The election season has begun. So has the season for opinion polls. We have already had at least three nationwide opinion polls, and all of them forecast a reduction in the tally for the BJP and its allies. But none put the BJP below 200 seats. This reinforces the political conclusion that the BJP is in the race to form the next government, with Narendra Modi as the PM if it does a little better — or without him if the tally is on the lower side.
I believe these polls are wrong. There is no way the BJP can touch the 200-seat mark necessary for it to form the government in 2019.
I do not think the pollsters are deliberating misleading us. I have been in this business of election forecasting. I know that pollsters hate to get their forecast wrong. And they pay big professional price when they do. The reason for a possible failure is technical.
To understand this, let us turn to Prannoy Roy’s distinction between Type-1 error and Type-2 error in opinion polls. Type-1 is the most common error made by pollsters. They are risk averse, and hence play it safe. They get the winner right, but underestimate the extent of victory. Typical recent examples would be the AAP’s or the BJP’s sweep in Delhi or Uttar Pradesh, respectively. Most pollsters predicted the victor but almost no one got the extent of the victory right.
Type-2 error is where one gets the victor wrong or predicts a big sweep that does not happen. This is every pollster’s nightmare, and often happens when the pollster takes a risk and goes strictly by what the numbers indicate. My own misadventure in projecting a clear defeat for the BJP in Gujarat in 2017 illustrates Type-2 error. I saw two credible polls showing a decline for the BJP in the last four weeks and simply projected it onto the final outcome. Pollsters who predicted a clean sweep for the Congress in Rajasthan based on correct pre-poll data made the same error.
Pollsters prefer to make Type-1 error and avoid Type-2 error as much as possible. Yet there is a special case where a series of Type-1 errors can lead them into making Type-2 error. A tendency to play it safe can land the pollsters into the most risky situation of getting the victor and the loser wrong. This is what is likely to happen with the pollsters in 2019.
The same thing happened in 2004 as well. Atal Bihari Vajpayee had completed a seemingly successful five-year tenure. His popularity ratings were way above that of any opposition leader. The economy was allegedly “shining”. And the BJP had won the three state assembly elections in Rajasthan, Madhya Pradesh and Chhattisgarh. When Vajpayee advanced the general elections by six months, every poll expected the NDA to come back with over 300 seats. Pre-election polls closer to the election date gave an average of 271 seats to the NDA. The average of all the exit polls was 255 seats for Atal-led alliance. Everything pointed to another term for Atalji. When the results came, the NDA won 187 seats. With the BJP’s own tally at a paltry 145, he was nowhere in the race.
What went wrong in 2004? Basically, all the pollsters played it safe in estimating the loss of seats for the BJP. But since they made the same error in a number of states, the accumulation of small errors led to a big blunder. Normally, Lok Sabha elections are safer to predict, as Type-1 errors at the state levels tend to cancel each other out at the national level. You over-estimate a party in one state but under-estimate it in another state. But in 2004, all pollsters played it safe by understating the BJP’s losses in most of the states, especially in states where the party was in direct contest with the Congress. Instead of cancelling one another, forecasting errors stacked up. Thus, a typical Type-1 error resulted in a Type-2 error, one of the most embarrassing polling blunders of recent times.
Election forecasting for 2019 threatens to be a replica. The normal logic of errors cancelling one another out may hold good for the 317 seats in the area outside the Hindi belt. For example, an underestimation of the extent of the DMK alliance victory in Tamil Nadu may be balanced by an under-estimation of the YSR Congress.
Similarly, over-estimation of the Congress in Gujarat may be balanced by its under-estimation in Karnataka.
But in the case of the 226 seats in the Hindi belt, all the errors are likely to be loaded in one direction. The BJP had won 191 seats (202 if you include allies) here in 2014. It can only shed seats everywhere. Any forecaster who plays it safe will understate the BJP’s losses in every state. So, if the poll shows 12 seats for the BJP in Uttar Pradesh, the pollster is likely to jack it up to 20. If it shows 10 seats for the Congress in Chhattisgarh, a conservative pollster would slash it to seven. Since all these errors would favour the BJP, it would add up to anything between 30 to 50 seats of over-estimation for the ruling party.
I am not recommending that you stop following the opinion polls and the forecast this election. On the contrary, I have always argued that a half-decent survey is more informative than drawing room or newsroom gossip. I would just suggest that you focus on the real and interesting trends, instead of just looking at the forecast of the number of seats for each party. The more interesting and useful information in an election survey is about percentage of votes for each party, its distribution across social segments, and the evaluation of the government and its leaders. If you do look at the number of seats predicted for different parties, just mentally chop off 30-50 seats from the BJP’s tally.