Monday 18 May 2009

Why the surveys were all wrong

Apparently the Congress win in the recent was not predicted by any pre-pool or exit survey. I find that rather difficult to digest. However, let's say that is indeed the case. How can it be possible? Two possibilities come to mind:
All the surveys were badly done. They didn't go for representative samples; instead they just interviewed people near their homes and hotels.

Many seats were decided by the narrowest of margins. This was reflected in the surveys, but was left of by the publications. The newspapers and news channels left out the uncertainty so that the innumerate millions are not confused.

Also, the Congress, because it contested more seats this time (thanks to Ekla Cholo), won more of those marginal seats. A person who calls many tosses is almost sure to win more (numbers, not percentage) than one who calls few.

And, hey, where's the landslide? Congress & Co are 11 short of getting a simple Vote of Confidence. We don't have a hung parliament, but Vajpayee's and Dr Singh's terms show that that doesn't necessarily mean fatal crisis. Let's get real for once.

PS: One thing I heard no commentator talk about was the inordinate length of this election, the consequent presence of central security throughout the country, leading to free and fair (one assumes and hopes) voting. We all agree that corruption is a great ill, yet discuss matters as if it is not even a factor!

Second, I am sure someone has tried to find the role of luck in elections in our country. Very simply, we may look at the correlation between vote shares and seats, after adjusting for state size (In a state with one or two seats, the correlation will probably be weak). Do political parties employ statisticians? Shouldn't they?

No comments: