In the 1948 US presidential election, even before the official results were out, most newspapers declared Republican Thomas E Dewey the 31st American president on the basis of poll surveys. The one to take office, however, was democrat Harry S Truman.
Election Commission officials seal an Electronic Voting Machine (EVM) prior to the start of voting at a polling station in Dibrugarh, Assam. (AFP photo)
In India, too, exit polls – almost indigenously developed by the pioneering Delhi-based Centre for the Study of Developing Societies (CSDS) in the 1960s – have had a fair share of failures. The last four Lok Sabha elections show they have suffered wide inaccuracies, barring the one in the 1998 general elections.
Two back-to-back flawed predictions — the 2004 and 2009 exit polls — have posed a question mark on their reliability. Some see an existential threat.
“A correct forecast by media opinion polls this time will wash away the sins of not getting it correct in the last two Lok Sabha elections and restore lost credibility. But if they go wrong, they will do irreparable damage to the opinion-polling industry, driving a wedge between the media and those who practise survey research,” says Praveen Rai, a psephologist with CSDS.
In the 1999 elections, forced by an early collapse of the Atal Bihari Vajpayee government, most polls overrated the NDA’s win. They gave the alliance a thumping 315-plus seats, although it actually scraped through with 296.
The 2004 exit polls, however, were a glaring failure. All pollsters did not see the Congress clawing back. While some forecast a swing in favour of the incumbent BJP, others saw only a minor dip in its tally. But all exit polls predicted a win for the BJP.
For the first time in many years, the Indian economy grew at 8%. The BJP launched its glitzy India Shining campaign, which became a catchphrase for an emerging economy. However, a disconnect ensured a shock defeat.
In 2009, the exit polls weren’t quite spot on. Although most predictions put the UPA ahead, they entirely missed the key trend: big swings in its favour. The UPA bagged 40 more seats without necessarily adding anything to its vote share.
The first serious media poll surveys started surfacing in the 1980s, with psephologist Prannoy Roy teaming up with David Butler. Their studies culminated in the iconic book The Compendium of Indian Elections by Prannoy Roy, David Butler and Ashok Lahiri.
The advent of satellite television lent exit polls blockbuster prominence, since the state-run Doordarshan’s commissioning of a countrywide exit poll in 1996 to CSDS.
While most surveys have a margin of error, the survey method itself and sampling are the keys to accuracy. A large sample size may not be adequate if it does not have enough representation of diverse sections of the electorate, such as castes.
While academics prefer random sampling from actual voter lists, which are deemed more accurate, market research firms tend to go by quota sampling, or picking samples from different electorate groups, such as farmers, professionals and labourers.
As a forecaster, Praveen Rai finds the prediction of the NDA winning seats in the range between ‘220-plus to a majority’ with such definitiveness this time intriguing. “It blurs the divide between the predictive prowess of forecasting the results and the actual unfolding of the election outcome,” he says.