The recently released NDTV and CNN-IBN opinion polls have predicted a categorical win for the NDA, giving it between 238 and 258 seats. The polls sponsored by ABP News, Zee and Pew Research point in the same direction. Does this unanimity of opinion, with most print media reports pointing to the same, suggest that the results are a done deal?

The markets seem to believe so, as do some opinion makers. But there is still a small band of sceptics that believes the NDA will be the single largest formation, though it may struggle to form the government – they feel the media and pollsters may have overestimated the swing in the NDA’s favour, even as they acknowledge an anti-UPA wave.

In a multi-party situation, even if NDA falls short of estimates by, say, 35 seats, it may struggle to get to the magic number of 272. In such a situation, the balance of power would tilt in favour of the NDA allies vis-à-vis the BJP, or — what seems less likely — even open up a possibility for Third Front players.

Unjustified criticism

Where will an unexpected election result leave the pollsters? Would it be right to point an accusing finger at opinion polls, condemning them as a sham? That’s overstating the case, if not getting it wrong altogether. The Congress has argued for a ban closer to the elections, saying they are manipulated by vested interests and aimed at influencing the poll verdict.

This is unacceptable, both on a point of principle and fact. Exit polls are rightly banned for their potential to influence voting behaviour, as they are beamed in real time. But the link between opinion polls and voting patterns is hard to establish.

For instance, in the 2009 elections, the Congress won 206 seats against poll estimates of about 155 seats; can it be said the polls induced a push in favour of the ruling party?

Or that, in the 2004 elections, the NDA lost vote share against all expectations precisely because pollsters were expecting it to win comfortably?

Banning opinion polls is a slippery slope; it could lead to curbs on any kind of journalistic forecasting on print and television, curtailing freedom of speech and expression. Opinion polls supplement media reports as an important source of information for political parties.

Media watchdogs should insist on transparency in the modalities of the survey, rather than reach for the scissors.

Why the errors

Opinion polls can go off the mark on two counts: in estimating vote share and in arriving at the correct number of seats on the basis of vote share. For instance, in 2004, most polls estimated an increase in vote share for the NDA and estimated the coalition would get 280-300 seats. However, in 1999, they were largely right in predicting a Congress win on the basis of vote share.

The Congress ended up on the losing side, though, with 114 seats, despite a 28.3 per cent vote share, while BJP got 180 seats with a vote share of just 23.8 per cent. The conversion of vote share into seats remains an area of darkness, and the arcane formulae of psephologists do not seem to help.

In the 2004 elections, both the Congress and the BJP lost 1.6-1.7 percentage points in their vote share vis-à-vis 1999, but the Congress ended up with 145 seats, 31 more than in 1999, while the BJP was down by 42 seats to 138. As in 1999, Congress retained a lead in vote share. Vote share to seat conversion becomes complicated in a multi-party scenario.

A wave election?

Now, for the predictions on the ongoing elections. The polls are expecting a wave in favour of the NDA – a 6-8 percentage point increase in vote share from 24.6 per cent in 2009.

To put this ‘swing’ in perspective, let us examine the vote share changes in previous ‘wave’ elections: 1971 (Garibi Hatao), 1984 (Indira Gandhi’s assassination), 1989 (corruption/Bofors and Ram Mandir) and 1991 (Ram Mandir).

The Congress’ vote share increased by 3 percentage points in 1971 over 1967 — from 40.8 per cent to 43.7 per cent. In 1984, the Congress vote share went up by 6.3 percentage points over 1980 to 49 per cent. In 1989, the Congress share was down 10 per cent from 1984. In 1991, the BJP secured 20 per cent of the votes, against 11 per cent in 1989 — a 9 percentage point swing in its favour. Does the mood in the current elections compare to any of these? The polls have captured the pro-NDA sentiment. But the devil lies in the detail.

comment COMMENT NOW