Decoding Bengal SIR data: Of 123 margins, 49 in closer focus

Special Intensive Revision (SIR) was definitely the biggest talking point of the West Bengal election. The final result was nothing short of a political earthquake, with BJP winning 207 seats, surpassing most expectations, the mighty TMC humbled to 80, and other parties reduced to a mere footnote.

The anti-incumbency was so acute that it probably blew away much of the discomfort and anger around SIR, which TMC had wanted to leverage, many pundits believe. Some have previously argued that the deletions were done in a politically motivated manner to tilt the odds in BJP’s favour, especially pointing to the “logical discrepancy” factor introduced late in the SIR process.





Here, we decode the data based on different thresholds.

To understand the possible effect of SIR, the most obvious starting point is to check the seats where voter deletions were larger than the margin by which the seat was won.

This analysis uses constituency-wise deletion calculations by Sabar Institute, mapped with the 2026 result-margin data. Net deletion here means deletions for reasons other than death. The net-deletion table covers 294 Assembly constituencies and adds up to 66,62,010 deletions.

And this is where the first big number comes in.

In 123 of Bengal’s 293 declared seats, net deletion was larger than the winning margin. BJP won 83 of these seats, TMC 38, and Congress 2. BJP’s tally of 83 out of 123 is roughly two-thirds, broadly in the same zone as its overall dominance in the election.

There is a second, sharper layer. Supplementary deletion, from the supplementary/adjudication layer, crossed the victory margin in 49 seats. This refers to under-adjudication voters whose names were finally deleted. At final count, 27.16 lakh of the roughly 60 lakh under-adjudication names were deleted. When mapped constituency-wise, BJP won 26 of these 49 seats, TMC 21, and Congress 2.

For clarity, we are using the word stress in a narrow arithmetical sense. A seat is counted as a deletion-stress seat when the number of deletions is larger than the victory margin. So if a constituency was won by 5,000 votes and net deletion was 12,000, that seat enters the stress list because deletion was more than twice the margin.

This does not mean the result would have changed. It only means the deletion count was large enough to be electorally material when compared with the margin.

All these 49 supplementary-deletion stress seats were already inside the larger 123-seat net-deletion stress universe.

These numbers matter because they stop the debate from becoming vague. In 65 seats, net deletion was not just larger than the margin; it was more than double the margin. In 20 seats, it was more than five times the margin. Even by the narrower supplementary-deletion test, 23 seats crossed the 2x mark and 10 seats crossed the 5x mark.

That is not a clerical footnote.

Take Rajarhat New Town. BJP won by just 316 votes. Net deletion there was 50,274. Supplementary deletion alone was 24,132. By the supplementary test, deletion was more than 76 times the margin; by the net-deletion test, it was more than 159 times the margin.

In Satgachhia, BJP’s margin was 401, while net deletion was 17,783 and supplementary deletion was 8,785. In Kashipur-Belgachhia, BJP won by 1,651, while net deletion stood at 39,278.

These numbers do not say the result would have changed. They say the deletion figure was far too large to ignore.

But here is where the story becomes more interesting than a partisan talking point. High deletion stress did not always mean a BJP win.

Look at Samserganj. TMC won by 7,587 votes. Congress came second. Net deletion was 83,662. Supplementary deletion alone was 74,775, almost 10 times the margin. If deletion stress automatically translated into BJP benefit, Samserganj would not look like this.

The sharper political question, therefore, is not merely where deletions crossed margins. It is whether these seats also saw sharp vote-share churn. Did BJP rise sharply? Did TMC fall sharply? Did the deletion-margin map overlap with the political swing map?

The clean benchmark comes from the 129 seats that moved directly from TMC in 2021 to BJP in 2026. In these seats, BJP’s average vote-share gain was 10.63 percentage points on the adjusted basis, while TMC’s average fall was 8.90 points. The average two-way churn was 19.53 points.

The strongest churn signal comes from the overlap of the top 50 BJP-gain seats and the top 50 TMC-drop seats, both measured in percentage-point terms. Thirty-five constituencies appear in both lists. In these, BJP gained 15.93 points on average, while TMC fell 12.35 points. That is the real churn zone, where BJP’s rise and TMC’s fall happened together.

This matters because the 123 deletion-margin seats are not one political type.

Some are BJP-conversion seats, where deletion stress and churn moved together. Some are TMC-erosion seats where the beneficiary was not always BJP. Some are arithmetic-only stress seats, where deletion crossed the margin but the vote-share movement was not dramatic.

Bhabanipur is a good example of the first type. It is not part of the 49-seat supplementary-deletion stress list. Supplementary deletion there was smaller than the margin. But broader net deletion was 2.66 times the margin. At the same time, BJP’s vote share rose 17.86 percentage points, while TMC’s fell 15.52 points.

So Bhabanipur is not an adjudication-deletion story. It is a net-deletion-plus-churn story.

Jadavpur tells a similar story. Net deletion was 1.25 times the margin. Supplementary deletion was small. But BJP’s vote share jumped 21.29 points, while TMC fell 11.58 points. Jadavpur is not in the supplementary core, but it is very much part of the broader deletion-margin and churn map.

Nandigram, however, is different. It technically falls into the net-deletion stress zone, but only just. BJP won by 9,665. Net deletion was 9,891, just 226 more than the margin. Supplementary deletion did not cross the margin. BJP’s vote-share gain was only 1.88 points. TMC’s fall was 1.09 points.

Inside the 49 supplementary-stress seats too, there are different political types.

Some are clear BJP-surge seats. Jangipur, for instance, shows supplementary deletion more than three times the margin, BJP gaining 20.73 points, and TMC falling 30.88 points. Rajarhat New Town, Kashipur-Belgachhia, Manikchak and Monteswar also fall into this stronger bucket: deletion crossed the margin, BJP rose sharply, and TMC fell sharply.


But another set tells a different story. Farakka, Raninagar, Lalgola, Raghunathganj, Mothabari, Suti and Samserganj show TMC erosion under high deletion stress, but the beneficiary was not always BJP. In parts of Murshidabad and Malda, Congress and local contest structure mattered. A falling TMC vote did not always become a rising BJP vote.

Then there is a third bucket. Raina, Pandabeswar and Jangipara had high deletion-to-margin ratios, but weak BJP-TMC churn. Pandabeswar is especially useful as a warning. BJP won and supplementary deletion was more than four times the margin, but TMC’s vote share actually rose by 0.29 points. That cannot be called an anti-TMC churn seat.

In a smaller set, especially in Kolkata and urban-adjacent belts, margin stress overlapped with a sharp BJP rise and a sharp TMC fall.

Hence, there is no definitive way to say, seats worst affected by SIR disproportionately helped a certain political party. It broadly mirrored the prevailing ground situation at that given constituency when looked at a micro level. SIR had “special” in its name, but the result did not stick out as a statistical sore thumb. What stood out was the margin math it left behind.