Here is the Wikipedia page on Law of Large numbers. It defines the law as: "the average of the results obtained from a large number of trials should be close to the expected value, and will tend to become closer as more trials are performed." A variant of this (again from the same Wiki page), called Borel's law of large numbers states that if an experiment is repeated a large number of times, independently under identical conditions, then the proportion of times that any specified event occurs approximately equals the probability of the event's occurrence on any particular trial.
So, here is a question that IIT JEE would expect you to know the answer of.
If two persons, "A" and "B" pick up the same telephone directory, and randomly select 10000 persons each, and find out the average age of their group of 10000 persons. Which of the following statement is true:
- Average age of "group A" would be approximately same as that of "group B"
- Average age of "group A" would be substantially higher than that of "group B"
- Average age of "group A" would be substantially lower than that of "group B"
- The two random groups can not be compared.
If you are like one of our students, you would have marked the first statement as the right answer. Congratulations. You have learned the law of large numbers. But wait a minute. Let me test you further.
Let the person "A" pick up the telephone directory of Mumbai, and person "B" pick up the telephone directory of Delhi. Would that make a difference. If you answered that it won't make a difference, you are being reasonable, though someone must test it.
Now, let us consider a hypothetical city, SanghiNagar. The city has made a few interesting laws. It does not allow a phone to be in the name of anyone less than 21 years of age. Further, it wants all old people to have a phone, and hence it gives a tax rebate on all phone connections in the name of the people whose age is more than 60 years.
Now, let person "A" pick the telephone directory of SanghiNagar, and person "B" pick the telephone directory of Delhi, and select randomly 10000 names each. Would you expect the average age of the two groups to be same in this case.
If you are a student of an IIT, you would have answered it in negative. The reason is that the two populations are statistically very different. You would expect the average age of the selected group in SanghiNagar to be higher than the average age of the selected group in Delhi.
Now, ask your Director the same question. He will tell you how wrong you are. He will tell you that the law of large numbers will operate over two separate samples and give out the same statistical quantities, that once the quantities become really large, they cannot be statistically different.
When we teach in our classes, we encourage our students to question us, find errors with what we have taught in the class, and that is how our students learn, and we learn too. But, we can't question our directors.
If you take the group of CBSE students and the group of a state board students, can you say that the 90 percentile of CBSE group has similar academic learning as the one at 90 percentile in the state board.
Notice that state governments put in very minimal amount of money into school education. They don't have good infrastructure, they don't have enough teachers, and there is very little accountability. On the other hand, CBSE affiliates a large number of private schools, which are more serious in providing education, at least better than what a typical government school in most states would. CBSE also affiliates several central government funded schools (like Central Schools) which are much better managed, endowed, etc.
Are the two groups statistically similar. Let us look at the statistics. If you consider the number of "PCM" students in the country, CBSE only has a little more than 15 percent of them, but if you see the IIT students, more than 45% of the students are from CBSE board. Recently, I was in a meeting in one of the states. I was told that about one percent of 12th class science students in the state are from CBSE, while 98% are from state board. But if you look at the entrance exam results, the number of CBSE students in the top 1000 is several times more than 1%. (This is when a large number of CBSE students would not give the entrance exam, since they know that only 1% seats would be for them - they have a board based quota for admission.) We can check the same thing across the country in a variety of tests. And note that I am not talking about minor differences here and there. We are looking at a performance which is a multiple of others' performance.
And the reason is easy to see. As I said above, state governments are simply not putting enough resources in school education.
But our Directors have a different take. Every board is a large board. And their understanding of the "law of large numbers" is that all large populations will give the same statistical results, and any proof to the contrary will be put in to the waste paper basket.
And they can, therefore, go ahead and pronounce to the world that a 90 percentile in CBSE is same as 90 percentile in a state board, and therefore, the percentile marks can be used for admission purposes.
What is the impact of this. A board which has done reasonably well in terms of having a good curriculum, inculcating a decent pedagogy, insisting on minimal infrastructure in every school that they will affiliate, ensure a better attendance of students than other boards, carry out exams in somewhat better conditions, less cheating, and so on, and the board works to attract good schools to affiliate with itself, is going to be discriminated against, just because they deliver better quality.
The message being sent is simple, but yet effective: We want to reduce the quality of education in this country. If you are going to join a school affiliated to a better board, we will make sure that your chances of getting admission in IITs, NITs, and other CFTIs reduce. You still have time. Leave CBSE. Join your state board. With the same effort, you will be able to get a much higher percentile in a low-quality board than in a high quality board. So why join the high quality board. After all, you shouldn't be worried about school teaching any way. Whatever you need to learn, you will learn in a coaching class. It is just a matter of giving some exam. Give the simplest exam.
I am sure the Directors don't deliberately want to attack quality. They had good intentions when they were appointed Directors. The only problem is that they don't know the law of large numbers, but they are not willing to admit it.