Elections 2024: What's the Secret Sauce Behind a Good Exit Poll? Experts Answer

How and why an exit poll is conducted and what determines its accuracy–read what psephologists say in our deep dive:

As the 2024 Lok Sabha elections conclude, all eyes are now on exit polls to be released on Saturday, 1 June. How and why is an exit poll conducted, what determines their accuracy – psephologists answer these questions for you in our deep dive.

(Image: The Quint/@Vibhushita Singh)

As the high-octane 2024 Lok Sabha elections – being touted as the world’s largest democratic exercise – come to a close, their results are now piquing the stakeholders’ interest.  

But before the Election Commission (EC) declares results on 4 June, exit polls shed light on larger polling trends – will the BJP-led NDA government win a third term, are the odds favouring the INDIA alliance, who will be the next prime minister of our country, among others.

Check out key takeaways from various exit polls here.

“An exit poll is conducted by surveying voters exiting the polling booths on the day of voting. That is why it is known as an exit poll,” explained Sanjay Kumar, Co-director at the Centre for the Study of Developing Societies (CSDS).

He added that an exit poll is different from an ‘opinion poll’ or even a post-poll survey. 

How is an exit poll conducted, what was the need to conduct them, what determines their accuracy – psephologists answer these questions for you in our deep dive: 

What is the Need to Conduct Exit Polls?

One would wonder, what purpose do exit polls solve when the EC declares the result in 3-4 days after polling concludes? 

“The larger objective of exit polls was to understand why a voter polled in favour of a party or candidate. And to identify a pattern in case this reason varies based on the class, caste, gender, religion, location and education of the voter,” said election analyst Ashish Ranjan, who is also the Co-founder of Data Action Lab for Emerging Societies (DALES).

Ranjan added that the poll body does not share this data, that indicates which section of the society favoured which party. It only declares how many seats were won by individual political parties and breaks down polling percentage in the ratio of male: female.  

Kumar, however, said that the purpose of exit polls is only to “satiate the curiosity” of the leaders and political parties, who have contested the elections.  

What is the Difference Between Exit Polls and Opinion Polls?

Surveys around elections can be classified as pre-poll, which are conducted any time before elections are held and post-polls, wherein surveyors speak to voters at their houses or in their localities after the elections have concluded. 

“Exit polls are always conducted on the voting day, when the voter is exiting from the polling booth. But an opinion poll can be conducted any time before an election – be it two months or 10 days before election – and it gathers political opinion,” Kumar described.

He added that an opinion poll is a political survey consisting of a lot of questions which elicit the political/economic/social opinion from the people. It is more useful because the data gives a picture about the anxieties of the voters and can help candidates strategise their campaigns.

Post polls – conducted after the conclusion of elections – also have several questions that help understand the issues which weighed on a voter’s mind as he/she cast their ballot. 

“Since an exit poll is conducted outside a polling station, there is a natural limitation to the number of questions that can be asked. But the question on which candidate or party they voted for is asked, which gives us the vote and seat estimates,” Kumar reasoned.  

Opinion and post-polls, on the contrary, are more comprehensive and are conducted at the voters’ house. 

Timeline of Conducting Exit Polls in India

Exit polls were being conducted in the West for a long time but in India, they were conducted for the first time in 1957 with the help of Indian Institute of Public Opinion (IIPO). The institute was founded by Eric da Costa, an Oxford-trained economist who is popularly known as the “father of Indian polling.” 

The first scientific polls that were held in India, keeping the larger objective in mind, were commissioned by CSDS Founder Prof Rajni Kothari in 1967 to understand the issues of the people, Ranjan stated. 

“The aim of the survey is not to find who won how many seats but to bring out people’s issues and understand how people view the government’s activities and performance.”

Ashish Ranjan, Co-founder of Data Action Lab for Emerging Societies (DALES)

In 1980, Prannoy Roy and Ashok Lahiri had sought an opinion poll for the India Today magazine, marking the introduction of opinion polls in India.  

“In a regular manner, it started happening since the 1996 Lok Sabha elections, and since then many exit polls are published/broadcast after both Assembly and Parliamentary elections,” Kumar added. 

How is a Constituency Picked for Surveying?

“The survey need not be a big survey. The guarantee of an accurate survey is not the size of the sample but how representative the sample is,” said political analyst Rajan Pandey, who is also associated with survey agency People’s Pulse.  

Pandey illustrated the process through an example: 

• Let’s take an Assembly segment with 1 lakh voters. The segment has two blocks – A and B, and two towns – C and D. Now, an agency with more focus on sample size conducted a survey with 10 percent of the voters, i.e. 10,000 voters. But the agency surveyed all voters from town C.  

• Another agency, which laid emphasis on picking a more representative sample, first gathered the population of blocks A and B as well as towns C and D.  If A forms 25 percent of the Assembly segment’s population, then the sample should have 25 percent voters from A. Similarly, if 50 percent of the Assembly segment’s population is Muslim, 50 percent of the sample should be Muslim voters.  

“Even if the second agency surveyed 1,000 voters, it will be far more accurate than the first one because the sample is more representative; it represents the actual constituency,” Pandey described. 

The Lok Sabha elections are conducted in 543 constituencies across India.

“But surveys are usually not conducted in all 543 seats. We sample seats too based on regions. For instance, if Purvanchal region of Uttar Pradesh has 25 seats, we’ll sample five seats which represent the area. Similarly, if Bundelkhand region has six seats, we’ll sample one.” expounded.  

How is the Data Collected?

Pre-poll and post-poll surveys can be conducted in many ways – over the phone, through an app, qualitative, quantitative, etc.  

“Nowadays, call centres are roped in. The phone numbers of voters of a constituency are arranged and randomised. Call centre executives call voters, ask a set of questions and record their answers on the computer,” Pandey said. He added that apps are also being used to record data in real time. 

Kumar said that earlier, secret ballots were used to conduct surveys for exit polls, but the method is increasingly becoming obsolete. On being asked about if the trends on seat share are more important than vote share or vice versa, he explained:

“The data is always collected about voting percentages. You ask an individual which party they have likely voted for. So, you are collecting the political preference or voting preference of different people. You aggregate those voting preferences – for instance, if out of 100 people, 70 have voted for party A, you get a sense that A will win this seat. The vote share is converted into seats retrospectively."

Sanjay Kumar, Co-director, Centre for the Study of Developing Societies (CSDS).

Which Exit Poll is More Accurate, Reliable?

While the accuracy of a poll is determined by several factors – sampling, questionnaire design, fieldwork, how one collects and interprets data, etc – it is important for any polling agency to release the data and all the details of the poll.  

"But the sad thing is opinion poll agencies don’t release this data. It's like a doctor prescribing you a medicine without telling you what illness you have. If the exit poll only gives you data on seats, how can you distinguish which one is more reliable?” Kumar remarked. 

Ranjan, however, claimed that in the last ten years, the focus of exit polls has shifted to only the number of seats a political party is winning, followed by their vote share. “This undermines the larger objective of the exit polls,” he said. 

Another major limitation, according to Ranjan, is that agencies do not declare how representative their sample is, which is a major indicator of the poll’s accuracy.