Mission not accomplished

For the exit polls that crowded television screens as soon as the last vote was polled in the UP elections, it was once again a case of too near yet too far. While television anchors screamed out adjectives such as “sabse bada khulasa” (biggest revelation), “election ki chaabi” (the key to the election) and “jaadu ka pitara” (magic box) to their exit polls, election results proved otherwise. Here is a roundup of what the post poll surveys said.

The News 24-Today’s Chanakya poll was first off the block and it gave the SP 185 seats, the BSP 85 and tied the BJP and Congress at 55 seats each. The biggest shift of votes – eight per cent – was projected to be in favour of the SP while the BSP was set to lose seven per cent votes. A similarly big gain for the SP was predicted by the Star News-AC Nielsen survey which gave the SP 183 seats and the BSP 83 seats. It projected the BJP as the third largest party with 71 seats and gave 62 seats to the Congress and its allies.

The Sahara News poll also put the SP at the top with 188 seats, confined the BSP to 83 seats, gave the BJP 64 and the Congress 55 seats. It also predicted that the SP would increase its vote share in minorities, snagging 44 per cent of the votes against the 2007 figure of 38 per cent. While BJP was projected to increase its vote share among general voters (although falling behind with the OBC and SC voters), for the Congress-RLD combine the OBC/general vote share was projected to have increased.

However, the NDTV polls gave SP only  95-105 seats. The BSP was tipped to emerge the biggest party with 120-130 seats, the BJP was shown to vastly improve its 2007 tally with 115-125 seats while the Congress was left with just 35-45 seats.

The Headlines Today figures, with 203 seats in the SP’s kitty were much closer to the actual figures. It gave the BSP 88-98 seats, pegged the BJP at 50-56 and the Congress alliance at 38-42. Zee News also came close to the actual SP tally with a figure between 190 and 210 seats. India TV-C Voter on the other hand gave 141 seats to the SP, 126 to BSP, 83 to BJP while wrapping up the Congress at 36.

The CNN-IBN-The Week (CSDS) poll looked at likely vote percentages and gave the SP 34 per cent of the share. The BSP got 10 per cent less at 24 per cent while for the BJP the figure was 14, better than the Congress-RLD share of 12 per cent. Notably, it put the Congress share at even lesser than the others which were likely to get 16 per cent of the vote share.

Numbers apart, all polls converged around a few points. One, they gave the maximum number of seats to the SP (NDTV being the noteworthy exception). With the exception of Headlines Today and Zee News, all put down a clear verdict as an impossibility. Three, all predicted a decline for the BSP, though the figures stretched from 83 to 130. Four, almost all surveys limited the BJP and Congress to the last two spots but with vast differences in seat projections. Also, all surveys predicted that both parties would improve their 2007 tally.

But these projections were rather obvious. Of course the BSP was suffering anti incumbency and would slip. The biggest gainer would of course be the SP and some of the gains might flow the way of the BJP. And again, of course the BJP and Congress could scarcely be expected to work miracles in the elections. On the actual test, all exit polls plummeted.

No one for instance predicted the SP’s monstrous tally of 224 (the second highest in any election in the state since Independence). Similarly the BSP’s fall to 80 from 206 was not foreseen by any survey. Exit polls also fumbled in deducing that the BJP would improve its tally. The improvement for the Congress was similarly a niggardly 28, up six from its 2007 tally of 22.

Vote percentage too (as calculated by the CNN-IBN-Week survey) were not a faithful barometer of the number of seats finally won. Thus, while SP captured some 29 per cent of the votes – a four per cent lead over the BSP, in terms of seats, this translated into 129. Now set this against the 2007 result to understand why percentages and seats do not necessarily bear a direct relation. In 2007, the party had seen a marginal increase in its share of votes, yet in terms of seats its count fell from 143 to 97. Contrast this with the Congress – an increase of two per cent in vote share, but in terms of seats – only six more. For comparisons’ sake, the science of exit polls proved to be as (un)reliable as the projections of bookies in Lucknow which laid the greatest bets on chances of President’s rule in the state (80-84 per cent). The possibility of a SP-Congress alliance stood at 76-79 per cent chance followed by a BSP-Congress tie up at 45-55 per cent.

