JEDDAH, 16 May 2004 — The much sought-after pollsters, who confidently projected their views before and during the general elections, have generally ended up with egg on their faces after their predictions proved way off the mark. Now with the results out the same pollsters are wishy-washy – wishing to wash away their projections with cautious embellishments.
No TV channels or newspaper projected a victory for the Congress and its allies, giving them a maximum of 160-200 seats. The NDA was projected to win 230-260 seats. But when the results came in, the shoe not only did not fit, but it bit. The Congress alliance got 224 seats and the NDA ended up with 189 seats throwing all the forecast awry.
Obviously, most surveys published ahead of the ballot and exit polls broadcast at every phase of the voting failed to read the voter’s mind.
This is not the first time that pollsters’ have gone horribly wrong. In the last West Bengal Assembly elections, almost every one of them had written off the Communists who have been ruling the state for a quarter of a century, and heralded the Trinamool Congress led by Mamata Banerjee as the new light of Bengal. But the Left Front, without the charismatic Jyoti Basu, romped home with a sweeping majority.
Even in this parliamentary elections, the pundits gave Mamata’s party the edge marking Trinamool as a major player in West Bengal’s politics. Come result day, only Mamata saved her seat.
Similarly, in the 1999 assembly elections in Bihar, the forecast was Laloo Prasad Yadav’s Rashtriya Janata Dal (RJD) was facing a rout with the prospects of winning only 50-60 seats with the NDA forming the next government in the state. Again the predictions proved hollow as Laloo’s party won 124 seats and the NDA 52 seats.
In a vast country like India errors are unavoidable. What is inexplicable is the margin. Yes, there are problems in conducting exit polls in this country. The attendant who works the sample has to run a whole gamut of it. Security forces prevent surveyors from going near polling booths to talk to voters. The voters too hesitate in telling their preferences. Communication facilities are also poor. And it is not necessary that the area, randomly chosen, of sample register a trend. So, invariably the margin of error in Indian projections is slightly higher than is the case in Western democracies.
There are other factors that affect the estimations. The exit poll is affected if the voter turnout is very high or very low. This is because exit poll projections are usually based on likely swing compared to the last election.
In a democracy of over a billion, a huge sample size is needed for any credible predictions. This is not feasible. Moreover, the voters pollsters talk to are mostly from urban areas, overlooking, deliberately or otherwise, the most important factor that India lives in villages and the voters of the remote areas are the ones who decide who rules in Delhi.
Now these pundits, who were confidently spouting their calculations without compunctions are taking cover with the banal assertion: “We may have been off the mark, but we were right about the general trend.”
This argument carries no weight. If the purveyors of survey seek to stay in this prediction business, they should get it right. This they can do only if they get their act together.
