Wednesday, December 4, 2013

Exit polls should be on entertainment channels!

I was a little surprised to hear over TV last night that an exit poll involves going to the homes of respondents and asking them who they voted for. I was under the impression that voters leaving polling booths were intercepted immediately after they had voted and made to vote again in a dummy exit-poll-booth. Done this way, it would be a fairly accurate process. But apparently that’s not the case. Explaining this, Rajdeep Sardesai declared honestly that his channel would only be able to put out the final exit poll numbers by 8 pm Thursday (since visiting voter homes takes time). The other channels did not even bother to issue this clarification!

This methodology of exit polls explains why it is practically impossible to forecast the final outcomes accurately. Going to voter homes is difficult and costly. Can surveyors actually go into the homes of the rural folks, the urban poor, the SCs and STs, and in the case of Chhatisgarh, the Naxal dominated areas? I say “practically impossible”, because it’s not a technical impossibility, but only a commercial one. No TV channel expends enough money to do surveys accurately. This is why almost all of them go so wrong.

Take the 2009 general elections, when exit polls were broadcast on TV on 13th May. Polls were conducted by CNN-IBN, Times Now, Headlines Today, India TV-C-voter, NewsX, Star News-Nielson, Zee Business, UTVI, News24 and NDTV. The average of these polls for the NDA was about 200 and for the UPA, about the same. What the final result was is well known, in fact, much better than anyone expected giving the UPA a conclusive win (source: tweet from PM’s media advisor, Pankaj Pachauri). If the exit polls were conducted correctly, why did they not pick up the “wave” for the Congress?

In fact, some have argued that given the Congress’s rural/poor support base, exit polls typically give it 20-25% lesser seats than it finally gets. Likewise, the more urban focused NDA gets a higher score by the same quantum. Such poll errors are not limited to the Congress and BJP alone. They affect all parties that are in rural-heavy states. In 2012 in UP for example, DNA ( reported, “Except for two, all the exit polls predicted a hung assembly in Uttar Pradesh with Samajwadi Party on top but the results showed that the Mulayam Singh Yadav-led party romped home with a thumping majority”. Such is the reputation of these polls that DNA quotes then Election Commission Chief SY Qureshi in the same article as tweeting “opinion and exit polls should at best be on entertainment channels”!

The serious flaws in these exit polls is visible in the forecasts made for the AAP in Delhi this time, with one poll (ORG-India Today) giving the party just 6 seats, and another (Chanakya) giving it 31. All these polls claim to have statistical errors of around 3% or so. If that is so, then how come there is such a wide variance?

And even if one goes by the 3% error, that can wreak havoc on the forecasts. The biggest voting % difference between the BJP and Congress, as predicted by the exit polls, is in MP and Rajasthan, but its only 5%. A 3% correction one way for the Congress, and the other way for the BJP, can hand over victory to the Congress in both these states. The gap between the two parties in Chhatisgarh is just 1%, and no one should have any doubts about the uncertainty of the poll results in this state. Ditto with Delhi, where the vote shares are too close to allow making any prediction.

Then there is this other problem about converting vote shares to seat shares, an exercise which involves estimating the “concentration” of votes. A 1% lead for the BJP over the Congress in Chhatisgarh for example, if uniformly spread across the state, is enough to give the party a clean sweep. On the other hand, the 5% in MP and Rajasthan may only give it a marginal lead. How then do these media outlets speak with such authority on the basis of these polls? A news channel yesterday, with a known rightist bias, went so far as to make its headlines sound as if they were based on final outcomes, dropping even the caveat of them being based merely on exit polls! Exit polls may be good content strategy for TV channels, but for politicians, psephologists, political analysts and media pundits to engage in 3-4 hour long, highly intellectualized debates is plain stupidity!

All this is not to say that the Congress is not doing poorly in these four major states. It is. In MP, the Congress has no chance, with the state having been completely saffronized (like Gujarat) over the last 10 years (so saffronised that people don’t mind the huge corruption, law & order and other problems). In Rajasthan as well, known for its “flip-flop” pattern of electing governments, the Congress will likely lose. What is important to see however is what happens in Delhi and Chhatisgarh. For the Congress, wresting Chhatisgarh will be a big morale booster. With respect to Delhi, it will be an acceptable outcome if either it emerges at the top or if the BJP doesn’t come at the top. Half this battle is about the Congress winning; the other half about the BJP losing!

Honestly, the Congress should not worry too much about these poll results, no matter how they go. In 2003, the same pattern had been observed – the BJP won MP, Chhatisgarh and Rajasthan (but not Delhi). Yet they were trumped in the General elections of the 2004. This is not an exact analogy, since the central government then was the BJP’s and this time it is the Congress’s. But still, six months is a long time in Indian politics, and the Congress can still make a recovery. A few bold and innovative moves and the sentiment could shift. The sentiment against the Congress will also drop, if it is seen as having been punished already.

One last point. The BJP knows not to trust these exit polls. That is why its spokespeople were not willing to say if they would attribute the results to Narendra Modi. What if they lose Chhatisgarh, and don’t emerge as winners in Delhi?! Likewise, the Congress was unwilling to credit Rahul Gandhi with the outcomes of these polls. Quite clearly, the political parties know the polls are unreliable!

The real truth is that what former CEC SY Qureshi said about the entertainment nature of exit and opinion polls. Rather than featuring on news channels, they should be carried on entertainment channels. But then, if this principle were followed, most of the content of the news channels would end up being on entertainment channels!

No comments:

Post a Comment