Is exit poll science or a means of entertainment? Understand how victory or defeat is assessed

Writer- Atan Biswas

After voting in the Lok Sabha elections, exit polls will dominate the media today. It is surprising what use these exit polls can be, knowing that the actual results will be known only after three days. Many people have been waiting for the results for several weeks after voting. They can certainly wait for three more days.

A hollow exercise?

Exit polls are predictions that have no bearing on elections. They will not influence voters (as opinion polls can in many ways), and many people probably don’t trust them. It is hard to know which exit polls to believe, if any, because there are so many of them, and their predictions vary widely. Any party trailing in the exit polls can claim they are wrong. However, such distrust cannot be faulted.

Strange history

The reliability of exit polls was not so low nearly two decades ago. Exit poll results bolstered Georgia’s 2003 ‘Rose Revolution’, which resulted in the ouster of Eduard Shevardnadze from power. The recall referendum of Hugo Chavez in Venezuela in 2004 caused a massive uproar because of the wide discrepancy with exit polls. Exit poll estimates also played a role in the 2004-05 ‘Orange Revolution’ in Ukraine.

Richard Charnin, in his/her 2010 book Proving Election Fraud, discussed possible irregularities in Democrat John Kerry’s 2004 US presidential election defeat. Political commentators argued that the election had ‘behaved badly’, although Charnin took into account the possibility of undiscovered votes and ghost votes. The real evidence was the final national exit poll, which showed a 6.5 percentage point difference in vote share between Kerry and George W. Bush!

Exit vs Opinion Poll: What is the difference between opinion poll and exit poll, clear your confusion before the result

How was your experience in India

Interestingly, around the same time, the credibility of exit polls began to decline globally. The UPA’s surprise win in the 2004 Lok Sabha elections went against the predictions of most exit polls. Even though most surveys predicted the NDA’s victory in 2014, many were unable to predict the massive win. What a disparity there is in 2019 too. One exit poll predicted 242 seats for the NDA, while another predicted 365 seats.

Not even in the headlines at the global level

This is the case across the world. Be it 1992 or 2015, exit polls in Britain have failed spectacularly. Exit polls in the 2000 US presidential election showed that Al Gore was winning in states like Florida. Being ‘fooled’ is not only due to ‘randomness’ but also due to many other reasons. For example, Republicans in the US have always been reluctant to participate in such studies. This is called ‘non-response bias’. A similar ‘shy Tory factor’ was seen in Britain in 1992 when exit polls wrongly predicted John Major’s victory. The story of ‘shy Trump voters’ gained momentum after the 2016 US presidential election.

Explained: Exit polls will tell whether BJP will form the government or not, know how accurate they are

Possible Solutions

There can be different types of non-responses in India too, depending on the situation. To circumvent this, a small-scale survey can be conducted before the exit poll to find out the percentage of supporters of various parties who are interested in participating in the survey. Additionally, a statistical method known as ‘random response sampling’ can be used to determine the actual percentage of people who support certain parties. But do today’s pollsters follow fundamental statistical principles while planning, selecting and interpreting their data? Theoretically, an exit poll should interview every fifth or eighth voter of a particular booth, which is basically a systematic sampling. Importantly, enthusiastic voters willing to express opinions in between should be ignored. In fact, most pollsters do not disclose sample size, sampling frame, sampling and estimation techniques or summary data at various margins of uncertainty. Thus, from a statistical point of view, it is almost impossible to comment on the quality of their forecasts.

A big gamble!

It is clear that exit polls do not necessarily indicate the final results of elections. It is quite common that some exit polls predict a win for Party A, some for Party B, while others show a close contest. Since pollsters around the world have made election prediction a fun game, why exit polls are still conducted may be a very relevant question. Is it to entice impatient voters in the final hours of their agonizing wait, or because it is an instant revenue-generating option for the media? Either way, exit polls today are just another game show amid the endless excitement of elections.

(The author is Professor at the Indian Statistical Institute, Kolkata)