Who actually voted in Delhi?

It is not surprising that pollsters got the Delhi Assembly elections right. The biggest problem with Indian polling is the conversion of vote share into seats. Even if the sample is representative and the data reliable, this is an intractable problem: Multiparty elections are often won by less than the statistical margin of error.

Delhi was a two-horse race with margins comfortably in favour of the Aam Aadmi Party (AAP). The Congress (INC) and Bahujan Samaj Party (BSP) played spoiler in only about eight seats out of 70 where they polled more votes than the winning margin.

Aggregated 2020 vote shares were about 53.6 per cent for AAP, 38.5 per cent for the BJP and 4.3 per cent for the INC. Compare to the 2015 Assembly elections where the AAP scored 54 per cent, the BJP 32 per cent, and the INC 9.7 per cent. Also consider the 2019 Lok Sabha elections where the BJP won all seven seats in Delhi (and 65 assembly segments) with 56.6 per cent votes, while INC scored 22.5 per cent and the AAP, 18 per cent.

The collapse of the INC vote in 2020 is most obvious. The Assembly elections are comparable. The BJP gained over 6 per cent in vote share between 2015 and 2020 though they only won five more assembly seats. The AAP retained its vote share. One glib conclusion is that the BJP ate into INC vote share — that is, erstwhile INC voters are now voting for the BJP. This is a tempting hypothesis since the drop in the INC vote share between 2015 and 2020 almost exactly equals the rise in the BJP vote share.

The second more nuanced hypothesis is that some of the lost INC vote share transferred to the AAP, while some of the erstwhile AAP vote share transferred to the BJP. We’ll need more granular data to come to any conclusion but the polling pundits suggest that this latter scenario is the case.  

There is also the statistical certainty that votes that went to the BJP in the 2019 Lok Sabha elections must have gone to the AAP in 2020. What does one make of this? The obvious answer is that voters treat national and state elections differently and rewarded the AAP for delivering at the state level.

Perhaps, but balanced against that, the BJP swept the Delhi municipal elections in 2017 presumably because they were expected to deliver on local issues. Maybe the mood has changed in the last three years and, maybe, the BJP’s blatantly communal campaign did not inspire that much confidence in voters this time around.

So far, we’re looking at a “swing voter” scenario: The postulate that the same person may have voted INC in 2015, voted BJP in 2019, and voted AAP in 2020.  There is another way to look at this. Some US pollsters have asserted that there are actually few swing voters, but even committed supporters of a given party don’t always vote. Therefore, ensuring that the committed base actually votes, is more critical than campaigning to convince swing voters.

Say, that 40 per cent of voters favour one party, and 30 per cent favour another party, while the rest of the vote is scattered across “swingers” and less favoured parties. The “also-ran” party may easily win, if its base turns out 100 per cent, while only 50 per cent of supporters of the most favoured party bother to vote.

We know 67 per cent of Delhi voted in 2015 and roughly 62 per cent in 2020. There were about 1.28 crore eligible voters in 2015 and 1.42 crore voters in 2020. Perhaps, many INC supporters did not bother to vote in 2020 and, maybe, the BJP turnout was also relatively low?  It is theoretically possible that the AAP won with over 53 per cent vote share, with only 33 per cent support, if every AAP supporter actually voted.

I don’t know whether the data would support this hypothesis but it’s worth investigating since it has a bearing on strategy. Meanwhile, the loss of yet another state could impact the BJP’s ability to manage the good and services tax system, where states have a say. This gradual reversion to a more federalised tax structure has deep implications, both good and bad.