By Ashutosh Varshney

All opinion polls have predicted that the National Democratic Alliance (NDA), headed by Narendra Modi, leader of the Bharatiya Janata Party (BJP), will form the next government in India, even if it does not get an outright majority of parliamentary seats. While this may well come true, we need to ask some questions about how precisely opinion polls can predict winners in India’s parliamentary elections.

Let us begin with the record of opinion polls in recent years. In 2004, the “Shining India” campaign of the NDA was expected to win hands down; it was soundly defeated. In 2009, virtually no one gave the Congress party even the slightest chance of winning more than 200 seats: it did. Predictions have gone wrong even in the assembly elections of larger and complicated states, such as Uttar Pradesh (UP). Few, for example, anticipated in 2012 that the Samajwadi Party (SP) would win a majority of seats. In contrast, polls correctly predicted recent winners in the states of Gujarat, Rajasthan and Madhya Pradesh.

Why should there be such variance? An answer to this question will also give us clues about the current election scenarios.

First, the basic problem of predicting India’s parliamentary elections remains unresolved: no formula has yet been found that can precisely convert votes into seats. The best polls manage to get the vote percentages of the parties right, but there are so many regional, even local variations that an accurate transformation of votes into seats defies even the best statisticians. Statisticians have worked on this problem for decades without succeeding.

Predictions are, of course, easier when electoral battles are basically bi-polar, as was true of Rajasthan and Madhya Pradesh elections in 2013 (though there, too, the extent of the BJP’s victory was underestimated) and Gujarat in 2012. In all three, the BJP and the Congress had a straight fight. But UP in 2012 had a four-cornered battle, as it has had since 1999.

In India’s parliamentary elections, the five biggest states today – UP, Maharashtra, West Bengal, Bihar, Tamil Nadu – are all tri-polar or more, as are some of the middle-sized states such as Karnataka, Seemandhra and Odisha. These states account for 320 of India’s 543 Lok Sabha constituencies. No opinion poll is so large that it can have a representative sample of each of these constituencies that could, at least in principle, allow precise predictions.

Even if such huge samples could be drawn, a second complication is strategic voting. Many voters wait till the last minute to judge for themselves which way the wind is blowing and then vote either with the wind, or go with whichever party is best placed to beat the party they dislike most. If Muslims, for example, are feeling anxious about a BJP victory, who would they vote for in UP, Bihar, Karnataka and Telangana: BSP, SP, Congress, JD(U), JD, RJD, TRS? The answer is not likely to be the same in every constituency.

This example is not simply theoretical. In an estimated 200-odd Lok Sabha seats of India (out of 543), Muslims tend to account for 10 per or more of the electorate. As the prospects of Modi’s victory rise, India’s Muslims may well elect to vote strategically, complicating the final tally enormously.

Since 1989, India’s parliamentary elections have come to exemplify a core principle of democratic theory: that there should be uncertainty about election results. The uncertainty is, of course, not radical, for we do know that the incumbents would lose this time. Democracies tend not to elect the incumbents a third time. Being in power too long almost invariably sullies the record.

The record of the Congress and UPA, too, is full of blemishes. Extraordinary levels of corruption have cost it a great deal of popularity. The other big issue has been the lack of correspondence between the power and governance structures. Prime Minister Manmohan Singh is universally admired for his personal integrity. But he had no great political stature and legitimacy. He never won a direct election. Being from the Upper House, his subservience to the Nehru-Gandhi dynasty, entirely expected, hurt governance immeasurably. The great surprise is that he lasted as long as ten years in power.

In short, while the UPA will almost certainly lose, an NDA government is by no means certain. Indeed, the greater the possibility that the NDA might win, the higher the chances of strategic voting by the minorities and those Hindus who do not like Modi. After May 16, one should not be surprised if intense political bargaining precedes the formation of government.

Ashutosh Varshney is director of the India Initiative and professor of political science, Brown University; and author, most recently, of Battles Half Won: India’s Improbable Democracy (Penguin).

Related reading:
Guest Post: the rural/urban divide dies out, beyondbrics
Guest Post: working class drive EM consumer growth, beyondbrics
Guest post: lack of party alliances risks instability, beyondbrics

Get alerts on Markets when a new story is published

Copyright The Financial Times Limited 2020. All rights reserved.
Reuse this content (opens in new window)

Follow the topics in this article