2004 exit polls: When surveys got it horribly wrong

New Delhi, May 20: Exit polls can predict a general trend and give a rough idea about what the actual election result, but these surveys can by no means be taken as gospel truth. The entire nation seems to believe that the poll results have been declared after exit poll predictions came out. BJP supporters are congratulating each other, sending messages on whatsapp as if it were certain that Modi would take oath as next PM.

Enthusiasm over Exit poll predictions is fine, but going overboard is certainly not required as Exit polls do not have a reliable track record of forecasting electoral outcomes.

In the 2004, the average of all tallies of exit polls greatly exaggerated the BJP-led National Democratic Alliance's tally. Atal Bihari Vajpayee's term had just ended and the BJP was upbeat about the Lok Sabha elections 2004. India had witnessed steady economic growth under Vajpayee, and BJP in general had become synonymous with progress due to reforms.

What happened then before the elections, the general perception, the survey predictions and the way India Shining campaign seemed to strike a chord, all seemed favourable for the NDA.

The NDTV-AC Nielsen exit poll predicted 230-250 seats for the NDA and 190-205 for the Congress, while other were expected to get 100-120 seats. The Aajtak ORG-MARG exit poll gave NDA 248 seats and Congress 190, while predicting 105 for the others. The StarNews C-Voter survey predicted 263-275 for NDA, 174-186 for Congress, and 86-98 to others.

The average of exit polls gave the NDA 252 seats, whereas it actually got 181. The BJP could win just 138 seats and Congress got 145. The final tally was this - UPA-208, NDA-181, Left Front-59. Partywise, Congress-145 and BJP-138.

In 2009, exit polls put the NDA at 187 and the UPA at 196, but the actual results were 159 and 262 respectively.

Even in the United States, practically all polls favoured Hillary Clinton, but it was Trump who eventually became the President. Exit polls may not be able to guage certain subtle undercurrents that effect voting choices. Sample may not be a true representative of the population, no matter how expertly the poll is designed.

Exit polls and accuracy:

Most of the exit polls have predicted Prime Minister Narendra Modi's return to office. Exit polls predictions can be suggestive but never be conclusive. This is because of the inherent limitations in the methodology used to conduct an exit poll survey. Exit poll predictions are based on a sample size, now is the sample a true reflection of the population? Does it take into account all the possible diversities in the society? Experts who design these surveys do their best to ensure that the sample reflects the diversity in actual population as much as possible, but the errors cannot be ruled out. That said, the surveys cannot be total opposite of the actual outcome.