LS polls: Surveys to 16 May, many a slip between cup and lip

The last of the major opinion polls will be concluding today (4 April), and one thing is clear: the BJP-led NDA is clearly well ahead of the UPA, and seems set to form the next government after 16 May. The only doubts relate to the scale of its victory.

The Lokniti_-CSDS-CNN-IBN Election Tracker_, which concluded its state-wise poll results yesterday, also seems to point in the same direction, while the ABP News-Nielsen survey gave the NDA 236 seats (UPA - 92). NDTV (in its March survey ) gave the NDA similar numbers (230), and it is set to conclude its latest survey tonight before the Election Commission’s deadline tomorrow. (See the NDTV poll results so far here ).

However, if you think it is a slam-dunk for Narendra Modi and the BJP, you would be well advised to keep your fingers strongly crossed for another 42 days. Remember 2004? The polls predicted an NDA victory, and we ended up with the UPA. The results in 2009 were less wayward, but nobody predicted the scale of the Congress victory.

So, what are the surprises and pitfalls you should look for this time?

First, the trend is unlikely to be wrong, but the seat-count could be significantly different. The biggest pitfall in predicting an Indian election comes from converting vote shares to seats. This is because concentrated vote shares can give some parties more seats, while diffused votes can cost other parties seats. In 2008, the BJP won Karnataka despite polling fewer popular votes than the Congress. Last year, the Congress swept the polls despite increasing its vote share by just around 2 percent.

So, even though the BJP appears to be doubling its vote share this time, according to many opinion polls, a lot of the extra votes may be coming from areas where it may be getting no (or few) seats. Example, Tamil Nadu, Kerala, West Bengal – where it could get 11-12 percent vote share and practically no seats. So BJP partisans should wait awhile before ordering the bubbly – or chai, if that’s their preference this time, in honour of the chaiwala they plan to send to 7, Race Course Road.

Second, there is the bull-whip effect . In any election, the undecided voters tend to make up their minds closer to voting day, when they know which way the wind is blowing. This is called the bullwhip effect, where a party that is ahead suddenly reaps a huge harvest of last-minute votes that push its seat totals in a favourable direction – beyond what the opinion polls indicated. Opinion polls cannot capture the bullwhip effect – though exit polls can. (We have to wait till 12 May for those to kick in).

The bullwhip effect gave the Samajwadi Party a clear majority despite getting less than 30 percent of the vote in 2012. This could happen in 2014 too for the BJP, if the undecideds finally decide to vote for Modi in the belief that he is the winner. Nobody likes voting for losers.

Third, the minority vote may make a huge difference this time. And opinion polls may be undercounting their voting intentions. The Lokniti-CSDS-CNN-IBN Tracker polled a total of 20,957 voters through a system of random sampling to avoid bias. But even with all the care one takes, there is a possibility – even probability – that the minority vote may spring a surprise.

For example, in Bihar, the minorities have a 16.5 percent share of population, but the opinion poll spoke to only 10.9 percent Muslims. In UP, the comparative percentages are 18.5 percent and 14.6 percent polled. Since it is well known that Muslims may be voting tactically this time to defeat the BJP under Modi, one cannot truly estimate the impact their voting will have on the BJP’s chances. If they truly vote tactically in all seats in UP and Bihar, the BJP tally could even fall below 30. If they split their vote between two parties in many places, the results could be more favourable to the BJP.

However, the chances are the impact of the Muslim vote is being underemphasised this time in both UP and Bihar – because they may be undercounted, because they may be more motivated to come out and vote, and because they may have preferred to mislead pollsters for the time being while they decide who to cast their vote.

On the other hand, there is also the reverse possibility: any hint of a Muslim polarisation against Modi could prompt the reverse among the majority community.

Apart from minorities, Dalits also tend to get mis-estimated in opinion polls. This is why Mayawati could be the X factor in Uttar Pradesh this time. If both Muslims and Dalits turn up to vote in large numbers – the former to defeat BJP and the latter to vote for Maya – UP could spring a last-minute surprise.

That said, the Lokniti-CSDS survey indicates that SCs have been polled higher than their share of the population in Bihar, and lower in UP. We don’t know how that will ultimately play out – especially since Modi is said to be making inroads in this voter base too, given his most backward status.

Fourth, there is the couch potato effect. If too many Modi voters think their man has already won and don’t turn up at the polling booths, preferring to watch the action on TV, the BJP could still end up with fewer seats than its vote share predicts. The real issue is those who feel threatened by Modi may have a superior motivation to vote than those who don’t. The polls predict armchair voting intentions better than actual votes. So that’s another uncertainty.

Urban voters also tend to be lazier than rural ones – and urban voters are key to Modi’s vote.

In the ultimate analysis, political parties can’t count their chickens unless they can herd their voters to the booths on polling day. Parties like the BJP, which may not have built their party organisations sufficiently between 2009 and now in critical states such as UP, may find that there is a gap between voter intentions and votes. This may be a bigger threat to the BJP than the other parties.

Fifth, opinion polls tend to undercount relatively new undercurrents and overcount obvious trends. This was demonstrated clearly in Delhi, where the undercurrent was against the Congress, but most of this vote was presumed to go to the BJP.

The Aam Aadmi Party’s vote shares got massively undercounted in opinion polls before December 2013 because many voters did not believe the party could win – and yet they voted for it. The opinion polls failed to get this correct – and some pollsters were changing their stories on voting day as they witnessed massive concentrations of AAP voting in many areas.

Nobody predicted Sheila Dikshit’s defeat – and that too at the hands of the AAP.