2024 general elections: Exit polls can go wrong; take them with a pinch of salt

HYDERABAD: “Anything that can go wrong will go wrong.” This is the epigram of Murphy’s Law. It has an extension too – at the worst possible times.”

This aptly suits the exit polls and the prediction of electoral results. More often than not, exit polls in India went wrong stunning the pollsters, much to their own embarrassment, and to the consternation of political parties.

An interesting study of the previous exit polls has laid bare the shallowness of exit polls. Most pollsters were, however, closer to the truth in some cases, while most of their predictions went haywire in most other case.

When it comes to Telangana Assembly elections in 2023, most pollsters got it right and could make a perfect forecast of the rout of the Bharat Rashtra Samithi and the victory of the Congress with a fewer seats more than the half-way mark.

India is going through the biggest ever election currently and the exit polls would be out once the last day of polling ends on June 1, 2024.

While the people are eager to know the denouement of the political soap opera that lasted almost for over a year across the length and breadth of the nation, the pollsters are busy doing their typical number crunching and making a last ditch effort to get the result nearly right.

Even if they factor in providence, there is always this proverbial element of a slip between the lip and cup. ß

Sometimes, who contracted the poll agency and the leaning of the media organization that hired the pollster too play a role in giving out the numbers.

For instance, let’s see how the exit poll results went off the mark on many an occasion.

The exit polls done by almost all agencies, except Axis My India, went totally wrong in predicting the results of Karnataka in 2023. Credibility of the past exit polls is one thing every agency hark back to boast their accuracy. However, the theory of probability is what most agencies try to harp on. By the same logic, the accuracy of the pollsters to can go right or wrong.

Except in the case of Telangana in 2023 Assembly polls, where the writing on the wall was more than a stark reality, all major pollsters went wrong with regard to Chhattisgarh. Axis My India predicted a close call between the Congress and the BJP, while the BJP in reality won an absolute majority with 54 seats. My Axis foresaw the Congress bagging 40-50 seats as opposed to the BJP’s 36-46. But none of that was true.

In the case of Rajasthan, India TV-CNX, New 24 – Today’s Chanakya and India Today – Axis My India went completely wrong. ‘All enabled School of Politics’, ABP-C Voter and TV9 Bharatvarsh–Polstart could predict nearly correct result indicating who the winner is. Jan ki baat, Republic TV – Matrize, Times Now – ETG predicted the possible outcome accurately after the exit polls.

In the case of Madhya Pradesh, ‘All enabled School of Politics’, India TV-CNX, Republic TV-Matrize, and India Toady – Axis My India came closer by providing a directionally correct result in the exit polls. ABP- C Voter, Jan ki baat, Times Now-ETG, and TV9-Bharatvarsh-Polstart goofed up predicting wrong results.

The Lok Sabha elections held in 2004 that happened soon after the elections to the Assemblies of Madhya Pradesh, Rajasthan and Chhattisgarh. The BJP-led National Democratic Alliance (NDA) was projected to win the polls in the exit polls. While pollsters gave over 250 seats to the NDA, but the actual result gave the Atal Behari Vajpayee dispensation just 187. The Congress-led United Progress Alliance (UPA) was projected to win 170 to 205, as against the actual outcome of 219.

When it comes to 2014 Lok Sabha polls, the pollsters were “impressed” by the UPA’s decade-long rule, leaving them to predict just an edge for the NDA, which indeed triumphed bagging more than 300 Lok Sabha seats and the BJP individually crossing the half-way mark of 272 seats. The exit polls went wrong and the results had upset the proverbial apple cart of the Congress and the UPA, limiting them to a niggardly 60 seats.

The exit polls for the Bihar Assembly elections in 2015 expected a close call between the contestants and did not predict any majority for any of the parties. However, the results had taken the pollsters by shock. The RJD-JDU-Congress coalition secured 178 seats as opposed to the BJP and its allies which were limited to 58.

The Delhi Assembly polls or 2015 had sprung a surprise to the entire nation with the Aam Admi Party literally sweeping the polls. It secured 67 out of the 70 seats in the Delhi Assembly. Though the pollsters predicted a victory for the Aam Admi Party, they did not expect such a devastating defeat to the BJP and the Congress. While the BJP made it in just three seats, the Congress came a cropper.

In the 2017 Assembly elections of Uttar Pradesh, the exit polls expected a backlash of the demonetization by the Narendra Modi regime. Most exit polls expected a hung Assembly in the State. However, they had to eat crow as the BJP posted a resounding victory winning over 300 seats. This was way too beyond its erstwhile number of 47 seats in the Assembly, trampling the opponents under its feet. The UPA (Samajwadi Party-Congress combine) was inflicted a humiliating defeat with just 19 seats, as against the 325 of the BJP and 47 of Bahujan Samaj Party (BSP).

What are the factors that contributed to the Murphy’s Law to come true?

Pollsters actually study the vote share in different constituencies and then convert them to seat share which would miss the mark generally. Without spelling out the methodology, Today’s Chanakya comes close to the truth on most occasions. With enormous sample sizes, Axis My India does a nearly professional job. All others too predict the electoral outcome.

The factors that dent the accuracy of the pollsters include:

1. Going wrong with the basic premise, for the pollsters assume that the voters have given them accurate information on who they had voted for. Most occasions, the voters would not tell the truth and largely their response depends on who were all present at the polling station (political party agents) and also the marginalized sections do not take the pollsters seriously. Neither they can respond in English nor are they willing to share the information.

2. When the election is tightly contested, the usual error margin widens from 1-3 percent up to 5 percent. This naturally affects the poll prediction.

3. The pollsters would hesitate to touch base with women voters who are usually silent on what they chose in the EVM. Always the women voters’ sample is not right.

4. The lack of efficient research and background work before taking up the exit polls and also reducing the expenditure on the field staff naturally affect the process. When there is no financial input the desired result always remains elusive. The ‘lazy psephology’ owing to lack of resources – human and financial – and casual approach down the line too can put out wrong predictions.

5. The tight turnaround time (TAT) to send the feedback without tabulating the data effectively obviously makes the calculation clumsy and the outcome of the prediction can be skewed.

6. Field staff’s reluctance to go deep down into the constituencies, choosing booths that are convenient for them, the most important human error in assuming and filling the forms on a templated basis,, dependance on the historical data of the previous elections and the dearth of accuracy in the socio-economic and caste data sets would all contribute to the failure of the pollsters in predicting the poll result.

Therefore, the exit polls must always be taken with a pinch of salt. They are not the actual results. Some of them may give out a forecast based on an informed guess coupled with whatever field exercise they have done. But most occasions they are not accurate.