better over another, it adds one point to its score.

In the end, all the b-schools are arranged in decreasing order of their scores and voila, there are the rankings.

The number of legitimate respondents taking the PaGaLGuY B-school Rankings survey 2010 was 9,576. That is approximately 16 times the sample size of the next largest perception-based b-school rankings in India conducted by a well-known business magazine (comparison with other magazine rankings is invalid as they use data provided by b-schools). However, we have observed in the past three years of this ranking's existence that after about 600-700 respondents have taken the survey, the rankings more or less stabilize and the movements between ranks thereafter are limited to +/-2 ranks. In that sense, we have been surpassing the methodology's cutoff point by thousands of respondents. It also proves that the methodology accurately captures the perception of stakeholders, which is further corroborated by the finding that the rankings calculated by taking two random samples of 1,000 respondents each from the total is the same.

Before calculating the final rankings, we undertake a thorough data-cleansing exercise, wherein bogus or duplicate respondents are identified using a vast pool of technological resources and built-in security functionalities of the survey. Every year, we come across bunches of over-enthusiastic b-school students who try to circumvent the survey system and vote multiple times to get their b-school up in the rankings. While a very vital feature of CommunityRank insulates the survey from duplicate voting to a great extent (read Detailed explanation of Stage 2 to understand this feature), we still identify and delete hundreds of duplicate votes from the system before releasing the survey, so that it is free of taint.

Last year, about 15 students in total from two well-known b-schools created a total of 50 fake duplicate votes and laid the votes of about 500 genuine voters waste. For indulging in such an unethical frenzy, we disqualified both the b-schools from the PaGaLGuY B-school Rankings 2010 in full public view.

We shall not shy away from disqualifying b-schools in this year's survey too, should such behaviour be identified again. So our advice to b-school students taking this survey is — Don't even think about it.

Detailed explanation of Stage 2

Like we said, CommunityRank captures the perception of MBA stakeholders about b-schools by making respondents compare each of the 104 b-schools with every other multiple number of times.

To understand how this works in detail, let's take a simpler example. Suppose that you have six items, say A, B, C, D, E and F, which you want to rank according to which is better using CommunityRank.

To compare each of the 6 items with every other once will result in a total of 15 comparison questions. This is better explained in the following illustration.

We can create a questionnaire out of the above 15 comparison questions and pose it to a respondent. After each comparison question is answered, the item chosen by the respondent as 'better' gets one point added to its score.

After all the 15 questions are answered, each item would have accumulated a certain score. Arrange the items in the decreasing number of scores and you have the ranking ACCORDING to this particular respondent.

You can make more number of people answer the above survey and keep updating the scores of each item. The more the number of people answer the survey, the more representative the rankings will become of a large population. You can replace the six items with the names of six b-schools, and you will have the rankings of the six b-schools.

That was a 'clay model' version of the PaGaLGuY B-school Rankings methodology. The actual rankings compared a total of 104 b-schools. And this is where some complexity sets in.

Ranking 104 items by comparing each item with every other will require a total of 5,356 questions. While anyone can answer a questionnaire with 15 questions, almost nobody can humanly answer a questionnaire with 5,356 questions!

So here's what we do. We factorize the number 5,356 and discover that 5,356 = 103 x 52. With this new insight in hand, we randomly split the pool of 5,356 questions into 103 questionnaires with 52 questions each. Read that again.

Each time a new respondent clicks on 'Take Survey' in the PaGaLGuY B-school Rankings system, we randomly serve one questionnaire out of these 103 to him/her. When the next respondent clicks to take the survey, we serve him/her another randomly selected questionnaire from the pool. As with the previous example, each time a respondent chooses a b-school to be 'better' in his/her perception, we award one point to that b-school's score. Gradually, there comes a time when 103 people take the survey and each of the 103 questionnaires have been answered. When the 104th respondent selects to take the survey, we restart the cycle and serve him a randomly selected questionnaire comprising 52 questions from a fresh pool of the 5,356 questions. And so on.

As you might have figured out, 103 people collaboratively answer all the 5,356 questions. Or, 103 people together do the same task that one person did in the 'six items' example.

After 1,030 respondents anser the survey, each of the 104 b-schools has been compared with every other 10 times. After 10,300 respondents, the comparison cycles have repeated 100 times.

Such a system gives us two really good fringe benefits.

1. It keeps the survey tamper-proof. B-school students who are hoping to mass-select their b-school as 'better' in all the comparison questions of their questionnaire might be stumped, because their randomly selected questionnaire just may not contain even a single question comparing their b-school. Or have a very limited number of questions containing their b-school.

   Of course, despite the above measure, we still look for duplicate voting in the rankings and cleanse the data of such redundancy before releasing the final rankings.

2. Demographic rankings. We can select a demographic sub-section of the respondents and see how they ranked the b-schools. For example, we can create cycles out of only the women respondents from the survey and see which b-schools women prefer the most. Or how do those with work experience rank the b-schools differently from the freshers. There's lots of data to mine, lots of valuable insights into how India thinks about its b-schools.

Finally after the survey ends, we are left with a large collection of completed comparison cycles. We then cleanse the data of duplicate voting, calculate the score of each b-school and after arranging the b-schools in decreasing order of their scores, we get the PaGaLGuY B-school Rankings.