Opinion polls and exit polls go wrong. A journalist with her ear to the ground can perhaps detect sounds not heard by the surveyor.
I am late by half an hour for Mayawati's rally in Sultanpur. Yet, no harm done. The ground is just filling up and there is time for a quick snack at the nearby dhaba. As the car reverses, my driver Sunil, a Thakur by birth but nonetheless an ardent fan of Mayawati, points to a most astonishing sight. Thousands and thousands of women, babies tucked under their arms, are trudging on foot to the rally ground. By the time we return, the venue is overflowing with people, the rush of women we saw making up nearly half the audience.
The gentle morning sun grows fierce but there is no sign yet of the Bahujan Samaj Party (BSP) supremo. Through the five hours that follow, the crowds sit motionless, not taking a break, not even asking for water. I ask a local intelligence officer, who is quietly watching over the crowd, to estimate its strength. Fifteen thousand, he says, adding, "But it is not just the numbers. Look at the quality of the people. They are not curious onlookers. They are voters."
Indeed so. The audience listens in rapt attention as a score of lesser functionaries make speeches extolling Mayawati and her multi-caste social engineering experiment. Sarvajan (all-caste) unity is also the theme of a qawwali rendered in a mix of Bhojpuri and Awadhi. The only time the large assemblage of men and women stirs is when slogans are raised at the end of a speech: "Behen Mayawati zindabad, Sarvajan zindabad," they shout after the speaker.
Mayawati's arrival breaks the discipline, but only momentarily. Men and women rushing forward for a glimpse of their leader are motioned back to their seats by volunteers of the Bahujan Samaj Force, a paramilitary-like organisation in charge of security and crowd control at Mayawati's rallies.
The BSP chief is not the best of orators. Her tone is monotonous, her speech without humour and without the flourishes of her rivals. It does not help that she reads out from a prepared text. But her listeners hang on to every word. She tells them the BSP is now a party of all castes, all religions, even as she matter-of-factly provides the caste and denominational break-up of her candidates: 139 from the forward castes, 110 from the Other Backward Classes (OBCs), 93 from the Scheduled Castes and 61 Muslims.
"Haathi nahi Ganesh hai, Brahma, Vishnu, Mahesh hai" (it is not elephant but it is Ganesh, it is the Hindu triumvirate of Brahma, Vishnu and Mahesh), she declares to thundering ovation. This is followed by a message especially addressed to her core constituency - Dalits, who know just what is at stake in this election.
Mayawati's list of dos and don'ts for voting day misses no detail. "Keep a fast for a day," she tells the women, urging them not to light the chulha (stove) until they have finished exercising their franchise. The menfolk are advised not to fall prey to distractions. "People will try to divert your attention. They will say one of your fellow Dalits has had an accident. Even assuming this is true, first cast your vote and then attend to your friend," she says, adding, "And yes, you'll find leaflets distributed in my name, asking you not to vote for x or y candidate. Pay no attention. This is a ploy to prevent you from voting."
As she winds up her speech with a full-throated cry of "Jai Bhim", the gathering, too, rises to its feet, filling the four corners of the pandal with deafening shouts of "Jai Bhim".
The rally is educative. The day before, travelling between Lucknow and Barabanki, I had heard Rahul Gandhi and Lal Krishna Advani. Rahul attracted rapturous but inattentive crowds. They wanted to see the celebrated Gandhi heir, not hear him. Virtually the same lot turned up to hear Advani. This time they were visibly restless. What caught their fancy was a foreign television crew filming the rally.
Advani spoke on the strengths of Indian democracy, dwelling at length on United States President George Bush's doubt-filled first election. The crowds were waving to the camera. Exasperated, Advani remarked: "I suppose now you will all clap for the camera." And, indeed, they clapped for the camera, much to the BJP leader's chagrin.
Clearly, as the local intelligence man in Sultanpur put it, the quality of people turning up to hear Mayawati is vastly superior. They are not rally-hoppers. They are voters. For me, searching for pointers to which way the vote would go, Mayawati's captive audience is a valuable clue. But there are other indicators.
In Lucknow, where I start my journey, local journalists breathlessly talk about an election that has not been this free and fair in decades. They eulogise the Election Commission of India for making this possible and speak of Dalits in the remotest villages trooping out to cast their votes - in many cases for the first time since Independence. "This is a miracle," they say.
As I travel onward from Sultanpur, I hear reports of unprecedented crowds at Mayawati's rallies. In Ghazipur, I miss the lady by a whisker but hear the local people gush about the 50,000-strong turnout at her rally. For me this is crucial piece of evidence. At election time a crowd of 10,000 is rare; 50,000 is unheard of.
