EXIT POLLS

Exit polls provide estimates on how people voted in an election. They are based on interviews with voters immediately after they exit the polling stations, combined with other calculations related to voter data.

Public Interest in Exit Polls

A large number of Indians place as much importance on exit polls as they do on the actual results. Typically, exit polls are released on the last day of voting because agencies conducting these polls are mandated by the Election Commission of India (ECI) to wait until polling has been completed in all phases. This rule exists to prevent influencing voters who have not yet cast their ballots.

Release and Public Perception

When exit polls are finally released, pollsters often present varying estimates. People tend to show interest in those exit polls that align with their political preferences. They generally disregard the vote share estimates and methodologies. More often than not, the accuracy of exit polls is judged by personal opinions on political parties.

Upcoming Exit Poll Release

When exit polls for the ongoing Lok Sabha elections are released on the evening of June 1, this phenomenon is likely to be repeated. Many television channels will rush to air the polls. Unfortunately, there is more competition to be the ‘first’ to show the exit poll numbers than to ensure superior data quality.

Recent Inaccuracies

In recent years, many exit polls have proven erratic, often showing conflicting results. Last year, many polls incorrectly predicted winners in the Legislative Assembly elections for Chhattisgarh and Madhya Pradesh, while some were way off the mark in Rajasthan.

No Predictable Pattern

There is no consistent pattern in the accuracy of exit polls. An agency may have accurately predicted the results for Madhya Pradesh but made errors in Rajasthan. There were states where all exit polls were accurate (Telangana) and others where none were correct (Chhattisgarh).

Factors Influencing Accuracy

Today, some people judge the accuracy of an exit poll by the survey agency that conducted it or the television channel that commissioned it. Others consider the sample size, believing that a larger sample size indicates a more reliable poll. In reality, these should not be the sole indicators for judging the accuracy of an exit poll, which depends on many different factors.

• The science of surveys, including exit polls, operates on the premise that data are collected from many respondents using a structured questionnaire, whether conducted over the telephone or face-to-face.
• This method is not new; it began in 1957 during the second Lok Sabha elections when the Indian Institute of Public Opinion conducted a poll.
• However, no guesswork or estimation can replace the necessary methodology. Without a structured questionnaire, data cannot be collected coherently or analyzed systematically.

Since the inception of exit polls in 1957, there has been significant improvement, particularly in sample size. Gone are the days when a national sample of 20,000 to 30,000 respondents was considered large. Today, survey agencies conduct exit polls with samples as large as 1 million, and samples of a few hundred thousand have become commonplace.

Historical Context and Methodological Advances

Though the Lokniti-Centre for the Study of Developing Societies (CSDS) typically does not conduct exit polls, it has done so occasionally. For instance, in the 1996 Lok Sabha elections, Lokniti-CSDS used a sample size of 17,604 to conduct its first exit poll, resulting in a very accurate national projection of both vote share and seats.

Importance of Representativeness Over Sample Size

Lokniti-CSDS has continued studying voting behaviour through post-poll surveys. The 2019 Lok Sabha election post-poll survey had a sample size of just over 25,000. While seat projections sometimes missed the mark, the vote share estimates were very close to actual results.

Although a large sample size is important, representativeness ensuring the sample reflects various voter profiles is more crucial. Recently, television channels, often the sponsors of these exit polls, have pressured for larger samples, resulting in increasingly bigger exit polls.

Accuracy and Challenges

For example, Lokniti-CSDS’s predictions for the Chhattisgarh Assembly Elections were incorrect in both 2018 and 2023, despite being based on post-poll surveys. In the 2017 Uttar Pradesh Assembly elections, while the winner was predicted correctly, the final tallies for different parties were off the mark. Increasing the sample size would not necessarily have improved accuracy, suggesting other issues, such as potential fake interviews, may have affected the results.

Technological Solutions and Remaining Challenges

Technological advancements such as call-backs to respondents, images of interviews, field phone calls, WhatsApp groups, and similar tools have helped mitigate some issues. However, there is no foolproof method for ensuring accurate predictions.

The count method for estimating election results is both time-consuming and labour-intensive, requiring an estimate for each seat. When agencies claim to have made seat-wise estimates, it is often presented as the most comprehensive poll, especially when the sample size is several hundred thousand respondents.

Innovations in the Count Method

Some agencies have introduced innovations in the count method, maximizing efficiency by spending less time and resources. While an exit poll might claim to cover all constituencies, in practice, polls are not necessary in certain seats. For example, it may be unnecessary to conduct a poll in Varanasi where the Prime Minister is contesting, or in Gandhinagar where the BJP president is contesting. By carefully eliminating such constituencies, accurate estimates can still be achieved.

Focus on Swing Constituencies

After applying the elimination method alongside the count method, surveys are required only in a limited number of critical constituencies, often referred to as swing constituencies. Innovative exit polls focusing on these areas can be more accurate than those using traditional methodologies.

Limitations of the Count Method

While traditional methodology estimates vote share and analyzes voting behaviour based on various socioeconomic backgrounds, the count method has limitations. It can hardly provide an estimate of vote shares, and any systematic analysis of voting behaviour remains elusive.

• Many exit polls merely present several seats without providing vote share or methodological details. Should these even be considered exit polls? It is time to distinguish between real exit polls and estimate polls.
• A vote share estimate is mandatory for any poll. If a poll does not estimate the vote share, we must question the poll's purpose and methodology.
• What innovative method was applied that predicts seats without estimating votes? Without vote share estimates, the credibility and accuracy of the poll are highly questionable.