Friday, August 02, 2013

A POLLSTERS’ REPUBLIC - Opinion polls as a business opportunity

By Swapan Dasgupta

As a precocious and somewhat impressionable 15-year-old in Calcutta, I was enthralled by the political drama that was being enacted in 1970-71, between the Congress split and the general election. I devoured all the political news from The Statesman and the Madras edition of Indian Express (which my father used to receive at his office), and I also followed the political jousting between the two Parsi rivals, R.K. Karanjia of Blitz and D.F. Karaka of Current.

This over-reliance on the printed word, which I also assumed to be the definitive version of politics, led me to the conclusion that the so-called Grand Alliance would prevail over Indira Gandhi in the 1971 election. Consequently, I was in a state of shock when the results started trickling in and indicated that the Congress-R (as it was then called) was heading for a conclusive victory.

The experience shattered my faith in the infallibility of the Indian press and prompted a healthy scepticism of the election assessments of journalists—a scepticism that persists.

Mercifully, I soon realised that gullibility wasn’t my exclusive preserve. Throughout the 1980s and till the 1991 election, when the reach of TV was patchy, I recall being accosted many times by political workers in the localities and the idlers in the tea shops of eastern Uttar Pradesh with the information that the BBC had predicted that so-and-so candidate was winning in such-and-such constituency. It was never clear to me why the BBC (which enjoyed a formidable reputation in those days of media squeamishness) would even care to forecast the results of some obscure constituency where there was no one of any consequence contesting. But the fact remains that the phrase ‘BBC ne bola hai’ resonated throughout the Hindi heartland at election time and added to the general tamasha.

I guess Indians have become more sophisticated. The myths surrounding BBC psephologists in distant London are no longer in circulation. Instead, political buffs are inclined these days to sit for hours before a TV set watching the election returns from opinion polls hosted by excitable anchors. From the mid-1990s till the 2004 general election which upset all pre-existing calculations, the last word belonged to the India Today and NDTV polls. As the media has mushroomed, opinion polls and exit polls have multiplied with the added complication that each TV channel believes their poll constitutes the last word on the subject. Today we have the somewhat ridiculous spectacle of spokespersons of the political parties being grilled and taunted by aggressive anchors who take it for granted that their poll findings constitute the final results.

It is not merely the media that has succumbed to political forecasting as entertainment, each of the major political parties spend significant sums of money engaging in-house pollsters who advise them on everything from candidate selection to constituency-wise resource allocation. Over the past fortnight, for example, I have been forwarded opinion polls conducted by a political party which not only indicate levels of popular support, identifies issues but even suggest the choice of candidates for particular constituencies. Polls, it would seem, have become an additional input in the larger process of factional lobbying.

At one level, the departure from purely instinctive politics to a more focussed approach is welcome. India hasn’t quite reached the levels of parties in the more advanced democracies where tactics and strategies are decided by the feedback from focus groups, but it is clear that gut feel no longer counts as much as it did earlier. But this shift to a more ‘scientific’ orientation has also generated a tribe of charlatan pollsters who manufacture polls to suit conclusions that politicians have already arrived at. In an ideal world, pollsters are meant to convey ground realities to their clients with a premium on accuracy; in India there are just too many exercises that are aimed at conveying good news.

This disingenuity is also a consequence of large-scale ignorance of pollsters on polling methodology. Just as the hallmark of good leadership lies in the ability to separate bad advice from good, political leaders must possess the ability to distinguish between polls with robust methodology from polls that merely boast a gigantic sample size.

Some years ago, a polling organisation claimed that random sampling (preferred by statisticians) has no relevance in India and should be replaced by the ‘cluster’ method. On probing, I discovered that the ‘cluster’ method was a euphemism for polling people at bus stops and tea shops—the preferred method of journalists. Indeed, there are polls which rely almost exclusively on the aggregate of journalistic assessments, a methodology that has the advantage of cutting costs. One canny pollster has actually dispensed with fieldwork altogether and depends almost exclusively on private information channels—much like the Intelligence Bureau and satta market assessments that make their way to the news reports. What is astonishing is that he often gets the actual results right.

Of course, there are legitimate problems in translating popular vote share into seats in a fractious, multi-party environment. Although some people have developed mathematical models that minimise errors, their findings are invariably tempered by the wisdom of the editors of media organisations. Only too often, and immediately after an unexpected outcome in the actual elections, I have been told by ‘editors’ that the statistician had got it right but that the outcome had been ‘toned down’ on the strength of gut feeling of the editors.

Likewise, it is accepted by all reputed pollsters that the larger the area of consideration, the greater the chances of getting the aggregate outcome right. Yet, it is routine for pollsters to present political parties with constituency-based projections. Recently I saw an all-India survey that even suggested a seat tally for Goa, a state that sends two MPs to the Lok Sabha.

What seems sufficiently clear is that opinion polls have become a business opportunity for both the genuine and the carpetbagger. Some political parties have also concluded that “paid polls” should also complement “paid news” at opportune moments. In 1999, for example, a well-known media group in Maharashtra allowed itself to be used to publicise a clearly dubious poll showing that the Congress would emerge as the largest party. And in 2002, a weekly magazine allowed a pollster to transfer the entire bag of uncommitted votes to the Congress in Gujarat.

The tendency of Indians to subvert any worthwhile endeavour is well known. But the answer to this problem lies in clients becoming more discerning. It does not lie, as the government seems to think, in pressuring the Election Commission to ban all opinion polls from the day of the notification—which can be a full two months before polling day. Such a draconian order will drive motivated polls underground and we may be confronted with a situation where wild rumours, based on internet postings which can’t be controlled, have a field day. There is insufficient evidence to suggest that opinion polls mould voting intentions in any significant way. And the present over-zealousness of the Central Government, it would seem, stems largely from a desire to underplay the massive negative ratings of the UPA-2.

There are already restrictions in place for the publication of opinion and exit polls. These are worthwhile restrictions and there is no need for further curbs. What is needed is for politicians to educate themselves on the design and interpretation of polls, and for TV anchors to undertake a crash course in the responsible dissemination of poll findings. 

The Telegraph, August 2, 2013

No comments: