Exit poll predictions vary

Deepshikha Ghosh

New Delhi, April 21:

Indian political parties and pundits engaged in a war of words today after exit polls gave wildly fluctuating projections after Round One of the staggered parliamentary elections.

Depending on which way the verdict swung, political parties welcomed them or trashed them, but pollsters who conduct these exercises admit exit polls are not accurate. "You have to take these polls with a pinch of salt," said an official of A.C Nielsen, which has executed the fieldwork for Indian Express-NDTV.

"We operate under extreme constraints - the time is limited, since we have to produce results for channels to flash them by 5 pm when the balloting is over, and we have no control over the sample size," the official said. After the first phase of polling yesterday in 140 of the total of 543 parliamentary constituencies across the country, the Express-NDTV exit poll gave 75 seats to the Bharatiya Janata Party (BJP)-led alliance, 53 seats to the Congress and its allies and 12 to the rest.

Of the 140, the BJP-led National Democratic Alliance had won 88 seats in 1999 and the Congress and its allies 45. The channel AajTak projected 93 seats for the BJP-led alliance, 44 for the Congress and allies and three to others. Zee television put the figures at 63-78 for the BJP plus allies, 37-50 for the Congress plus allies and 16 for others. STAR News projected 80 seats for the BJP-led coalition, 53 for the Congress and its allies, and seven for the other parties. The Sahara-DRS exit poll projected 82 for the BJP plus allies, 55 for the Congress plus allies, and three for others. The AC Nielsen official explains why the procedure is faulty.

"When we cover a constituency on polling day, we have to take a sample of one booth. We have from 7 am till about 2.30 pm to speak to voters." So at a given time, a bulk of voters could support one party. Thus the exit poll conducted by different agencies is bound to vary.

Pollster GVL Narasimha Rao, who has conducted the Sahara exit poll, said he could not

completely disagree with the observation.