The women dropping off the voter list
When India set out to clean up its electoral rolls, more women than men disappeared from them — and the documents the exercise demanded are precisely the ones women are least likely to hold. A look at the gendered edge of the Special Intensive Revision, and at the limits of what the published numbers can prove.
Between June and September 2025, the Election Commission of India ran a "Special Intensive Revision" (SIR) of Bihar's electoral roll — a door-to-door re-verification of every voter, the first of its kind in decades. The stated aim was housekeeping: remove the dead, the duplicated, and those who had moved away. When the final roll was published, it had shrunk by about 47 lakh names, from 7.89 crore to 7.42 crore. Then a second, larger phase rolled the exercise out across a dozen more states.
Buried inside the aggregate was a pattern the headline number hid: the people removed were disproportionately women.
In Bihar, the Election Commission's own figures record 22.7 lakh women deleted against 15.6 lakh men — women were about 59% of the attributed deletions, though they are under half the electorate. Bihar's voter sex ratio fell from 914 to 892, dropping in 37 of 38 districts. In the West Bengal phase, women were 61.8% of all deletions. The documents the SIR accepted as proof lean on exactly the papers women are least likely to have.
More women than men, in both states measured
The Election Commission attributed 22.74 lakh deletions in Bihar to women and 15.55 lakh to men. That is not a subtle skew: women made up roughly 59% of these deletions while forming under half the roll. When West Bengal's phase produced a frozen list months later, the tilt was sharper still — women were 61.8% of deletions there, and a majority of deletions in 219 of the state's 294 constituencies.
View data table
| Group | Deletions (lakh) |
|---|---|
| Women | 22.74 |
| Men | 15.55 |
The sex ratio of the roll tells the same story from another angle. Bihar's electoral sex ratio — women voters per 1,000 men — fell from 914 at the start of 2025 to 892 on the final roll, sliding in every district but one. A subsequent investigation by The Hindu found that the constituencies with the highest female turnout in the previous election tended to see the largest drops in roll sex ratio afterwards: exactly the seats where women had shown up to vote were the seats where women's names most often fell away.
The papers women don't have
To stay on the roll, a voter had to establish identity and descent from a list of accepted documents. That list excluded the papers poor and mobile citizens most commonly hold — Aadhaar, the voter ID card itself, and ration cards — and leaned instead on the birth certificate, the matriculation (class-10) certificate, and a parent's documents to prove lineage.
Each of those requirements has a gender edge. In Bihar, NFHS-5 records that about 39% of women have never attended school (against 18% of men), and only around 16% of women have completed twelve years of schooling (against 28% of men) — so a matriculation certificate as proof of identity excludes far more women than men by construction. And because marriage was the reason for roughly two-thirds of all female migration in the last census — with most women moving to their husband's village — a woman's birth and school records typically sit in her natal district, not the marital constituency where she now votes.
View data table
| Measure (Bihar, NFHS-5) | Women | Men |
|---|---|---|
| Never attended school (%) | 39 | 18 |
| Completed 12+ years of schooling (%) | 16 | 28 |
| Literate (%) | 55 | 76 |
The Supreme Court intervened repeatedly: it ordered the Commission to publish a booth-level, searchable list of the deleted names with reasons, and later directed that Aadhaar be treated as an additional identity document — while maintaining that Aadhaar is not, by itself, proof of citizenship. In May 2026 the Court upheld the constitutional validity of the revision itself.
Where women voted most, women's names fell most
The Hindu's data team lined up each of Bihar's 40 parliamentary constituencies two ways: how strongly women turned out to vote in the 2024 general election, and how far the constituency's roll gender ratio fell after the SIR. The seats where women had voted in the greatest numbers tended to be the ones that then lost the most women from the rolls. The relationship is real — if only moderate.
