Exit polls are generally a better barometer of election results and trends than opinion polls. The general consensus among exit polls is that theNDA is coming back to power. BJP by itself, may just fall short of a majority. The exit polls have broken the trend of the past few years where they didn’t present a clear picture. On most occasions, we found the exit polls to be hung themselves, but this time there is unanimity, but numbers and ranges differ.
Exit polls may not be able to accurately predict the seats and vote shares, but they do generally indicate the direction of the trend correctly.
A caveat, we will have to wait for the final results on 23 May, which could get delayed due to VVPAT reconciliation. While exit polls in India went horribly wrong in India in 2004, we have seen them misreading the Trump elections, the Brexit vote and lately, the Australian polls.
In India, given the diversity, population growth, number of first-time voters and inconsistency in polling station coverage, historical elections data is far less reliable than say in Europe, the Americas or Australasia, where there are minimal changes in population and no significant addition of new voters.
While most pollsters agree that the NDA is returning to power, there are huge variations in the seats tally. Take a look:
Variations in Tally
- Nielsen predicts the lowest tally for NDA, still +5 from simple majority, implying BJP falling short of majority.
- Axis predicts the highest tally for NDA, +30 compared to 2014, +9 considering the current constituents of NDA (including AIADMK as NDA in 2014).
- Axis predicts the lowest tally for UPA, +18 seats compared to 2014.
- VMR predicts the highest tally for UPA, more than double the 2014 numbers.
- Axis predicts Others will lose half heir seats, while C-Voter predicts that they will maintain their tally. I go with C-Voter in this case, Others can’t be so low.
Let’s take a look at the vote shares:
- VMR predicts the lowest vote share for NDA, still +2.6 percent versus 2014.
- Ipsos predicts the highest, +10 percent compared to 2014. NDA’s vote share is expected to go up due to gains made in Odisha and West Bengal, and maintenance of vote share in UP.
- Ipsos predicts the lowest vote share for UPA, +1.6 percent compared to 2014.
- VMR predicts the highest, more than 7 percent.
- Ipsos predicts Others will lose almost 12 percent vote share, but their seat tally will fall only by 23.
- C-Voter predicts vote share of Others will fall by 10 percent, but seat tally will be same as 2014.
Which brings us to state-wise numbers:
- In UP, Ipsos predicts for NDA twice the number of seats that Nielsen has prophesied.
- In Bengal, Axis predicts 5 times more seats to NDA than Ipsos. This, despite the fact that the lower end of overall seats of Axis are same as Ipsos’ overall tally.
- In Odisha, Axis numbers for NDA are twice the number predicted by Nielsen.
Method to the Madness of Exit Polls
The exit polls method is a four-pronged selection process of PCs, ACs, polling booths and voters using random sampling method. PCs / ACs / booths are selected keeping in mind swing seats, low margin seats, stronghold seats. The number of PCs, ACs, polling booths and voters depends upon the total sample size intended. Surveys are usually conducted on the election days, when the voter comes out of the station after polling. Detailed questionnaires are prepared; also translated into vernacular languages. The data is then weighted to take into account age, gender, caste, religion and turnout.
Why Such Disparity in Results?
Polling entails employing scientific tools. Many voters – mostly rural – and also field investigators, may not understand the intricacies and respond in a scientific manner. The nature of contests also results in different seat calls. This time, the contest is close in UP, MH, Bengal and Odisha; triangular fights adding to the complexity.
The basic assumption on which the polls are made – that the voter is telling the truth – may not always be correct.
Some may deliberately lie (annoyed with interviews by a host of polling agencies) while others like minorities / Dalits / the poor / deprived – basically, the marginalised – may not want to speak the truth for fear of backlash.
Shortcomings of Historical Poll Data
A lot of reliance is placed on historical data which has its own shortcomings. Change in turnout, addition of new names to voter list, composition of polling booths, etc, pose many problems. Caste in a constituency is not available, as the last caste census was done in 1934. The new pollsters could struggle even with this data. The depth of research and cost-cutting, hampers data collection.
There is intense turnaround pressure. Advanced technologies are being used, but there is a human-angle element in the field.
Many times, investigators sample only convenient booths like urban / semi-urban booths. Given the above challenges, each pollster ends up with different sets of numbers even if they go to the same polling stations and interview the same set of people.
The Top Pollsters May Not Always Get It Right
C-Voter, which predicted an AAP victory in Punjab, was bang on in Tamil Nadu. Axis which got Bihar right, got Tamil Nadu wrong. Today’s Chanakya, which got the Lok Sabha polls right, got Bihar and Tamil Nadu horribly wrong. Both got UP right. The best of the pollsters by track record ‘Axis’ and ‘Today’s Chanakya’, have got it spot on (seat tally) only 50 percent of the time. The big ranges also confuse the voters.
Exit polls are like ice-creams. They quench our sugar-craving. But only till the time one has the next craving. Hopefully, by that time, the actual results will be out. Many reputations at stake.