Exit Polls May Prove Wrong - But On Modi, They're Clear

Whatever else the exit polls may or may not say, they do confirm the big election story - that Prime Minister Narendra Modi has almost single-handedly lifted the BJP in Uttar Pradesh from the depths of despair to one where they are predicted to win outright or at least emerge as the largest single party. In doing so, he has not only established himself as campaigner-par-excellence but politically put to bed any adverse impact of demonetization. And if those polls that project huge majorities in UP, Uttarakhand and even Manipur turn out to be correct, then the political landscape will have changed dramatically.

While this euphoria over the forecast of a BJP sweep, except in Punjab, may have those on Twitter celebrating, the political class itself remains much more circumspect. Even BJP leaders are not popping the champagne, as memories of Bihar's exit polls being completely wrong still linger in their minds. And if you look at the gamut of predictions for tomorrow, it is obvious that some, many, or may be all  pollsters could be wrong.

Take UP, the state steals the limelight in these elections, because of its political weight. If you plot the predictions of the BJP vote and seats (see chart below) you will find that most of them have coalesced around 33% vote share and anything between 165-200 seats.Two of them have, however, proclaimed big majorities for the BJP. Of these, the News 24 Chanakya poll gives the BJP almost the same vote share as the 2014 Lok Sabha election, but trims their seats by 50 for a final count of 285. This is almost the result that would come about if you took the 2014 Lok Sabha data by assembly segments and introduced the SP-INC alliance, making it a three-cornered fight. In other words, News24 is saying that the Modi Wave of 2014 holds good in 2017. While that may be true of the later of seven phases of voting, the BJP and its supporters all admit that in the first two phases, where 140 seats were decided, the BJP under-performed considerably as compared to 2014.

Some polling also shows that Muslim consolidation around the SP-Congress could possibly have triggered a Hindu consolidation around the BJP especially in the later phases of voting. In 2014, the BJP swept eastern UP, so such a scenario remains possible. If you don't buy the consolidation of votes and so exclude the News24 projections as an outlier, then the poll of polls changes, though it continues to show the BJP well ahead but just under a majority:The other concern for the BJP would be that almost all polls have shown their vote share in UP dropping to 32-36% from 43% in 2014. That is a greater drop than happened in Bihar, and again reflects the purported early setbacks. Similarly, the SP-Congress should be very disappointed that their vote share has hardly moved from 2014 - it has declined by 6% compared to what both parties polled together in 2012. And this probably reflects the reality on the ground, where SP leaders felt that Congress votes did not transfer to them.If the polls are correct, the Congress in particular has much to be unhappy about. If all the worst-case scenarios play out for the Congress: losing Punjab, a thumping in Uttarakhand, no comeback in Goa and possibly losing Manipur, Saturday could be a very bleak day for the party.The tragedy for the Congress in Punjab is that if they win, it will be because of Captain Amarinder Singh, and if they fail to get a majority, the Captain has said he plans to retire, so the Congress will be left rudderless. Similarly, not winning in Goa will shows its inability to win against incumbency (the BJP governs the state) and losing Manipur will underscore the growing strength of the BJP in the North East.

The exit polls in Punjab also reflect two other trends: the collapse of the Akalis and with them, the BJP, is reflected in all the polls; and the continued growth of AAP shows that 2014 was not a flash in the pan and they are emerging as a major national player.If AAP wins Punjab, as some polls indicate, then not only will the Congress be worried, having lost to them heavily in Delhi, so will the BJP, whom they plan to challenge in Gujarat later this year.

In less than 12 hours, the actual results will start coming in, and many exit poll predictions will be proved wrong. Wrong, because polling isn't a science: it is a projection from a sample. The problem is that in homogenous countries, it is easier to project from small samples, but in places like UP, where more than hundreds of sub-castes exist and caste-wise voting is often the case, pollsters face the enormously difficult job of choosing their samples correctly. Now if you launch such polls across seven diverse phases, the chances of being wrong increase. 

The Trump victory in the US elections highlighted the fact that small samples cannot capture the voting variations by region and other demographics. The US election also saw a lot of people who did not admit that they were voting for Trump. Similar trends have affected Indian pollsters for many years - so take the polls with a pinch of salt. They are indicators of a pattern, or a trend, and the  "trend" seems to be heading BJP's way, but whether these trends convert to a sweep is a more difficult call.

(Ishwari Bajpai is Senior Advisor at NDTV)

Disclaimer: The opinions expressed within this article are the personal opinions of the author. The facts and opinions appearing in the article do not reflect the views of NDTV and NDTV does not assume any responsibility or liability for the same.