How West Bengal’s “historic” 91.91% is an artefact of voter roll deletions — and why 10.51 lakh fewer people actually voted in 2026 than in 2021
Across television studios and press conferences on April 23, 2026, one number was repeated like a mantra: 91.91%. Chief Election Commissioner Gyanesh Kumar declared it the “highest ever percentage of polling in West Bengal since Independence.” The Trinamool Congress called it a people’s verdict. The BJP read it as anti-incumbency. And yet, beneath this celebrated figure lies a quieter, more inconvenient truth — one that the raw data makes impossible to ignore.
Fewer people voted in West Bengal’s Phase 1 this year than in 2021.
THE NUMBERS THAT DON’T ADD UP
Across the 15 Phase 1 districts, an estimated 84.16 lakh people cast their votes on April 23, 2026. In the same districts in the 2021 Assembly elections, approximately 94.67 lakh people had voted. That is a decline of 10.51 lakh actual votes — yet the reported turnout jumped by over 11 percentage points.
How is that possible? The answer lies not in the ballot box, but in the voter roll.
Before the 2026 elections, the Election Commission of India conducted a Special Intensive Revision (SIR) of electoral rolls — the first such exercise in West Bengal since 2002, a gap of over two decades. In the process, over 91 lakh names were deleted statewide. Across the 15 Phase 1 districts alone, 27.07 lakh voters — approximately 22.5% of the pre-SIR electorate — were removed from the rolls. The stated reasons: death, migration, duplication, and untraceability.
The roll shrank. The number of people who actually showed up shrank too. But because the denominator fell faster than the numerator, the percentage soared.
THE ARITHMETIC OF ILLUSION
If we apply the actual votes cast in 2026 against the pre-SIR electorate — the same baseline used for 2021 — the Phase 1 turnout collapses to just 69.81%. That is nearly 9 percentage points lower than 2021’s 78.52%, not 11 points higher. Pre-SIR electorate (Phase 1) 120.57 lakh Post-SIR electorate (Phase 1) 93.50 lakh (27.07L deleted, 22.5%) Votes actually cast in 2021 94.67 lakh Votes actually cast in 2026 84.16 lakh (▼ 10.51 lakh) Reported 2021 turnout 78.52% Reported 2026 turnout 90.02%(looks like a surge) Counterfactual 2026 % (on pre-SIR roll)69.81%(the true picture)Inflation from roll shrinkage+20.21 pp Of the +11.50 pp “surge” → roll shrinkage +20.21 pp (176%) Of the +11.50 pp “surge” → genuine votes −8.71 pp (−76%)
The entire reported surge — and then some — is a mathematical artefact. Genuine voter mobilisation, measured by actual bodies at booths, not only failed to increase — it fell by 8.71 percentage points on a comparable base.
THE DISTRICTS TELL THE STORY
District SIR Del% 2021 Turnout 2026 Turnout Δ Actual Votes Inflation Effect Murshidabad 40.5% 79.72% 92.93% ▼ 4.51L +37.61% Malda 32.6% 76.22% 89.56% ▼ 2.23L +29.18% Uttar Dinajpur 35.8% 77.32% 85.00% ▼ 2.31L +30.46% Cooch Behar 25.7% 79.26% 94.54% ▼ 0.85L +24.29% Birbhum 17.2% 80.22% 93.70% ▼ 0.23L +16.16% Jalpaiguri 15.0% 78.92% 93.23% ▲ 0.03L +13.98% Bankura 12.8% 79.88% 89.91% ▼ 0.13L +11.50% Paschim Medinipur 10.1% 81.07% 90.70% ▲ 0.04L +9.17% Dakshin Dinajpur 17.4% 81.49% 94.85% ▼ 0.14L +16.50% Purulia 10.4% 75.97% 88.00% ▲ 0.19L +9.19% Alipurduar 12.6% 77.53% 86.00% ▼ 0.12L +10.85% Jhargram 14.3% 81.04% 90.53% ▼ 0.14L +12.93% Darjeeling 10.3% 76.54% 86.49% ▲ 0.06L +8.89% Paschim Bardhaman 8.3% 75.44% 83.00% ▲ 0.04L +6.92% Kalimpong 20.0% 74.07% 81.98% ▼ 0.19L +16.40% Phase 1 Total22.5%78.52%90.02%▼ 10.51L+20.21%
Inflation Effect = Reported 2026 % minus Counterfactual % (2026 votes on pre-SIR roll)
Of the 15 Phase 1 districts, 10 recorded fewer actual votes in 2026 than in 2021. Murshidabad alone shed 4.51 lakh real votes. The district with the highest deletion rate produced the highest inflation. Paschim Bardhaman, with the lowest deletion rate at 8.3%, has the most honest headline — and accordingly, the lowest reported turnout in the phase.
WHAT THE STATISTICS CONFIRM
A Pearson correlation of deletion percentage against reported 2026 turnout yields r = +0.18 — weak and statistically insignificant (t = 0.66, df = 13, p > 0.05). Against 2021 turnout, r is virtually zero. Deletion intensity does not predict voter behaviour. It predicts the size of the arithmetic distortion.
A NEW KIND OF ELECTORAL ARITHMETIC
This is not an allegation of fraud. The SIR process is legally mandated, backed by Supreme Court directions, and conducted under ECI supervision. Many of the deleted voters may well have been genuinely ineligible.
But accuracy and enfranchisement are not the same thing. When you shrink the denominator by 22.5% and real votes fall by only 11%, the percentage rises automatically. It is not democracy at work. It is arithmetic.
A turnout percentage is only meaningful if its denominator is stable and credible. When the roll changes by 22.5% in a single revision, the percentage becomes a ratio of two moving targets — and the political narratives built upon it deserve to be measured against the one number that does not lie: how many people actually stood in line and voted.
In Phase 1 of the 2026 West Bengal Assembly elections, that number was 10.51 lakh fewer than in 2021.
Methodology: Pre-SIR electorate estimated by adding confirmed deletion data to final post-SIR roll sizes. Deletion figures sourced from The Quint, Business Today, and Outlook India citing ECI data. 2026 final turnout from ECI via DD News and Open Magazine. 2021 district turnout from News Today, cross-referenced with ECI phase-wise data. Counterfactual turnout = (2026 votes cast) ÷ (pre-SIR roll). Pearson r tested at p < 0.05, df = 13.
Indian Political Debate 