People's Democracy

Poll Khul Gayee?

 Nachiketa

 THE phrase in Hindi, Poll Khul Gayee [which means the cat is out of the bag], is perhaps most appropriate when applied to various pollsters sticking their necks out to give their numbers on Gujarat. Whilst India Today [ORG-Marg] announced that it was Narendra Modi who was the clear winner, Outlook was singing a different tune. The polling agency - C fore said that the BJP would manage only 80-85 of the total 182 seats. Another agency - CSDS, through Frontline agreed with India Today, and if one were to go by what they say, the BJP has a clear 15-percentage advantage! The Week, is trying its best to hedge its bets, and gives exactly, 85-90 seats to both, the BJP and the Congress. If the Election Commission was buying all these magazines, and J M Lyngdoh  wasn’t so busy, they may well feel inclined to simply do away with the polls! What’s the point, when it has all been settled and sorted by media gurus and psephologists. The thing to note is that different polls have given completely divergent results, and maybe that’s why the Election Commission has decided to go ahead with the polls!

This is not the first time that opinion polls have been carried out. It has been the trend for at least the past ten years, to try and lend authenticity to assessments and reportage. Even if one was to not dig way back into the past, but just look at how the pre-poll surveys dished out in the run up to the assembly elections in West Bengal and Tamil Nadu fared. Surveys by the same old agencies, suggested a “photo-finish”, a “close call”. But the real results, were completely out of line with the pre-poll estimates.

Psephologists and agencies in the pre-poll survey business defend their surveys vorciferously. It is all scientific; random samples of a suitable size are picked, and enthusiastic youngsters get busy trying to gauge preferences. But just think about this. For an electorate of 3 crore and 28 lakhs, The Week claims to have spoken to 3204 voters across 24 constituencies, and CSDS to just 1,775 voters, in 27 constituencies. Is that enough? You might argue that speaking to all these voters is a darned sight better than randomly trying to get a feel of the land and making rough estimates. But is it?

CHOOSING EASY ROUTE

The business of journalism, what the trade involves, and means - all this comes under review when pre-poll methodologies are discussed. The reporter with conversations at tea stalls, knocks on doors, and an attempt to visit real issues, is of course likely to go wrong. And sometimes seriously wrong. And anybody who has reported before on elections in India would vouch for how discerning the Indian voter is, and how well he/she has read the Indian constitution, zealously guarding the right to a secret ballot. Reporters have learnt by burning their fingers how little people disclose and how they are tempted to tell you what they believe you wish to hear! More interestingly, the person who talks the most may be least indicative of the general mood in that particular area.

Factoring all this in articles written about election trends is very tedious and difficult, and it is this that perhaps first made media czars to look to the `safe’, easy and perhaps a more fool-proof statistical route? But quantifying social trends, and trying to make a pre-poll assessment seem very precise - and therefore more credible, seems to be doing exactly the opposite. It is worthwhile remembering, that even in a far less complex, much more homogenous environment, with a two-party system like that in Britain, opinion polls have gone hopelessly wrong. Even the otherwise very cautious BBC, on the basis of opinion polls, had declared Labour victorious in two consecutive general elections, and ended up with egg on its face when Margaret Thatcher romped home with a huge majority! Experience taught the BBC to not allow or commission opinion polls in the future, and also to report very cautiously on polls done by others. The lessons learnt there haven’t put anyone on notice here.

It is easy to dismiss the cynicism about pre-poll survey fever as being old-fashioned, and inspired by a general hesitation to adapt to new technology but fighting this dependence on pre-poll surveys is not a negation of science. It is in fact an acknowledgement of the potency of scientific and mathematical tools, recognition of how hopelessly things can go wrong when a small error creeps into the sample. When some ordinary voters speak to pollsters as normally as they would to journalists, then instead of being able to pick up the nuances or discount for the complex calculations that the voter makes before he/she commits himself to the journalist - the formulae used for pre-poll surveys actually run the risk of building on the errors - exponentially at times. The result, ending up as a completely incorrect prediction.