Friday, Oct 07, 2022

No Proof Required – Bihar elections: what happened and why

It was a lousy campaign. The BJP, Narendra Modi and Amit Shah ignore this reality at their peril.

bihar elections, bihar elections live, bihar elections results, bihar polls results, bihar news, #BiharResults, bihar results, india news, bihar elections photos The next time BJP officials say they lost big just because of electoral arithmetic, tell them that argument has as much credibility as the discredited notion of the BJP losing because of caste-based voting. (Illustration: Pradeep Yadav)

Practically the whole class failed in forecasting the Bihar assembly elections. And what a failure! It wasn’t as if the pollsters missed it by a few seats here and there — what is euphemistically called “sampling error”. It was not retail error, it was wholesale. To add to the mystery of what happened, one exit poll got the results almost bang on right. The Axis poll, conducted by CNN-IBN, had forecast that the Mahagathbandhan (JDU+) would win 168-183 seats — the final tally was 178. Couldn’t get it more right. However, the Axis poll was not allowed to be telecast because somebody higher up (not the editors) felt that the poll results would prove to be embarrassingly wrong for the channel.

I have followed opinion and exit polls for some 40 years in the major democracies, and for 35 years in India. I have never, ever, observed what happened at CNN-IBN — where an expensive commissioned exit poll is buried. I can understand (but don’t approve of) an opinion poll being pulled because the powers-that-be feel it would sway the election. After all, even the Supreme Court believes opinion polls sway the electorate, so why can’t TV owners? But an exit poll? There’s nobody left to sway except the odds in the satta bazaar. I just don’t get it.

If you thought it could not get murkier, it does. Today’s Chanakya, which shot to fame by correctly calling the Delhi December 2013 election for the AAP (no one else was close) — and followed it up by correctly forecasting an overwhelming NDA win last year (not alone, but one among a few), sealed the fate of the JDU+ by stating in an opinion poll that the BJP+ was well ahead across a variety of indicators, but did not give a vote share or seat prediction. For that, the hidden ace up the sleeve was a bumper crop prediction for the BJP+ — in the exit poll, a seven-point vote lead and a 155-seat victory (to 83 for the JDU+). Errors are made — it happens.

But wait. Chanakya, in an attempt to preserve its reputation, claimed after all the official results were in that a simple data error caused their mammoth failure — the computer code had inadvertently switched the seats, votes, predictions, for the BJP+ with the JDU+! India is not only a nation of snake charmers, but apparently has the highest incidence of snake-oil and bridge salesmen a la George Parker — George, an American conman, conducted several sales of property he did not own, including the Brooklyn Bridge.

Subscriber Only Stories
JK Cement’s SPSU Udaipur Launches ‘Golden Batch 2022’ In Collaboration Wi...Premium
Appendicitis in Children- A new lifestyle disorderPremium
Re-Defining The Tradition In Folk Art: An Art Educator’s PerspectivePremium
Symbiosis School of Sports Sciences (SSSS) launches undergraduate program...Premium

The best path for dealing with errors was shown by the seasoned Prannoy Roy, the pioneer in opinion and exit polls in India (along with Ashok Lahiri). He also got the exit poll grievously wrong. It happens. He didn’t apologise for the error, and nor should he have. What Roy did apologise for was jumping the competitive gun and forecasting an election victory for the BJP+ (140 seats or so) on the basis of very early, mostly postal, ballots. That was a grievous error, and Roy rightly apologised for being hasty.

Opinion and exit polls have come a long way, and in my recollection, the last time they got an opinion poll so wrong was in 1991. That opinion poll was published by the ABP’s Sunday magazine (Vir Sanghvi was the editor) and I was the psephologist associated with that poll. The two polls I had conducted previously — the 1989 general elections and the state polls that followed soon after — were pretty much spot on. That is, I forecast a Congress defeat in both. But in the 1991 opinion poll, I forecast 302 seats for Rajiv Gandhi — the Congress obtained only 239. So what went wrong? Coding error?

As I explained in the article, “Getting at the truth in opinion polls” (written in August 1991), I had managed to get every state right except for two — Bihar and Uttar Pradesh. These two states had newcomers in the form of Lalu Prasad and Mulayam Singh Yadav. My forecast for the Congress’s seats in these two states was 66. It actually got five seats. Subtract 61 seats from 302 and one obtains 241.

Advertisement

Why did I get Bihar and UP so wrong and everything else so right? I believe it was lying on the part of some voters, something likely to have happened in Bihar in 2015 as well. Is there a fraction of opinion poll voters who always lie? Absolutely. But this percentage is mostly very small and not relevant for vote or seat forecasts. In any case, the translation from votes to seats causes more errors than not correctly accounting for some voters lying.

What does cause deep errors, as likely occurred with my 1991 forecast and Bihar 2015, is when people are afraid to reveal their vote and, therefore, lie. For example, the raw vote in Bihar for the Congress in 1991 was 59 per cent. After adjusting for lying, I brought down the Congress vote share to 43 per cent. In reality, the Congress got 28 per cent.

The fear of revealing their true vote can be caused by various factors. In 1991, it was most likely the fear of upper-caste retaliation (the serf against the landlord) that drove many Mandal OBCs to hide their assertion of power. In Bihar 2015, the atmosphere against the minorities was so bad that many Muslims and Dalits felt secure in hiding their vote from the pro-establishment pollsters.

Advertisement

A repeat of the 2014 Lok Sabha vote shares meant that the JDU+ would get 45 per cent and the BJP+ 40 per cent — a five-point gap. The vote count for Bihar 2015 — the JDU+ got 41.9 and the BJP+ 34.1 per cent — shows a 7.8 percentage point gap. In 2014, 20 per cent of Muslims, who constitute 17 per cent of the Bihar population, are believed to have voted for the NDA. According to the accurate Axis poll, only 3 per cent of Muslims are believed to have voted for the NDA in Bihar in 2015. In other words, if the Muslims had just voted as in 2014, the BJP+ vote share would have been 2.9 percentage points higher (17 percentage point decline associated with 17 per cent of the population) at 37 per cent. The JDU+ would have got 2.9 percentage points less, or 39 per cent, that is, instead of the vote gap being 7.8 percentage points, it would have been only 2 percentage points, or 102 seats for the BJP+ versus 125 for the JDU+.

So the next time BJP officials say they lost big just because of electoral arithmetic, tell them that argument has as much credibility as the discredited notion of the BJP losing because of caste-based voting. The BJP lost big because of their divisive campaign, whose purpose was to instil fear among those not preordained to vote for them. It was a lousy political campaign and the BJP, Narendra Modi and Amit Shah ignore this reality at their present (and future) peril.

The writer is chairman, Oxus Investments and contributing editor, ‘The Indian Express’

First published on: 14-11-2015 at 12:00:46 am
Next Story

Book Review: No Direction Rome

Latest Comment
Post Comment
Read Comments
Advertisement
Advertisement
Advertisement
X