It's a busy time for psephologists. As Gujarat's electoral drama moves towards its climax, the poll pundits who have words like "anti-incumbency", "vote swing" and "bandwagon effect" to explain political karma are on centre-stage along with the politicians whose fates they foretell.
The election results will throw up winners and losers among them too. Psephologist Naveen Surpaneni, Director of the Centre for Media Studies (CMS), says "By 11.30 a.m. we'll know who will form the next government".
Of four opinion polls in Gujarat, two said the BJP would win. One predicted a narrow Congress victory, while one had them equal. Two exit polls were also conducted. While Zee News gave the BJP 100-105 seats with a 57 per cent vote share, Aaj Tak gave the party 93-109 seats with a 44 per cent vote share. The variance in vote share figures indicates that both cannot be correct.
"There is some anomaly in conversion from vote share to seat share", says Professor P.M. Patel of M.S. University in Vadodara. He is the Gujarat coordinator and co-author of the CSDS-Frontline opinion poll, which gave the BJP a 15 per cent lead in vote share — enough for a landslide, he says.
Yashwant Deshmukh, the expert who's doing the poll analysis for Aaj Tak, says a two-horse race like Gujarat becomes difficult to predict if the difference in vote share is small.
{{/usCountry}}Yashwant Deshmukh, the expert who's doing the poll analysis for Aaj Tak, says a two-horse race like Gujarat becomes difficult to predict if the difference in vote share is small.
{{/usCountry}}"A small margin in vote share can then make a large difference in seats", he says. Surpaneni agrees. "In a close fight, the per cent of error becomes critical", he says. In psephology, a 3 per cent error is considered acceptable. The problem happens when only 3 per cent of votes separates winner from loser. Surpaneni expects the difference in vote share between the BJP and Congress in Gujarat to be around 3-4 per cent. The Aaj Tak poll says the same thing, and even the Outlook-Cfore poll, which had the Congress winning, had a similar figure.
In such a situation, the segment-wise turnout is critical, says Surpaneni. "For example, in Saurashtra, are the Patels voting or are they sitting at home?"
That's the sort of thing political parties pay to know. A complete election survey package can cost Rs 10 lakh, and are often financed indirectly by political parties. All of them consult psephologists while formulating strategies, though they may bash their own agencies on TV. The reason, says Deshmukh, is that "even if it's guesstimates, our guesstimates are better than theirs". And even if the psephologists themselves are, as Patel says, "groping like seven blind men fathoming the elephant".