Lok Sabha polls: Exit polls have little scientific basis, reject them

Numerous exit poll results are invading the news space and it is again a field day for the so-called election result experts. On each popular TV channel, they will sit for hours and give ‘Gyan’ to the public as to why the predictions are correct or incorrect. This farce will continue till May 23. Please enjoy the drama but don't believe a word you hear or a figure that you see on TV.

I will tell you why. Let’s assume an exit poll is being carried out with utmost honesty, the voter has given honest information of who he has voted for and the data collected is fed to the computers without any mistakes. Finally, a honest statistician, who knows his job, whose assumptions are statistically correct and who uses the most modern tools to analyse the data, analyses the data.

Even in such conditions, scientifically and statistically the soundest, standard error would not come down under 5%. Some statistician may also put the figure of standard error to 10% in such polls.

No, I don't want you to get intimidated by these technicalities. What it means in simple terms is that if I repeat the same survey using the same methodology, choosing the same respondents, with the help of same enumerators, the difference in results could be about plus or minus 5%.

Which means that about 10% of those who have said that they voted for party A may say that they voted for party B. In simple terms, a party which has been estimated to get 30% votes may get at the most 35% or at the least 25%. Do you realise that with this margin of error, nothing worthwhile can be predicted.

The Indian society is highly stratified, and these non-serious surveys do not take into account caste, religion, regional and linguistic biases. In other words, they consider the universe homogeneous which it is not by any stretch of imagination.

Most of the credible surveys, conducted on much less complex issues, meticulously stratify the sample and the smallest representative sample, so computed, is never less than 35,000 for Indian population.

I had harped on honesty right at the outset. Everyone will agree with me that honesty is not a strong point of the Indian society. A person may have given vote to a Congress candidate but when he comes out and sees BJP goons standing at the booth gate asking whom he had voted for, his prompt answer would be the BJP.

An enumerator, who is sympathiser of party A will also try to tilt results in its favour. Even if the enumerator does not have any political preferences, he will try to judge the inclination of the agency and record fudged data accordingly. Corruption and muscle-power are also important factors in our electoral system which cannot be included in a normal statistical analysis.

Instead of believing the exit polls, you should probably get out of the house and ask the first roadside baba you encounter to predict who is going to get how many seats. His prediction has a higher probability of being correct because he might have interacted with better-informed people.

Any psephologist, worth his salt, like a jyotishi, does not believe in his own predictions but he does everything to make you believe what he is predicting has a basis and that the basis is scientific. He pulls every trick out of his hat, makes you believe that data has been collected from a large cross-section of the society, sophisticated computer programmes have been used, statistical analysis has been done by experts, etc. But remember that even jyotishis have started using computer programmes.

However, in the era of EVM tampering, exit polls are important. They are intrinsic part of the political propaganda and tell you two things. One, who has invested more money to purchase services of exit poll survey agencies. In some cases, threat work better than financial benefits. The other aspect of the exit polls tells is far more serious.

It helps you understand which political party has tried to tamper with the EVMs and must use propaganda to make the country believe that the final results are correct. If party A has targeted 300 constituencies which happens to be an unexpected figure and people’s estimate is that this party will get only 150 or at the most 170 seats, then it is imperative to purchase as many survey agencies as possible to circulate a figure close to 300 seats.

Most exit polls on TV are giving close 300 seats to the BJP. Hasn't Amit Shah been saying for the past two months that his party will win 300 seats? Does it not sound like his master’s voice, being played by all the channels? But the beauty of tampering EVM is that even the best thieves cannot be 100% sure of results.

The reason is simple. In Indian conditions, whichever method of tampering you may use, it is impossible to steal all the votes. As per my estimates, tampering is done to target about 5 to 10% vote shift, but if the swing against the party that has mastered vote-stealing techniques is more than 10%, then you can still lose the seat and your prediction may go off the mark.

One of the important functions of exit polls is that they are used as an effective tool to demoralise the opposition. In the absence of such propaganda, unbelievable results may trigger people’s anger. Don't let anyone play with your emotions. Reject the exit polls, they have no scientific basis.

Realise that these figures will make only Bhakts happy. If the results are what is being shown on the television, then the election process has been compromised completely. Let no one convert your anger against subversion of Indian democracy into despair. That is the aim of most survey companies.