Imagine that you are travelling through an Indian village during the monsoon season. As you are slugging your wet shoes through the sticky mud on the road, you feel a sting on your ankle. You look down and find a small stream of blood trickling out of two bite marks. Looks like you have been bitten by … by what? A harmless snake, a scorpion, or a king cobra?
I learned that statistics can be rather dangerous, and one must be very careful about how to use it.
In panic, you look for a doctor. The villagers inform you that there are two doctors in the village. Both treat bite wounds. Doctor A has a lousy track record: about 90% of all his patients die, usually on the spot. Doctor B has a much better success rate: about 90% of his patients live. Faced with these stats, and with your IIT education in math, logic and statistics, you would choose Doctor A, right? If you did, you would be correct, making Professor Gopalan proud. Alas, you would also be dead!
This brings me to my story on statistics in an Indian village in western Maharashtra where I went to school for a few years. And in this village, I learned that statistics can be rather dangerous, and one must be very careful about how to use it. My village was a rather primitive place, but fortunately we were blessed with two doctors in the village. The first one was a witch doctor (Doctor B), who healed people with prayers, songs, dances and offerings to the appropriate gods. The second one (Doctor A) was my grandfather, who had a medical degree from the University of Calcutta.
Even though there were no radios or faxes or e-mail, somehow we kids always knew when someone was brought to the witch doctor, and we all gathered around to watch the show.
Early in the month of June every year came the monsoons, bringing with them a welcome relief from the heat and humidity that was building up for the last few months. They also brought with them lots of rain, filling up the snake holes with water. So the snakes would leave their snake holes, and seek shelter elsewhere. Every now and then, a villager would step on one of these snakes, and the snake would react by biting back.
Every time this happened, the villagers had two choices: they could take the victim to Grandpa, or they could take him to the witch doctor. In their hearts they believed in the witch doctor more than they believed Grandpa, so they would take the victim to the witch doctor first. Even though there were no radios or faxes or e-mail, somehow we kids always knew when someone was brought to the witch doctor, and we all gathered around to watch the show.
And it was a splendid show! There were prayers to the spirits to lift the curse from the victim. There were songs, and dances, and sacrifices of chickens. The spirits almost always demanded that chickens be sacrificed. If the victim or the folks who brought him in were better dressed than normal, the spirits demanded a goat. It was an awesome site to see a bunch of chickens with their heads cut off, running in circles.
This whole thing lasted for a few hours. And, amazingly enough, it worked most of the time! At the end of the ceremony, the victim would generally sit up, and even stand up and start walking.
So this was my lesson in Statistics 101, well before I heard of the word Statistics: be suspicious of just following numbers
But, of course, in one out of ten cases, the spirits were too angry, and would not agree to lift the curse. At the end of two or three hours, they would tell the witch doctor in no uncertain terms that they would not lift their curse, and that the poor fellow had committed sins too grave to be forgiven. When this happened, the witch doctor would turn to the victim’s relatives, informing them that the spirits were angry, and that there was not much he could do. Perhaps they could take him to the other doctor.
The whole crowd would then move a couple of blocks to Grandpa’s house. Most of the time, when they brought the victim in, he would be half dead.
Grandpa would examine the victim, check his pulse, and proceed to mix a shot of polyvenom serum, the combined antidote, and give it to the victim. Most of the time, it was useless. The victim would be too sick and his heartbeat too low to let the antidote have any effect. About 9 out of every 10 victims brought to Grandpa died.
So here were the undisputed statistics: 90% of all the snakebite victims brought to the witch doctor were cured, whereas 90% of all those brought to Grandpa died! Everyone knew this, including Grandpa. Now, you may not give the villagers much credit for knowing a lot of science and biology and such, but they knew their statistics, and it clearly told them that, statistically speaking, their chances were much better if they went to the witch doctor after a snake bite than to Grandpa. So that is what they continued to do, year after year.
This, of course puzzled me a great deal, so one day I asked Grandpa about it. Why is it, I asked, that 90% of the witch doctor’s snake bite patients live, whereas 90% of your die? Well, said Grandpa, it may have something to do with the fact that 90% of all the bites in this area are from non-poisonous snakes and mildly poisonous scorpions. The effect of their poison lasts only for about 2-3 hours, but the other 10% can be fatal, if not treated early. So the patients who come to me are those that are bitten by poisonous snakes, and then not given treatment for several hours.
So this was my lesson in Statistics 101, well before I heard of the word Statistics: be suspicious of just following numbers. Understand the phenomena underlying the numbers. And, finally, if the numbers don’t make sense, question them.
1 comment
As Mark Twain said:
There are three kinds of lies: lies, damned lies, and statistics