Modi may well win, but some of the exit polls will still be very wrong this time

The clutch of exit and post-poll surveys may be agreed on one thing — that the BJP-led National Democratic Alliance (NDA) will either secure a majority or be within a whisker of one. But the numbers they have thrown up are diverse — ranging from 248 seats to a staggering 340. Such polls should do more than tell us that Narendra Modi is a shoo-in to become the next Prime Minister — they ought to be able to tell us, and with a fair measure of accuracy, how many seats would go to which front or party. By the last token, at least some of the polls are going to be proved horribly wrong; interestingly most, if not all, have hedged their bets by making their predictions on a broad band rather than a specific number. The polls also reflect a sharp variance when making predictions at the State and regional level. For instance, in Bihar, the UPA’s tally was pegged at 2 by one agency and 15 by another. In Rajasthan, the variance was between 1 and 14 for the UPA; in Orissa it ranged from 2 to 10 for the BJP.

This suggests that our exit polls — which are expected to provide an accurate reflection of voter intent — have still a long way to go. Predicting elections in a country of a billion people in which constituencies are divided between religious communities, castes, and people from extremely diverse social and economic backgrounds poses its own challenges. The proliferation of political parties has led to candidates winning with much less than 50 per cent of the popular vote in a first-past-the-post system, making conversions of vote shares into seats somewhat tricky in regions where there are multi-cornered contests.

But given the ‘hit rate’ of our exit polls — remember how badly off the mark they were in 2004 and 2009 — should we really attach so much importance to them? In a letter to the law ministry in 2010, the Election Commission argued: “The object and purpose underlying the restrictions on exit polls is to see that the electors are not unduly influenced or misinformed by such polls, which as the past experience during the last several years has shown were far wide off the mark.” But voters could also be influenced even if the polls were always or almost invariably accurate. In any democracy, voters are influenced by perceptions of how other voters may vote — one may even argue that it is vital for people to be allowed to form an opinion on how others may vote before casting one’s own. To restrict opinion and exit polls on the basis that this information, accurate or otherwise, can influence voters is to misunderstand the way electoral democracy works. If opinion and exit polls lose their relevance, it will not be through bans, but by their own poorcasting.

(This article was published on May 13, 2014)
XThese are links to The Hindu Business Line suggested by Outbrain, which may or may not be relevant to the other content on this page. You can read Outbrain's privacy and cookie policy here.