How reliable are exit polls? Data says not much!

How reliable are exit polls? Data says not much!

A man arranges Electronic Voting Machines(EVM),and VVPATs at a distribution centre on the eve of the seventh and last phase of Lok Sabha polls in Mirzapur. PTI photo

The exit polls have a history of underplaying the United Progressive Alliance (UPA), while it has historically overplayed the National Democratic Alliance’s (NDA) performance.

A DH analysis of the data for the past three general elections, collated by global investment bank UBS, reveals at an average the exit polls have overplayed the NDA’s performance by 4.3%, while it has underplayed the UPA’s performance by 8.9% at an average across all three previous elections.

What is more surprising is the fact, that of the previous three elections, the two in which UPA ended up forming the government at the center, the exit poll had got even the direction wrong.

In 2004, the average of the exit polls had predicted 252 seats for the then incumbent NDA, giving it yet another term at the center, and 186 seats for the UPA. However, in the actuals, UPA ended up forming the government with 219 seats to its kitty, while the Atal Bihari Vajpayee-led NDA, with 187 seats ended up occupying the opposition benches. The analysis reveals that the exit polls had overplayed the NDA’s performance by 25.8% back then while underplaying the UPA’s performance by 17.7%.

Again in 2009, when incumbent UPA returned to power, the exit polls had predicted a loss of seats for UPA. However, UPA ended up with a huge gain in the number of seats. Back then, the exit polls had predicted that UPA’s tally to diminish to 196 seats. However, the coalition, in reality, won 262 seats, a variation of whopping 33.67%. On the other hand, NDA, which exit polls expected to gain turf, won 159 seats against predicted 187 seats – overplaying the performance by 14.97%.

In 2014, when the country was riding on huge Modi-wave, though the exit polls got their directions right, yet they were completely off the mark when it came to the victory margin of NDA, like many other political pundits. While the exit polls gave NDA 274 seats, it ended up with 336 seats, a variation of 22.7%. On the other hand, the exit polls predicted 115 seats for the UPA, which faced a complete rout ending up with 60 seats.

DH, on Friday, after speaking three different market analysts, had predicted that incumbent Bharatiya Janata Party (BJP) would get around 240 seats in the exit poll on the standalone basis, however, the actual tally would be far lower, the analysts said.

The exit polls, in recent times, have had a history of guessing the results wrong – both locally and internationally. Domestically, post-2014, the psephologists had predicted an astounding victory of BJP over the then Mahagathbandhan, which was completely opposite of the actual results. In Karnataka state elections last, the pollsters had predicted ruling Congress retain power, however, BJP turned out to be the single largest party.

Internationally also, pollsters had got the Brexit vote wrong, along with 2016 US Presidential Elections – where they had predicted a victory for Democratic candidate Hillary Clinton. The most recent embarrassment for psephologists was during the Australian federal election results on Sunday morning, wherein they were predicting a victory for Australian Labour Party. However, incumbent Scott Morrison of the Liberal Party of Australia surprised everyone by sweeping the elections.