Sanjay Kumar | The challenges and the risks over exit polls

Conducting an exit poll is a challenging but not an impossible task. If it was impossible to do an exit poll, it wouldn’t have become a multi-crore business involving several survey agencies. It is challenging as there is hardly any other business where the product’s quality is put to the test as quickly as it is with an exit poll.

For fairly obvious reasons, forecasting is easier in a bipolar political contest than in a multi-party one. For example, if there is only one student in a class, he doesn’t face any challenge and will always come first. If there are two students, there is some challenge and is no longer certain about coming first. Likewise, if there are many parties in the fray, it is harder to calculate which party will win how many seats. If the vote estimate is inaccurate, there is a huge chance of an inaccurate estimate of seats for parties and alliances.

The conventional view was that states like Bihar or Uttar Pradesh, where people are more opinionated, are more challenging as people may say something and act differently. But recent experience has shown that exit polls have become more challenging in smaller states like Goa or Manipur, where voting choices are often determined by a preference for a candidate rather than a party. Voters are usually clear on their choice of party, and find it easier to reveal their preference for a party than for a specific candidate.

Multi-cornered contests are also far more challenging when there have been spilts, mergers and defections from one party to another between elections and shifting of alliance partners from one election to another. The presence of a large number of castes and strong casted-based voting makes holding exit polls even more complicated. Uttar Pradesh, which is the focus of most political parties, has seen major changes in alliances between the 2017 and 2022 Assembly polls. In 2017, the Samajwadi Party fought in alliance with the Congress, while in 2022 it is in alliance with the RLD and some others. The defections in UP from one party to another also poses a challenge for pollsters as many leaders have changed their parties just before the recently-concluded elections.

Punjab could be equally challenging as the alliances there have changed. The Akali Dal and the BJP are no longer allies: the Akalis contested these polls in alliance with Mayawati’s BSP while the BJP formed an alliance with Capt.

Amarinder Singh’s new party. The Congress has seen major changes: the party is not in the same shape and size that it was in 2017. The Congress’ tallest leader, Capt. Amarinder Singh, had left and formed his own party and forged an alliance with the BJP. There is a new party set up by the farmers, which was also in the electoral fray in Punjab. All these changes led to a multi-cornered contest in Punjab.

The entry of a new political player in any state also poses a challenge as it become difficult to assess the electoral support base of the new entrant. An assessment of the vote share of the new entrant is also very difficult as it is for a new party formed after a split from its parent party. For applying a swing model for making seat forecasts based on vote share, an estimate for the new entrant is needed. Given that Goa, Uttarakhand and Manipur saw new entrants in these elections, the pollsters faced a major challenge. The Aam Aadmi Party fought in both Goa and Uttarakhand, while the Trinamul Congress was in the fray in Goa.

There was also a new tribal party in Manipur, which was contesting the Assembly polls for the first time. All these factors add to the challenge for pollsters.
Given these challenges, how do we read these exit poll estimates in the five states? If there is a “wave” election in any state, it eases the burden of pollsters as there are greater chances of getting the forecast right. The pollster may make an error in forecasting the extent of victory of a particular party, but getting winner right seems easier in wave elections. While almost everyone had suggested that Uttar Pradesh this time was a “wave-less” election, the exit polls seem to suggest the wind is certainly blowing in favour of the BJP and its allies. Despite various challenges in conducting an exit poll which applies for states like UP and Punjab, the pollsters seem to have got the winner absolutely right in both states. The pollsters should thank the voters of these states for making their task easier, when it initially looked there would be multi-cornered contests. All the pollsters predicted a victory for the BJP in Manipur, and there is no reason to dispute these findings. Except for one survey, all the others have predicted a victory for the BJP in a tight election in Uttarakhand. The presence of the AAP and BSP may have made the pollsters’ job even more challenging in Uttarakhand.

While the other states seem to have defied the conventional challenges that pollsters face in exit polls, voters in Goa don’t seem to be offering any help to pollsters. Different exit polls have predicted a neck-and-neck contest in Goa. The real test for pollsters in these Assembly elections will be who got it right in Goa and Uttarakhand, and not about who got it right in Uttar Pradesh, Punjab or Manipur.