The Swarajya Predictor Model: At 43 Per Cent Base Vote Share, BJP Is Hitting 300 In Uttar Pradesh

Now that a long election season in Uttar Pradesh has finally drawn to a close, Swarajya presents its predictions for the state before counting starts. This forecast was generated from a proprietary, in-house predictor model using historical election data, and trends forecasted by opinion poll surveys of January and early February 2022.

This model offers probable quantification of how the Bharatiya Janata Party (BJP) is expected to fare in the 2022 assembly elections. Its principal finding is that the BJP will poll at least 43 per cent of the votes, and win around 300 seats. These projections result from a conservative construct, since the data and trends indicate that the BJP could very well poll more than 43 per cent, and thereby win more than 300 seats.

The model was designed during November to December 2021, tested and populated in early January 2022, and the final forecast runs were made on 8 February 2022, when Times Now released their last survey for Uttar Pradesh.

Methodology and inputs:

The UP-BJP predictor model is a spreadsheet which converts vote shares to seats by using correlations. These conversions are made on curves generated by Swarajya’s detailed analysis of historical election data (basically plotting vote shares versus seats and trying to identify trends). This exercise revealed two key points.

One, we established numerically that outcomes are dependent on three main metrics – vote share, the index of opposition unity (IOU), and the vote differential, or margin, between the BJP and the main opposition alliance. For the past few elections, this has been a coalition led by the Samajwadi Party (SP).

Two, very interestingly, the correlation curves showed a distinct difference in results when the IOU was high, and when it was not. They also showed that this divergence begins to manifest itself once the BJP’s vote share crosses above 30 per cent.

Consequently, two primary curves were used when conducting the final forecast runs – a low IOU curve (colour-coded saffron in the chart), and a high IOU curve (colour-coded black).

The next step was to determine the extent of opposition unity expected in 2022. Both the latest opinion polls and ground reports indicated that there would be a significant consolidation of the Muslim vote under the SP banner, along with around half of the Jat votes in those seats where an SP-ally, the Rashtriya Lok Dal, has a presence, and, that the SP’s core Yadav vote base would remain loyal to the SP.

In 2022, the IOU factor also covers the fairly fierce rebelliousness within the BJP, which manifested itself in the disgruntled departure of a number of important personas from the BJP, to mainly the SP, because they were denied tickets. This infighting, along with the anti-incumbency normally expected at the end of a term, appeared to have been captured by most of the surveys, which projected the BJP’s vote share to drop below their 2017 figure of 41 per cent, and like a stone from their 2019 high of 51 per cent.

But, and this is a crucial point, none of the surveys reported a shift in voting preferences from other parties to the BJP. This was where Swarajya drew a limiting line to the value of these surveys, because a golden rule in current UP politics is that, if any one party or alliance seeks to organize a consolidation of the identity vote, rest assured that there will be a potent counter-consolidation towards the BJP. We saw it in 2014, in 2017, and again in 2019, when the anti-BJP IOU was the strongest in decades, with the SP, the RLD and the BSP surreally coming together.

How could it be that the Muslim vote shifts from the BSP and the Congress to the SP, but the surveys don’t show a resultant shift of votes to the BJP – especially when this is precisely what happened in 2014, 2017 and 2019? What is the value of a survey which reports a consolidation, but not an inevitable counter-consolidation which took place thrice in the past eight years in Uttar Pradesh?

Further, while the consolidation under the SP is palpable, and the contest is strongly bipolar between the SP and the BJP, the IOU has not gone up commensurately. This is because the vote shares of the Congress, the BSP, and ‘Others’ still total to near 20 per cent.

If the IOU were to be materially up, it would have to mimic the 2019 elections, when the combined vote shares of the two largest groupings totaled around 90 per cent. Otherwise, the determinant metric for outcomes would be vote differential, with the winning party needing less votes to come first. But that is not the case in 2022.

Under these circumstances, Swarajya concluded that the correct correlation, to make a more representative conversion from vote shares to seats for the BJP, would be less than the Lower IOU curve, but not by too much.

Third, the electoral data of the 2017 assembly elections and the 2019 general elections were rigorously analyzed seat by seat (called segments when it is a Lok Sabha poll). Multiple scenarios were crafted at the seat level, to see what outcomes were probable, depending upon how many votes shifted from one alliance to another.

Finally, two forecast runs were made. The first is at 39 per cent, which is the midpoint of three Times Now surveys conducted between January and February 2022. It is called ’VGN1’ and is the pink square on the curves.

The second is at 43 per cent, which is the base vote share Swarajya estimates for the BJP. It is called ‘VGN2’ and is denoted by a blue square on the charts.

In addition, three surveys by Times Now, two forecasts by Republic, and one by Jan Ki Baat are also plotted on the charts for comparison.

Results of the predictor model forecast runs:

To explain: the BJP’s vote share (historical, surveyed and predicted) is on the horizontal X-axis. Seats are plotted on the vertical Y-axis. The upper saffron curve is the Lower IOU one. The black curve is the Higher IOU one. Dummy data had to be added above 60 per cent vote share to make the curves bend under 403 seats, but they are only for visual and aesthetic appeal, and make no impact on predictions.

Readers may note how the results of the 2014 and 2017 elections conform to the Lower IOU curve (dark blue squares), as do the 2019 results to the Higher IOU curve (maroon square). The yellow squares are the Times Now surveys, the white ones are Republic’s, and the green one is Jan Ki Baat’s.

Admittedly, there is a case for classifying the results of 1993 and 1996 under the Higher IOU curve data set, as there is for putting the 1991 results in the Lower IOU column. But this is not necessary, since the divergence point, around 31 to 33 per cent, is well below the lowest predicted vote share for the BJP by anyone in 2022, so it does not affect the forecasts.

Analysis:

A zoom of the results is placed below:

We see that the Times Now seat forecasts (Yellow squares) for the BJP at 38, 39, and 40 per cent vote share lie closer to the Higher IOU curve; the seat range is 222 to 241 (midpoints). These are well below the VGN1 run (pink) at 39 per cent which predicts 260 seats.

The Jan Ki Baat poll of 28 January 2022 too, effectively represents a ground condition of very high IOU. These surveys aren’t wrong technically, since they all concur with Swarajya’s original premise that the contest in UP is going to be strongly bipolar. The only difference is that in 2022, bipolarity will be offset by the votes retained by the BSP, the Congress, and ‘Others’. In numerical terms, it means that the BJP will retain the edge, even if with a reduced margin.

The Republic survey (white square, ‘Republic-1’) gives the BJP 262 seats at 41 per cent vote share. This is only two percentage points more than the VGN1 run, which forecasts 260 seats at 39 per cent; since both fall within each other’s error margins, they may be treated as similar.

However, the big question is: what is the base vote share which the BJP may get in 2022? This is the key, because seats are calculated from vote shares, and not the other way around.

Interestingly, Republic too, constructed a scenario based on their primary survey, which predicts that the BJP would win 312 seats if its vote share went up to 43 per cent (white square, ‘Republic-2’). This is very close to the VGN2 run results.

Therefore, in conclusion, our base forecast is: there is a very high probability that the BJP will poll at least 43 per cent of the votes, and win 300 seats.

All electoral data from Election Commission of India website.