Karnataka 2023: Here's What This Prominent Opinion Poll Needs To Explain

As soon as the Election Commission announced its notification of assembly elections in Karnataka, there was a rush of opinion polls predicting either an outright Congress win, or a hung assembly.

Not one poll suggested that the incumbent Bharatiya Janata Party (BJP) would improve its performance. Instead, the theme of the day was that the Janata Dal (Secular) of Gowda family dynast HD Kumaraswamy, the JD(S), would be the ’kingmaker’.

But are surveys like these truly representative of ground realities, or are they efforts at perception management and narrative building? The answer to this vital question, as always, lie in the numbers.

Here is the 2018 assembly election data, plus that of the general election of 2019, alongside the ABP C-Voter survey:

One, the C-Voter survey says that there will be a minor swing of votes from the BJP, the JD(S), and ‘Others’, to the Congress – and only to the Congress.

This is a little difficult to digest, since over a dozen Congress legislators defected to the BJP in 2019 along with a handful from the JD(S), and most of them won their by-elections on a BJP ticket.

Further, ground reports from South Karnataka indicate that at least a small portion of the JD(S)’s vote has shifted to the BJP for multiple reasons.

These include increasing popularity for the BJP in a region where the party has fared rather poorly in the past, acrimonious dynastic tussles within the JD(S), and some loss of the JD(S)’s distinct political identity after it entered into an alliance of desperation with the Congress in the 2019 general elections.

Thus, such surveys which don’t reflect ground realities adequately and, instead, show votes moving unidirectionally from all corners to only one recipient need to be flagged, since they are curiously reminiscent of past ones which showed a non-existent ‘AAP wave’ in states like Goa and Gujarat.

Two, even if we take these marginal swings in vote share at face value for argument’s sake, the associated swings in seat tallies are, however, frankly indigestible.

There is no realistic, representative vote-to-seat conversion formula extant which can make the BJP lose 30 seats for a mere one per cent dip in vote share; or one which ‘gifts’ the Congress 43 additional seats for a two per cent rise in its vote share; or one by which the JD(S) loses eight seats to the Congress for a negligible change in vote share.

There are additional reasons why this survey’s vote-to-seat conversion is faulty.

In 2018, the BJP won its 104 seats by a very healthy average margin of 12 per cent. Of these they lost exactly two by the vote swing projected in the C-Voter survey. Therefore, the inference is that the BJP’s vote share would need to decline by a lot more than the survey’s given 1.4 per cent, if it is to lose 30 seats.

Similarly, of the 78 seats the Congress won in 2018 with a 38 per cent vote share, its victory margin was less than two per cent in over a dozen seats.

The inference, thus, is that the Congress’s core strength is a lot smaller and shakier than such surveys would have one believe. As a result, the Congress would need a higher vote swing, directly from the BJP, if it is to get an additional 43 seats in 2023.

Three, the survey’s inherent contradictions become more apparent when we analyse their findings at the regional level. ABP C-Voter predict that the Congress will win 8-12 seats in Coastal Karnataka.

As a table below shows, this really pulls the survey at the seams:

The state’s coastal belt is a BJP bastion. The Congress did get 13 seats here in 2013, but that was when the BJP was mortally hit by an internal rebellion. And even then, readers may note that the BJP fared ‘least poorly’ here with respect to a decline in vote share from 2008. 

Five years later, the BJP not only recovered its losses, but went on to sweep the coast with record vote shares and margins.

In 2013, the BJP won 15 seats with more votes than the Congress and the JD(S) combined. Only one seat, 81-Yellapur, was a close contest, and even there, the winning Congress MLA has since defected to the BJP.

Further, this is the one region of Karnataka which most closely mimics the fillip the BJP enjoys in general elections, which means the greater chances are that the BJP would improve upon its 2018 performance here.

It is the same with the survey’s forecast for Southern Karnataka, where it gives the BJP only 1-5 seats with no increase in vote share from 2018.

Again, this goes against both past data and present trends. Even at its nadir in 2013, the BJP won three seats here; and ground reports indicate that that the BJP will do better in the region next month.

In conclusion, nothing really prevents agencies from presenting surveys like the one analysed in this piece. However, if they do so, pollsters and media houses will run the risk of having their credibility brought under scrutiny.

Consequently, as India enters a long election season which will culminate only with the general elections of May 2024, they would do well to heed some friendly professional advice – that narrative-building and perception management are not psephological tools, but political ones.