Most television channels, including NDTV, were giving NDA a thumbs up, when it was the Grand Alliance that was winning the battle on the ground. Viveat Susan Pinto & Jency Jacob report
Barring CNN-IBN, no other television channel seemed to have got their numbers right on November 8, the day when results of the five-phase Bihar polls were declared.
Most television channels, including Prannoy Roy's NDTV, were giving the National Democratic Alliance a thumbs up, when it was the Grand Alliance of the JD-U, Rashtriya Janata Dal and the Congress that was winning the battle on the ground. The final tally: NDA 58; Grand Alliance 178.
What went wrong?
Here's what people in the know, who were part of counting day coverage, on Sunday said: Most channels subscribe to data provided by market research agency Nielsen.
The latter typically has people stationed at every counting booth, collecting leads after every round is verified by the returning officer and put up a board. Leads on the board are then conveyed to a central call centre, which is then given to all TV channels. Channels are connected to Nielsen's servers and have individual software to process the raw data received from the call centre, which is then quickly pushed on air.
Why CNN-IBN got it right is because the channel did not depend entirely on Nielsen numbers.
Radhakrishnan Nair, managing editor, CNN-IBN, said, "We have an extensive network of reporters and stringers from our regional channel ETV Bihar, who were deployed to give us minute-by-minute updates of what was happening on the ground. This helped us in putting out accurate numbers."
CNN-IBN was also the earliest in calling the elections accurately, precisely, at 10.03 am on Sunday morning. NDTV, in contrast, had to hastily re-calculate its numbers following an early mess-up in declaring the results in NDA's favour.
A producer with a national news channel said, "There was an error that crept in as the initial rounds which NDTV processed was based on postal ballots. This gave the wrong picture of the NDA leading, when that wasn't the case. Their speed proved to be their undoing."
On Sunday, Roy issued a clarification saying: "On every counting day, like today, all news channels get data from one agency. Again, a very globally respected agency. This morning, the first data that came in to all news channels was completely wrong. Our trend analysis was based on this data like it has been for 35 years -- it's never been wrong so far."
"We showed the BJP ahead -- that's what the data showed. The data, unfortunately, turned out to be incorrect. All news channels had to change their data halfway. Now, this has never happened before. And we have asked the agency for an explanation and we have heard from them that they are going to write and explain what went wrong once they look into the errors," he added.
Nielsen was not immediately available for comment.
Interestingly, the bad run for news channels began a few days before counting day, with most of them getting their predictions wrong.
But Yogendra Yadav, a psephologist and former member of the Aam Aadmi Party, who had predicted a day earlier that the Grand Alliance was likely to win the polls, said, "My judgement was based on electoral common sense and my experience as a pollster. I am of the view that expectations from opinion polls should be toned down. There has to be strong transparency norms of what goes into exit polls. There must also be a dispute redressal mechanism, in case there are inaccuracies in a poll and, last but not the least, more R&D must go into predicting polls."
But is that easy? As Today's Chanakya, the agency that had called the 2014 general elections accurately, says, that may not be the case.
The agency said in a statement on Monday: "We found that a simple computer template coding marking the alliances got interchanged at our end. Due to this, our seat numbers remained the same, but respective alliances got interchanged."
Pradeep Gupta, CMD, Axis Ad Print Media, the agency that did manage to call the Bihar polls accurately, said, "It is difficult to predict polls. But we can all keep trying and doing our best. In our case, we worked on the sampling, methodology, ensured we mapped it accurately to the demographics and socio-economic status of the people in the state, plus had a robust questionnaire. I think all of this worked in our favour."