Web Exclusive
Exit Polls 2019: Be careful; take their forecasts with a bucketful of salt

We are at that stage in this election season when the term ‘exit poll’ starts getting popular. This long, bitter and hard fought election is about to come to a close with only one phase of voting left. Fifty-nine seats will vote on Sunday, May 19, to elect their representatives and with it 543 seats would have completed this exercise.

As voting hour comes to a close this coming Sunday, an embargo will also get lifted. This will be the embargo on broadcast of exit polls by media organisations. The Election Commission prohibits this broadcast till the time every vote has been cast, so as not to influence voter behaviour.

An exit poll is different from an opinion poll. While an opinion polls seeks to find the views of a prospective voter before he exercises his franchise (at any stage before the act of voting), an exit poll is conducted immediately after a voter casts his vote and exits the polling station. Because this is recorded after the event, it is believed that an exit poll is likely to be more accurate in predicting the final result. While the results of General Election 2019 will be known on 23 May, exit polls may be of interest to various people. But how accurate have they been in India in general? We decided to take a look at past instances to see how precise or off the mark they have been in predicting the results, especially in general elections.

2004

We start by taking a look at exit poll numbers for General Election 2004. This was the 'India Shining' election when the Atal Bihari Vajpayee-led NDA government advanced the polls hoping to romp home to a majority. The table below shows the predictions made by pollsters after voting had been completed.

As the table above shows, the NDA’s hopes of returning to power were belied and it ended up winning fewer seats than the Congress and its allies. Nearly all the exit polls predicted that the NDA would win close to 250 seats (two predicted they could win a majority) while every poll (except one) put the Congress and allies at below 200.

Not only were the polls wrong on the count of seats, they were completely wrong even when it came to predicting the trend, as the NDA ended up trailing Congress and allies in a surprise result.

2009

After the shock defeat of the NDA in 2004, the Congress formed United Progressive Alliance and with the help of the left parties (CPI-M, CPI), formed the government at the centre with Dr Manmohan Singh as prime minister. In 2009, the UPA faced the L K Advani-led NDA in the elections, which took place in the shadow of the Indo-US Nuclear Deal and the terrorist attack in Mumbai on 26 November, 2008. The table below shows exit poll predictions for General Election 2009.

Like 2004, exit polls were found to be off-the-mark even in 2009. The UPA increased its gap over the NDA by performing better than expected in urban India. It won 262 seats and reached a comfortable position where it did not need the support of the Left parties either. Exit polls had predicted a very close election. There was not much to choose between the two alliances in terms of seats. Once again, exit polls were off the mark, not only in terms of seats but also in terms of trend.

2014

In 2014, the UPA was facing a challenge from a resurgent BJP, with Narendra Modi as its Prime Ministerial candidate. The then Gujarat CM’s popularity was at its peak even as Dr Manmohan Singh and the UPA had been badly affected by various corruption scandals. Exit polls had predicted that the NDA would trump the UPA. Some exit polls gave the NDA a clear majority and predicted a strong drubbing for the UPA.

The table above shows that while the exit polls were correct in picking the broad trend in 2014, they were once again wrong in making predictions on seats. The NDA won a major victory but the BJP surprised the country by winning a clear majority on its own. All exit polls (except News 24 – Chanakya) underestimated the losses for the Congress and UPA.

As the examples of three general elections since 2004 show us, exit polls have had a poor record of predicting the actual result of an election in India. This is largely true even for Assembly polls as well. We take a look at two assembly elections since 2014 for more on this:

Uttar Pradesh Assembly Elections 2017

In Uttar Pradesh, the incumbent Akhilesh Yadav-led Samajwadi Party tied up with the Congress party to stave off a challenge from the BJP in the assembly polls. Different organisations predicted different results in these polls. Some predicted a clear win for the BJP while others predicted a hung assembly.

The results of the Uttar Pradesh assembly, however, shocked everyone. The BJP won an unprecedented landslide victory. While two exit polls predicted a win for the BJP in the elections, only one predicted a big win. The actual result was a decimation of the opposition in the state with the BJP winning a 3/4th mandate in the state.

Punjab Assembly Elections 2017

After a decade-long rule by the Shiromani Akali Dal and BJP alliance in Punjab, there were two challengers to the throne for the incumbent. These were the Congress party and the new kid on the block, the Aam Aadmi Party. AAP had earlier surprised everyone by winning four Lok Sabha seats in General Elections 2014. The Amarinder Singh-led Congress party was trying to return to power, even as it was trying to ensure that AAP did not grow roots in the state. Exit polls predicted a big loss for the Prakash Singh Badal government. They were not sure about the winner of these elections though. Exit polls predicted an even split between the Congress and the AAP and a rout for the SAD-BJP.

The actual results were not close to any of these predictions though. The Congress won a landslide victory with 77 seats out of 117. The AAP was a distant second with just 22 seats and the SAD-BJP did better than any exit poll prediction by winning 18 seats.

Conclusion

Opinion polls and exit polls are now an integral part of our political universe. They are a part of the electoral news cycle and therefore will have their following on 19 May. But, given India’s complex demography, the scale of the sampling challenge and the incredible diversity, polling is a difficult exercise. The inaccurate results of various exit and opinion polls only highlights this challenge. For the general public, it would help to take these predictions with a bucketful of salt.