Reading the tea leaves

A quick note on exit polls

This has been doing the rounds of social media over the last couple of days:

It is a compilation of exit polls from the 2021 West Bengal elections. The actual outcome was TMC - 215, BJP - 77, Left+Cong - 0.

It is not an exception — among several other instances, last year’s Karnataka assembly election showed a similar gulf between what pollsters predicted (a hung assembly) and what the voter decided (a decisive mandate for the Congress).

The point sought to be made by those posting the table above is, what do pollsters know? That is reductive — it is as foolish to dismiss exit polls altogether as it is to believe them in toto.

Pollsters are generally good at reading trends, at capturing the general direction an election will take. This is particularly true if there is a pan-India wave in favour of one party or the other.

Such waves manifest when there is one over-riding national narrative that subsumes all others, including local grievances. 2019 is an example (as was 2014). In the run-up to that election, the attention of many (including myself) was focussed on the Modi regime’s many disastrous policies, chief among them being the 8 November 2016 demonetisation and the badly thought-out GST regime introduced less than a year later, on 1 July 2017.

I, among others, assumed that the impact of the twin blows would have sunk in among the voting public, and that this would impact the outcome of the 2019 polls, over-riding the 28 September 2016 surgical strike and the BJP’s campaign on a muscularly nationalistic platform with an Islamophobic subtext.

I was wrong, as were others, in gauging the impact of the surgical strike in the minds of the voters, and not seeing that the resulting (misplaced) nationalist fervour, the appeal of the dushman ke ghar mein ghuske maara campaign theme, would overwhelm more quotidian concerns.

In the event, the surgical strike, such as it was, triggered a wave and the pollsters picked up on it, predicting a big win for the BJP. Pradeep Gupta’s Axis My India was particularly spot on — the actual seats won by each party fell squarely at the mid-point of Gupta’s upper- and lower-end estimates.

So how do pollsters get some elections right and some wrong? An involved explanation would need to parse the quality and size of sampling, the demographics sampled, and numerous other factors. A quick explanation is, pollsters spot a wave when it washes over the electorate; they are not however as good at calculating the impact of sub-surface groundswells.

A perfect example would be the 2004 election, which polls got disastrously wrong — because while the BJP’s India Shining campaign was in your face and unmissable, the groundswell of discontent over a supposedly sparkling economy that had left the bottom of the pyramid untouched was beneath the surface and hence unquantifiable. In other words, 2004 was a bread-and-butter election — and such polls are far harder to read through random sampling.

2024 is one of those hard to read elections. There is no pan-India issue — this is what pollsters have begun to largely agree on, particularly after the fifth phase (What took them so long?). Against that, there is a visceral discontent among the electorate, across pretty much every state.

In such situations — particularly when taken together with a lacklustre ruling party campaign and a cohesive Opposition campaign led by local leaders as opposed to one central figure — vote share moves in two directions simultaneously: a decline for the incumbent, and an increase for the Opposition.

Whether these two opposing movements will be big enough for one to supersede the other is what the result of the 2024 polls boils down to — and I suspect that is not the sort of read pollsters, operating as they do in broad-brush binaries, can make with any great deal of accuracy. (I don’t have a quantifiable answer for that question either, by the way).

Beginning 6.30 PM, the exit polls will swamp TV screens (I don’t own one) and social media (which I’ll try to avoid for the duration). Follow along, if that is your thing — but keep the above in mind when deciding how to read what you see on your screen.

PostScript: When I said there is no national issue, I was speaking of how pollsters view elections.

• Unemployment, at its highest in decades, is a national issue.

• Crippling poverty, that forces 80 crore of our people to depend on free (and inadequate) rations is a national issue.

• The increased economic stresses on the middle class is, coupled with uncontrolled price rise, a national issue.

• Rampant corruption, most recently brought into stark relief by the release of the “unconstitutional and illegal” electoral bonds scheme, is a national issue.

• The ever-widening gulf between the top one percentile and the bottom 99 percentile is a national issue.

• The rising tensions between the Centre and states, particularly those south of the Vindhyas, is a national issue.

• The ripping apart of our societal fabric, the mainstreaming of hate, is a national issue.

• The collapse of the various pillars that support our democracy — the courts, the police, the media — is a national issue.

• A state collapsing into a civil war that has gone unchecked for 13 months and counting is a national issue.

• The deliberate degradation of the environment, and the rising impact of climate change, is a national issue.

• The gradual loss of friends and once-trusted neighbours along our borders is a national issue, particularly when coupled with China’s increasing belligerence.

These are all national issues. And the fact that we still talk of this as an election without any pan-national issue is the essential tragedy of our times.