A few minutes into watching primetime shows dissecting various exit poll numbers, one might ask: should exit poll results be just stated as information or analysed with reflexes that range from rejection and ridicule to an air of finality? Indian TV news media doesn’t seem to have made up its mind on that. That’s at the root of the muddled responses to exit polls on news television in the country.

As stopgap exercises in gauging voting trends, before the counting of actual votes by the poll body, exit polls feed an insatiable information cycle. Suddenly, months of ground reporting, anecdotal assessments and speculations are replaced with estimates in numbers. Its appeal as a pre-results build-up is derived from the numerical expression of voting behaviour—observations that could be flawed, accurate or approximate. The point is that irrespective of its methodological limitations or adequacy, exit polls present a number: the only time-filler news that consumers are looking for at the end of voting and in anticipation of actual results.

Last evening, there were, as could be expected, enough sceptics of the surveys predicting a comfortable win for the Bharatiya Janata Party-led National Democratic Alliance. While it is routine to hear dismissive responses from leaders of parties predicted to be on the losing side of the exit polls results, journalists mocking the polls and basing a primetime show on it only baffles viewers. Such scepticism about the figures could be better expressed by just limiting the exit poll figures to a piece of news and not conducting a panel discussion on it.  Such discussions can wait. In any case, no one is complaining about the lack of discussion shows on Indian news TV; instead, the problem might be too many of them.

Moreover, if a journalist has a different assessment of the popular pulse among voters, it should be expressed in numbers on a day when numbers are being talked about. In the election cycle, the time for “ground feel” is over, the slot is for estimated numbers now, whether inaccurate or accurate.

For instance, last evening on NDTV India, Ravish Kumar was anchoring a primetime show where “entertainment” was constantly being used to refer to the exit polls. It may very well be entertainment, as all prediction exercises are, but then a 60-minute discussion on “entertainment” clearly legitimises it with gravitas and importance. At one point, while discussing exit polls results from Bihar giving the NDA 29-34 seats out of 40, Kumar referred to how it is at variance with ground assessments he had gathered. However, unlike Kumar, many people in Bihar may not be surprised with these exit poll predictions. Even if one believes Kumar’s input, the conversion of such assessments into numerical estimates would have been useful after the conclusion of seven phases of polling.

On the same show, author and freelance journalist Revati Laul chose to offer her anecdotal account of a voter mood in Uttar Pradesh which was at odds with the exit poll predictions. Again, she may or may not be right, but what was obviously wrong with her take was the fact that anecdotes aren’t psephological data. She could have saved the former for another day, since May 19 belonged to a different exercise. Instead, the exit polls time slot made a detour to discuss unemployment, farm distress and other issues of policy analysis. On a day of seat-estimates discussions, such scrutiny could have made sense for viewers if it was supplemented with counter-data on seat share predictions. A numerical tabulation of such anti-incumbency trends could have been Laul’s contribution to the discussion. Yesterday wasn’t the evening for her anecdotes.

A psephological process requires a decent sample size, its inclusiveness of various sections, coverage of different areas of voting and the challenge of converting predicted vote share into the number of seats. Even with all that, inaccuracies can creep in because of a lot of factors—the daunting task of projecting seats from estimated vote shares being the most important of them.

On other channels, even channels that had commissioned exit polls like CNN-News18, India Today, Times Now and News 24, there was a high frequency of qualifying the predictions with warnings of the possibility of them being wrong. That anxiety is understandable, as the actual results may always leave the pollsters with egg on their faces. But if that anxiety is so palpable, why dedicate hours of analysis and time to figures that you’re accepting with a bucketful of salt? Either the figures could be explained by talking about the methodological rigour with which they have been arrived at, or just left for media consumers to make sense of.

Many media organisations, including online media and English TV channels like NDTV, insert a quirky disclaimer about exit polls while dissecting them, such as “health warnings” that exit polls often go wrong. Though the advisory is necessary and valid, pollsters in India may contest the use of the term “often”. That’s because, as Dr Prannoy Roy and Dorab Sopariwala’s book The Verdict recently said, there is an accuracy rate of 84 per cent for 431 exit polls conducted in India in the last four decades (1980-2019). More significantly, with the exception of the 2004 polls, the accuracy rate for exit polls for Lok Sabha elections in the same period is 97 per cent. There is obviously a case of survey agencies with less and more successful track records of predictions, though Roy and Sopariwala avoid naming them while choosing to rank them anonymously.

News channels commissioning such polls would do well to educate their viewers on how they just partnered with professional agencies to conduct surveys and aren’t conducting the surveys themselves. On Aaj Tak and News 24, when party leaders projected to be facing a grim verdict on May 23 lashed out at biased exit polls, the news anchors should have explained the nature of their channel’s limited role in it and the methodology was used for projections. In fact, even if charges of bias are accepted, incentives for giving agenda-driven projections aren’t obvious in exit polls because they have a short life of only three days before being proved or refuted.

While possible methodological issues may still falsify predictions and spoil the celebratory mood in the NDA camp, the chances of underestimation bias against the eventual winner (NDA, according to most exit polls projections) might push the UPA to a landslide defeat. Both seem probable in their own ways, though the chances of the former happening seem dim. Whatever be the case, May 23 will tell us. Till then, the news media should decide what it wants to do with exit poll numbers. There seems to be no mood to simply register the predictions; the media keeps talking about the numbers while telling you they’re wise enough not to trust the figures. That’s part of the problem.