When Prannoy Roy called the Bihar elections in favour of the Bharatiya Janata Party (BJP) at around 9 am last Sunday, a young bureaucrat was busy overseeing the counting process in a district headquarter, somewhere in western Bihar. He would have been greatly amused by Roy’s declaration (and, presumably more so, by Shekhar Gupta’s supporting rationalisation) had he been watching NDTV.

He wasn’t, of course: he had just about got over counting the postal ballots, and begun counting votes registered on the Electronic Voting Machines (EVMs). Postal ballot is a special voting arrangement for people from the state but not residing there, like those employed with the armed forces, paramilitary forces, diplomats posted abroad, and so on. It accounted for less than one per cent of votes in all constituencies in the elections. Also, any pollster worth his salt would tell you that postal ballot trends, of late, have been heavily in favour of the BJP. In any case, to call out any kind of definitive trend on the basis of postal ballot is the worst kind of journalistic hara-kiri, as Roy would discover very soon.

Why did then Roy, one of India’s most experienced journalists and well-known psephologist, commit such a novice mistake?  More relevantly, where were NDTV’s disastrously off numbers coming from? Because, to be fair to Roy, he was solely reliant on his earpiece into which his producer would have been giving updates. To be sure, the people behind the screen at NDTV continued to get the trends wrong till as late as 10.15 am as the following leads versus time graph shows.

Source: NDTV’s Twitter timeline

It was only at around 10.20 that the channel finally smelled the dung (come on, how is it even fair to write anything about the Bihar elections without a cow reference). Note that this wasn’t an exit poll, based on dubious claims of fickle interviewees. These were real numbers gathered from counting centres by people trained and paid to do so.  So what went wrong? To answer that, we tried to track the journey of these numbers from the counting room to the newsroom. And to understand the passage, Roy’s rather gracious (but also conditional) apology is quite helpful. He said on air that evening:

…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 half-way. Now, this has never happened before. And we’ve asked the agency for an explanation and we’ve heard from them that they are going to write and explain what went wrong once they look into the errors.

As Roy explains, all news channels indeed rely on a News Broadcasters Association-provided common feed, which automatically updates. The agency responsible for this feed was Nielsen. (In an interesting twist, Nielsen has now claimed its numbers were right, but we will examine that claim later.)

The second part of his explanation, though, which claims all channels got it wrong, is only partially correct, as the lead-versus-time graphs of the other channels would show.

Source: Twitter timelines

First, Times Now and India Today TV, both of who claim that they got it right all along on the counting day. Times Now predictably patted itself on the back on being India’s Election Headquarter and Rajdeep from India Today was outdoing Arnab at self-congratulations while having gotten it wrong for a good part of the morning. Till about 9.30, the two graphs (of Times Now and India Today TV) are strikingly similar to NDTV.  The critical patch here, though, is the portion from around 9.30 am to 9.50 am, when Times Now and India Today TV did some serious course correction and their numbers started deviating significantly from NDTV’s. Which obviously can mean just one thing: either NDTV or Times Now/India Today TV wasn’t using Nielsen’s numbers anymore. Meanwhile IBN was closer to the actual result, so were they not using the same source?

No. They weren’t using the “same source” as claimed by Roy at all. And, they were right on the dot. “CNN-IBN had people on the ground supported by reporters from our partner networks in the ETV network; we didn’t subscribe to any agency feed,” explained Bhupendra Chaubey, Executive Editor of the channel. The following illustrations would make the difference between the modus operandi of CNN-IBN and other channels clear.

In the case of the other channels, there is an extra entity (the agency that collates the data from on-ground sources) in the information flow – making the process slower. Also since the exercise doesn’t usually involve parallel processing, this leads to a lag in trends TV channels receive. For example, if there are 243 constituencies, the practice is to collect information on every constituency after each round of counting and then collate it in a centralised place. So to predict trends on the basis of a few early rounds of counting of postal ballot is foolhardy as the percentage of votes registered on EVMs, to be counted in subsequent rounds, is much higher.  Besides, these news channels also augment their numbers manually on the basis of inputs of their reporters and stringers on ground – which means there is confusion if the two sources do not corroborate. This appears to have happened across newsrooms this time.

While Times Now’s Arnab Goswami later asserted that the channel had got it right yet again, sources in the newsroom say things weren’t that simple.  The Times Group’s newspaper The Times of India, which was getting its number from the agency C-Voter, had trends that were completely opposite to that of Times Now.  Besides, reporters from the ground were also sending out conflicting information, leading to panic in the newsroom. To Goswami’s credit though, he, unlike Roy, played the waiting game – and waited for more trends to come in before calling the elections.

Interestingly, Nielsen has claimed Times Now – along with DD News – had remained faithful to their numbers throughout the day. However, we have learnt that Times Now, post 9.30, wasn’t only relying on Nielsen’s feed. Also, according to a pollster working in another agency providing the feed, the data Nielsen provided to DD News was not the same as the NBA feed. We have sent an email to Nielsen seeking to confirm the assertion but haven’t yet received a response. The story will be updated if and when we do.

The Times Now newsroom wasn’t the only one to be marred by tension amidst contradictory reports. In the India TV studio too, anchor Rajat Sharma, who was relying on the Nielsen feed, and co-anchor Yashwant Deshmukh of CVoter, were making completely different contentions. As the morning progressed, the latter turned out to correct. Thus, Nielsen’s claim that it never got the numbers wrong doesn’t quite cut ice.

“Postal ballots have always been in favour of BJP, and since the BJP’s been doing well of late, early trends don’t really change, but this was obviously a much more complex election, which editors in Delhi failed to comprehend,” said a political journalist who didn’t want to be named.  It is curious, though, that NDTV, which prides itself on accuracy and not speed, took the bait.

A well-known psephologist Newslaundry spoke to said the embarrassment could have been avoided had Nielsen and news channels made it clear that early trends were on the basis of postal ballots. In the NDTV studio that Sunday, another veteran psephologist, had cautioned Roy about exactly the same thing. Roy, instead, chose to listen to a journalist. Disaster was imminent.

The one thing that the Bihar election coverage exposed like never before is the inanity of ex post facto gyaan dispensed in TV studios. Journalist Mukul Kesavan called it out without pulling any punches here.  And let’s not even get into the exit polls being way off the mark almost across the board. In an age where the 24-hour TV furnace needs constant timber, it is unlikely that exit polls or the dodgy dissection of the result will go away. But at least now we know it’s to provide the drama and commentary that may or may not make sense until the real contest starts, much like what Siddhu and other anchors do before an IPL match. It’s not cricket, but it’s TV.

With inputs from Abhishek Choudhary and Mahima Singh