Before the Ballot Act of 1872, voters in England had to mount a platform and proclaim their electoral preferences to the world at large. While the secret ballot has deprived us of this entertainment, the modern pollster has restored some semblance of it on our television screens. In all their variations — pre-election opinion polls, exit polls and post-election surveys (conducted before the counting of votes) — polls seek to penetrate the veil of secrecy that cloaks individual political choices. No doubt many of those polled love to broadcast their preferences, but others cherish the secrecy of the ballot — not only because of a pathological passion for privacy but also due to concerns about social pressure, coercion and intimidation. The latter group may refuse to participate in the poll, thus biasing its results; worse still, they may participate but misrepresent their preferences, thus falsifying its predictions altogether.
When they do participate, what ensues is a fascinating game between the pollster and his target, the former seeking the truth, the latter considering whether and how to conceal it. All pollsters of course assure the voter that the secrecy of his choice will be protected. But all they can do to back up this assurance is to present him with a dummy ballot for insertion into a dummy ballot box in the hope that this will convince him that his secret is safe. This is a device that would convince only a dummy. Any voter with a modicum of sense knows that all that pollsters have to do to divine his secret is to open the box as soon as the voter is out of sight. The only question is whether they have the motive or the opportunity to do so.
Exit pollsters have neither. They are interested only in the aggregate voting picture, not in the vote of any specific individual; and even if they were, the continuous stream of voters passing through the exit polling booth would make it almost impossible for pollsters to peek into the vote cast by any one.
Things are very different for pre-election opinion polls or post-election surveys. Here pollsters randomly select a voter, visit him at home or work, quiz him about his age, his caste or religion, his income, his political beliefs and other personal particulars before offering him his fake ballot and fake ballot-box. In such situations, pollsters have the strongest incentive to discover the specific voter’s choice: the purpose of the whole exercise is to ascertain how certain age-groups, communities, castes or income-brackets vote and the only way to do so is to find out how specific members of these groups say they vote. Once the pollsters leave the voter’s home, they have ample leisure to open the fake ballot box and see whom the voter has voted for. The voter knows all this. Of course, pollsters may go a step further: they may assure the voter that even if they or their organizations somehow discover his actual vote, they will carry this secret to their graves. The voter, however, has no reason whatsoever to believe them. So, if he wants to conceal his electoral preferences, he has every incentive to simply lie, “voting strategically” in the opinion poll in a sense very different from that in which the Left intelligentsia and Muslim clerics had urged Muslims to do in 2014.
In consequence, all polls err systematically and irremediably. The systematic underestimate of the Bahujan Samaj Party’s share in votes and seats (except, of course, in 2014) and of the Congress’s successes in 2004 and 2009 are well-known. Of current interest is the total failure of all — except for the final exit poll by Chanakya — to predict the scale of the BJP’s victory in 2014. How, then, should one interpret the predictions of an opinion poll or the analysis of an opinion survey? Clearly they cannot reflect actual voting behaviour; what they indicate are only the voting patterns that voters are willing to admit to, or, more accurately, to claim. They are closer to what is considered politically correct (and possibly safe) than to reality.
In the 2014 elections, far more people voted for the National Democratic Alliance than were willing to admit to doing so. They voted clandestinely for a Modi regime in the face of real or imagined social pressure or threats. Some may have voted shame-facedly for Modi in the teeth of a consensus among the intelligentsia that as prime minister, he would be a national calamity.
What of assertions based on post-poll opinion surveys such as “Only 9 per cent of the Muslim electorate voted for the BJP”? The correct version of this statement should read, “Only 9 per cent of the Muslim electorate admits to voting for the BJP.” In fact, the only true measures of voting behaviour would be those extracted from votes actually cast in the elections, not from the dummy votes disgorged by the opinion pollsters’ dummy ballot boxes. Of course, the secrecy of the ballot ensures that no direct link can be established between the identity of an individual or his age, caste, religious, economic or educational characteristics and his vote. But correlations can be observed between the vote shares of parties and the socio-economic and demographic characteristics of constituencies.
Do such tests bear out the received wisdom about a profound Muslim antipathy to Modi? Consider the data from Uttar Pradesh, supposedly the focus of the BJP’s “polarizing” campaign. In order to focus on the serious players in this election, concentrate on the votes polled by the four major parties and their allies (Apna Dal and Rashtriya Lok Dal), ignoring the hundreds of little outfits that garnered a few hundred votes each in many constituencies. Of the votes of the Big Four, the BJP received 47 per cent in constituencies with less than 10 per cent Muslims, 45.5 per cent in those with 10-20 per cent Muslims, 46.5 per cent in those with 20-40 per cent Muslims and 42.3 per cent in those with more than 40 per cent Muslims. Its state-wide share in the votes of the Big Four was 46.8 per cent. This is hardly a picture of overwhelming Muslim hostility to the BJP. In fact, its vote-share seems virtually independent of the religious composition of the constituency.
For those infatuated with the belief that Modi’s victory has created a deep unbridgeable chasm between the communities, such data are not reality but illusion. They reject the simple explanation that quite a few Muslims must have voted for the BJP just as others did. They assert, instead, that Muslims remain unreconciled, still inveterately hostile to Modi (except of course for the 9 per cent) but rendered electorally impotent by a powerful Hindu counter-consolidation, a theory that establishes to their own satisfaction how divisive Modi can be. The question of how the BJP could win 42.3 per cent of the vote in constituencies like Moradabad and Rampur, where half the electorate is Muslim and a solid chunk of the rest is Yadav or SC, is one they ignore. They prefer to pose the counter-question: if many Muslims voted for Modi, why should they conceal the fact? The answer is obvious. When the clergy and the intelligentsia have for over a decade portrayed Modi as a bloodthirsty antediluvian monster, a 21st-century Tyrannosaurus Rex in hot pursuit of minorities, a Muslim may still vote for him, but to admit this would mean being branded a traitor to one’s community, a fate that no one would willingly embrace. Who knows? Perhaps it is this subterranean Muslim vote, invisible to the opinion pollsters, that is the key to Modi’s spectacular victory.