Where Exit Polls Fail: They Often Get Carried Away By The Noise, Losing The Silent Voter

After the Lok Sabha elections, 8 October has delivered a second stunning defeat to our pollsters. In June, almost all exit polls predicted the BJP’s return to power with a majority of its own, but the party faltered in crucial states like Uttar Pradesh, Maharashtra and Rajasthan, ending with 240 seats.

Today, as the results of the Jammu & Kashmir and Haryana assembly elections come in, it appears as if the pollsters have got it wrong once again in a crucial state.

While they got it more or less right in Jammu & Kashmir (J&K), where the exit polls directionally gave the National Conference-Congress alliance a near win but not a majority, almost all of them (without exception) gave the Congress a clear victory of 50-plus seats in Haryana. But the results have been entirely the opposite.

But before we analyse why they went wrong, we must empathise with the pollsters for two reasons. First, in a close election, the vote share-to-seat conversion is always a dicey affair. Second, when it comes to vote-share estimates, pollsters often tend to undercount the less voluble voices.

In the Lok Sabha elections, the whispers and unhappiness within the BJP and the Sangh Parivar were not given as much importance as the BJP’s confident claims of 400-paar. This happened in 2004 as well, when most pollsters got carried away by the seeming invincibility of the Vajpayee-led National Democratic Alliance. But we know what happened, especially with allies in Andhra Pradesh and Tamil Nadu in 2004.

In J&K too, where the pollsters thought new, radical parties like Engineer Rashid’s Awami Ittehad Party and the Jamaat were expected to dent the mainstream National Conference and People’s Democratic Party (PDP), the voters actually decided to give a clear mandate to the devil they know rather than grapple with the fringe. Small parties got a quarter of the votes, but very few seats.  

It is in Haryana that the exit pollsters met their Panipat.  

No pollster predicted that the BJP would win, and all of them put the BJP vote share significantly below that of the Congress.

But here is the problem: the BJP has nearly tied with the Congress with a 40 per cent vote share, which has made all the difference. One can get the vote share-to-seat conversion wrong, but the pollsters seem to have gotten the vote share detail itself wrong. Clearly, they undercounted the votes that stayed silent even as the Congress party’s Jat dominance was assumed to be enough to deliver a win.

On the contrary, even in Jat-dominated seats, the BJP outperformed the Congress, as the Jat vote was split between Congress and some independents. Meanwhile, the consolidation of the non-Jat vote — which must have been the silent vote — put the BJP over the top.

And what pollsters surely cannot explain is how the BJP performed better than in 2014 and 2019 — when the Modi wave was still to reach its peak.

The silent vote — often comprising women and other marginalised groups — can make all the difference in a close election.