Pollsters get it wrong
Voters in Rajasthan and Chhattisgarh upset the calculations of pollsters. All pre-poll surveys went awry for both states. Only two of four exit polls predicted a BJP victory in Rajasthan. Only one exit poll got Chhattisgarh right. Most pollsters agree there was a saffron swing in Rajasthan in the last few weeks before the polls. "In the last month things changed fast in that state," says C-Fore's Premchand Palety. Most pollsters failed to catch the shift as their polls were done three or four weeks before election day.india Updated: Dec 05, 2003 15:43 IST
Voters in Rajasthan and Chhattisgarh upset the calculations of pollsters. All pre-poll surveys went awry for both states. Only two of four exit polls predicted a BJP victory in Rajasthan. Only one exit poll got Chhattisgarh right.
Most pollsters agree there was a saffron swing in Rajasthan in the last few weeks before the polls. "In the last month things changed fast in that state," says C-Fore's Premchand Palety. Most pollsters failed to catch the shift as their polls were done three or four weeks before election day.
Yashwant Deshmukh, whose C-Voter did polls for Zee News, says their first two Rajasthan surveys gave Congress a decisive edge. The third survey, just 10 days before elections, caught a pro-BJP swing that was already touching 5 per cent.
Even this was partly underestimated. After all, the Congress had won Rajasthan with an 11 per cent margin last time.
But the BJP's Rajasthan campaign gained even more momentum in the last 10 days, says Ranjit Chib of AC-Nielsen. Senior Congressmen were few on the ground in Rajasthan. BJP heavyweights were everywhere. The result: turnout was seven percentage points higher than usual. "All this went to the BJP," says Deshmukh.
This sort of poll management proved crucial. Palety notes the Congress had four times as many rebel candidates than the BJP in Rajasthan.
AC Nielsen-Zee News-Outlook's was the only pre-poll survey that declared Chhattisgarh to be an even fight. Chib says the two surveys his outfit conducted had, in fact, given a slight edge to the BJP.
Deshmukh and Chib both cite bad vibes about Chief Minister Ajit Jogi as being decisive. AC Nielsen found a third of voters actively disliked Jogi. Pollsters say the Dilip Singh Judeo tape had little impact: voters believed Jogi "was the bigger crook".
Deshmukh says voters were also displeased at Jogi engineering defections in the BJP. "This was not part of Chhattisgarh's political culture and strengthened the anti-incumbency sentiment against Jogi."
One manifestation of Jogi-phobia was a higher than normal turnout — and a boost to the BJP. "Congress MLAs were popular. Their chief minister was not," said one pollster.
All polls were in agreement that the BJP would win Madhya Pradesh. However, only The Week-TNS-Mode's pre-poll survey captured how big a landslide wouldbury Digvijay Singh.
All the pollsters agree Singh was facing huge anti-incumbency sentiment. "There were three layers: voters didn't like the local Congress MLAs, the government in Bhopal and the chief minister," says Deshmukh.
Pollsters erred in assuming Singh would blunt the wave. C-Voter assumed his alliance with the BSP would give the Congress the edge in 30 or so seats. It didn't.Pollsters admit they trimmed the Congress' margin of defeat because of a feeling that Singh, who befuddled pollsters in 1998, had a trick up his sleeve. He didn't.
Delhi threw up the least surprises. Sheila Dikshit pulled off the sort of sweep all surveys had indicated. Most of the pre-poll surveys were off only in terms of margins. The HT/CSDS poll, for example, predicted a 48-seat margin between the BJP and the Congress. The final figure: 27 seats.
Pollsters differed, however, on whether the Delhi voter was exceptional given that all the other sitting chief ministers had been dethroned. Some argued Delhi voters tend to see the party at the Centre as the incumbent. Deshmukh says there is evidence of local anti-incumbency, with the two parties exchanging seats.
Others are unconvinced, noting that a majority of Delhiites place Dikshit and Prime Minister Atal Bihari Vajpayee in equally high esteem. Palety believes anti-incumbency was less important than poll management in all the elections. Pollsters admitted these elections were a reminder of the limits of their craft.
Palety says he was surprised at how many exit polls stumbled: "There are no undecided voters at this point. This should be more accurate." Chib disagrees: "We can't control who agrees to an exit poll. Samples are often unrepresentative." Worse, many exit polls closed shop after lunch and missed the late afternoon voters.
High turnouts threw off many calculations. Pollsters said there are certain unknowables in surveys. One is the effect of rebel candidates. Another is turnout.