Exit polls in 2012 have proved true to their image of being the staple of lazy but loud intellectualism. Flashback to 2007 when no survey predicted the BSP’s final mammoth tally of 206 and many had in fact predicted the BJP’s emergence as the single largest party to understand why Chief Election Commissioner YS Qureshi tweeted that such polls were best left to “entertainment channels”.

In 2007, at the top of the deluded heap lay NDTV (the trendsetter of post poll surveys) which predicted 108-118 seats for the BJP against the 113-123 for the SP. The actual tally was 51 and 97 respectively. Similarly against an actual of 206 and 22, the BSP was granted between 117-127 seats while the Congress tally was put at 35-45. CNN-IBN fared better in predicting 22 seats for the Congress and the lower limit of its prediction for the SP (99-110 seats) could also be termed passable. Once again the biggest slip happened in the case of BJP which was predicted to be getting 80-90 seats.

Times Now gave 116-126 seats for the BSP, 114-125 for the BJP, 100-110 for the SP and 25-25 for the Congress. Ditto for Star News which gave 108 seats to the BJP and only 96 to the SP. The figures for BSP and Congress were 137 and 27. Yet on comparative terms it redeemed itself by coming close to the actual figures for SP and Congress.

If the BSP appeared unconcerned by the exit poll numbers, it had history to go by. Not just in  the last election, in the 2002 polls too exit poll numbers had been all wrong – what with all surveys giving the party between 75-90 seats. It ended up with a total of 96 seats and 23 per cent of the total votes.

Umesh Chaturvedi, social scientist says that the great flaw in exit polls is that they address a yet immature electorate. “Our democracy is not as mature as that of America where voters openly declare their republican or democratic credentials. People vote for someone and say they have voted for another”, he argued.  The reasons for this is that in UP caste divides are strong and identities are sharp. The very latest example comes from Sitapur (UP) where SP workers set on fire the huts of supporters of an independent candidate. In such a scenario it makes no sense for the voter to give an accurate account of his voting habit.

Shafqat A Husain, a Delhi based statistician avers, “All exit polls start with certain perceptions such as Dalits will vote only for BSP. The variance in numbers comes only in the vote that lies beyond these assumptions. Mere statistical sampling will never give an accurate figure unless backed by ground research. The matter of chunavi hawa is the key.”
Kameshwar Choudhary, Dean, School for Management Studies, Dr BR Ambedkar University, Lucknow says, “Methodology and sample survey are crucial for exit polls to be reliable. Our society has multiple differences. To capture these accurately one will have to go to the remotest corners. Great professional expertise is required on the part of those who analyse survey results. It is unlikely that most pollsters would be willing to put in that kind of effort.”

There is also something to be said about the proliferation of news channels and the media’s fascination with opinion polls – in a country that polls everything from who is best suited to be India’s next prime minister to possible names for Aishwarya Rai’s newborn. Not all the agencies which conduct these polls are backed by scientific expertise. It also makes zero sense to draw conclusions from a sample size of thousands in a voter population that runs into crores. Even within sample sizes, to maintain the ratio of castes, sub castes and religions is a task next to impossible.

In UP, this time the task was further complicated by the large presence of first time voters. While family influence may prove to be the strongest deciding factor for which way the young vote goes, it cannot be discounted that the first timer would respond with a more politically correct answer when asked. Thus, it makes better sense for the youth to say they picked the SP because of its young face and not admit that they threw in their lot with the BJP, what with the party’s all pervasive communal taint. Similarly women voters might be loath to admit to voting for the SP that comes with the “goonda party” tag. Because exit polls require vocal responses from the sample, they are by default geared towards the more assertive sections of the population and are likely to miss out on minority voices. Moreover, while men in certain communities will speak, women will not. This distorts the picture further.

Thus, there are few (except projected winners) who will give credence to exit polls (on Akhilesh Singh Yadav’s facebook wall though, eager SP workers kept posting exit poll numbers to celebrate the party’s imminent coming to power). In fact, Digvijay Singh, Senior Congress leader and in-charge of the UP elections dubbed the polls a “fraud” as they tried to predict results from a small sample size. At best such polls can be indicative of trends. To offer them any more importance is to discredit intelligence.