The Dalit vote is typically silent. Yet the passion of this vote is difficult to miss. The motivation is visible in the rallies and in the queues outside polling booths. The overall voter turnout has dipped, thanks to the ECI's hawk-like watch over the process, yet this appears to have only helped consolidate the Dalit vote.
However, as important as the Dalit consolidation is the emergence of the Dalit-plus vote. This is not so strikingly visible and, indeed, there is enough in the poll arena to trip up the analyst. Add to a total of 117 parties 2,000-odd independent candidates and what emerges is a picture so fragmented that even the keenest mind cannot make sense of it. The crafty U.P. voter is a further problem. Stop at the village kasba (market square) and one can hear as many opinions as there are people.
A favourite answer is "sab maidan mein hain" (everybody is in the race). That is not much help, so one bombards them with questions: Who makes the best Chief Minister? What is the main issue in the election? Who ran the best government in recent times? Which are the biggies in the contest?
The answers are the same - here and elsewhere. Mayawati is the best Chief Minister; she ran the best government; she cracked down on thuggery, the main issue in the election; the contest is between "the BSP and another party". And finally, "Oh yes, there is a strong undercurrent of support for the elephant - across castes in this village, in other villages and throughout the State."
Before I return to Delhi, I cross-check my broad findings with officials in the Lucknow bureaucracy. If anything, they are more optimistic: the BSP could be heading for a majority. Yet back home on May 1, I am greeted by a profusion of exit and opinion polls, all of which forecast a hung Assembly.
The opinion and exit polls place the BSP in a broad range between 120 and 155 seats; the Samajwadi Party (S.P.) between 91 and 110 seats; and the Bharatiya Janata Party (BJP) between 90 and 125 seats. The BJP is positioned a close second to the BSP though on the election trail I detected the party only in a few urban pockets. Could I have gone wrong? Unlikely.
In 2002, the BSP won 98 seats for a vote share of 23.06 per cent. This time arithmetic and chemistry were on its side. That the BSP's Dalit vote had consolidated was not in dispute. That it was getting a fair share of votes from other social groups was also not in dispute. On the other hand, the BSP's rivals had to guard their flanks. Put all this together and you got a vote share that was way beyond anything the BSP's immediate competition seemed able to muster.
Yet what can a hack, armed with only a notepad, do in the face of overwhelming counter-evidence from reputed poll surveyors? In my final round-up for The Hindu (May 3), I scale back the BSP's seats to 180. As polling comes to an end on May 8, so does the race to call the election.
The final figures vary. One TV channel predicts a dead heat between the BSP, the S.P. and the BJP (117-127 for the BSP; 113-123 for the S.P.; and 108-118 for the BJP). A rival channel is more realistic in its assessment (152-168 for the BSP; 99-111 for the S.P.; and 80-90 for the BJP).
There are other pollsters in the market, each with their own figures. The overall seat range: the BSP between 116 and 168; the BJP between 80 and 124; and the S.P. between 96 and 123. The actual result is a vindication of the old-fashioned journalist: 206 to the BSP, 97 to the S.P. and 50 to the BJP.
Why did the opinion and exit polls go wrong and why did the ordinary journalist get closer to the actual result? Writing in Indian Express, political scientist Yogendra Yadav was honest enough to admit that "none of the polls, including ours, came close to suggesting a clear majority for the BSP".
Yadav, who anchored the opinion and exit polls on behalf of CNN-IBN, Indian Express and the Centre for the Study of Developing Societies (CSDS), noted further: "This was a clear instance of the triumph of old-style political journalism over the new-fangled number crunching. Political reporters may not have talked about a clear majority for the BSP, but they did capture the hawa in a way that the opinion and exit polls did not." Yadav cited three reasons for the failure of the exit and opinion polls: underestimation of BSP votes and overestimation of BJP votes in the exit polls; BSP voters' reluctance to speak up, which made for a response bias; and a skewed vote-seat equation in favour of the BSP.
Yadav said in conclusion: "By recognising this failure as failure, we begin to admit that election forecasting in India has a long way to go; that there is a big gap between what polls promise and are expected to deliver on the one hand and what they are capable of doing on the other. This allows us to acknowledge that the art of polling and forecasting has not seen any major methodological innovations since the path-breaking research by Prannoy Roy and the India Today-MARG team in the 1980s."
What can one say to this except that election forecasting is different from market surveys. A journalist with her ear to the ground can perhaps detect sounds not heard by the savvy surveyor who, though equipped with sophisticated tools, may not have the instincts needed to interpret political signals.