Correlation, Causation and superstitions

The chapter on correlation in my statistics course ended with a note which stayed with me for long. After learning all about the wonderful tools of regression and correlation, there I was presented with a WARNING : CORRELATION SHOULD NOT BE CONFUSED WITH CAUSATION.

Just to give you a flavour of the science I had learnt, correlation makes one capable of establishing if two variables move together in a particular direction. A positive correlation between say, population growth and pollution implies that when any one of the two variables increases, the other also increases since the variables are positively correlated. Yet this does not by itself imply that any one of them CAUSES the other. If X and Y are positively correlated, X may be causing Y, Y may be causing X or it may just be possible that both X and Y are being caused by a third exogenous variable!

A kingdom was once plagued by a pandemic that claimed a zillion lives. Tense, the king ordered a statistical enquiry. The statisticians presented the data - more the no. of doctors in the kingdom over the days, more is the occurence of the disease. The king took this relation of correlation for causation and ordered all the doctors in the kingdom to be executed!

Now lets apply the lesson - why do people believe in superstitions? A cat crossed the road before them, its a bad omen - they say out of experience. So if a sweet little pussy cat decides to cross the road just exactly when our jinxed hero steps out of his house, it makes him anticipate failure in his exam or the interview ahead. What explains this seemingly irrational and undoubtedly foolish viewpoint? Well, my theory goes thus. In the 'good' old days, it might have happened that in a good number of cases that the correlation between the 'crossing of a cat' and the 'occurence of failure' turned out to be positive, but people mistook the correlation for causation and concluded "Hence, the cat CAUSED the failure"!!!

Paul Samuelson listed something called a 'Post Hoc Fallacy' right at the beginning of his classic 'ECONOMICS'. The fallacy goes thus - If event Y follows event X, then you implicitly assume that Y was caused by X. If I sneezed before I slipped from the stairs, then the sneeze caused me to trip.

I guess our society needs to learn just a simple lesson and all the claptrap about black threads, wearing stones, reading horoscopes shall see the light of rationality.

P.S. If while reading this blog, lets say you slip off your chair ... kindly don't conclude that I caused it :P heehee :)