View data table (all 40 constituencies)
| Parliamentary constituency | Female turnout /1,000 men (2024) | % gender-ratio drop |
|---|---|---|
| Gopalganj | 1,218 | 8.38 |
| Kishanganj | 1,066 | 7.11 |
| Siwan | 1,140 | 3.96 |
| Sasaram | 887 | 3.49 |
| Jhanjharpur | 1,167 | 3.15 |
| Supaul | 1,155 | 3.09 |
| Maharajganj | 1,149 | 3.08 |
| Katihar | 1,080 | 2.72 |
| Aurangabad | 881 | 2.70 |
| Purnia | 1,024 | 2.65 |
| Sheohar | 1,100 | 2.54 |
| Karakat | 883 | 2.53 |
| Jahanabad | 917 | 2.17 |
| Madhubani | 1,168 | 2.15 |
| Purvi Champaran | 1,067 | 2.14 |
| Valmiki Nagar | 1,066 | 2.13 |
| Ujiarpur | 1,066 | 2.13 |
| Sitamarhi | 1,110 | 2.12 |
| Gaya | 884 | 2.07 |
| Samastipur | 1,077 | 2.04 |
| Bihar (average) | 1,017 | 2.01 |
| Khagaria | 1,107 | 2.00 |
| Begusarai | 1,054 | 1.99 |
| Buxar | 896 | 1.98 |
| Madhepura | 1,099 | 1.92 |
| Nalanda | 1,011 | 1.57 |
| Araria | 1,090 | 1.48 |
| Saran | 1,020 | 1.43 |
| Nawada | 899 | 1.29 |
| Jamui | 956 | 1.26 |
| Patliputra | 848 | 1.20 |
| Hajipur | 1,001 | 1.05 |
| Vaishali | 1,072 | 0.75 |
| Munger | 897 | 0.69 |
| Muzaffarpur | 1,018 | 0.40 |
| Paschim Champaran | 1,021 | 0.35 |
| Darbhanga | 1,082 | 0.15 |
| Arrah | 874 | 0.06 |
| Patna Sahib | 811 | −0.26 |
| Banka | 1,024 | −0.34 |
| Bhagalpur | 946 | −1.16 |
At the extreme sits Gopalganj, where 1,218 women voted for every 1,000 men in 2024, and whose roll gender ratio then fell 8.38%, from 967 to 886. But it is a tendency, not a law: some constituencies with below-average female turnout also saw sizeable drops, and three moved the other way. The point is not that the rolls targeted women who vote — the data can't show intent — but that the correction, whatever drove it, landed hardest where women's electoral presence was strongest.
View data table
| Assembly seat | Gender-ratio points lost |
|---|---|
| Kuchaikote | 95 |
| Gopalganj | 90 |
| Amour | 81 |
| Kishanganj | 80 |
| Barauli | 76 |
| Hathua | 76 |
| Bhorey | 66 |
| Baikunthpur | 65 |
| Sultanganj | 64 |
| Kochadhaman | 61 |
What this data can and can't tell you
This is a live, contested exercise, and the numbers demand more caution than most. Here is where the honest lines are.
- The gender split and the net figure use different clocks. The 22.7-lakh / 15.6-lakh women-and-men deletions are measured against the January 2025 roll; the 47-lakh net drop compares the pre-SIR total to the final roll and nets out new additions. They are not slices of the same pie — don't add or subtract one from the other.
- "More women deleted" is not "women deleted for being women." No published breakdown gives the reason for deletion by sex. Women being a larger share of deletions is consistent with the documentation mechanism described here, but the data cannot, on its own, prove intent or even the mechanism.
- The schooling gaps are context, not a measured cause. NFHS-5 tells us women are less likely to hold the accepted documents. It does not tell us how many deletions actually traced to a missing certificate — that link is argued by petitioners, not quantified in the roll data.
- The turnout-vs-deletion pattern is an ecological correlation. "High-female-turnout seats saw bigger sex-ratio drops" is a relationship across constituencies, not a statement about any individual voter.
- Bihar and West Bengal were measured slightly differently, at different stages of their revisions. Compare them as two readings of the same phenomenon, not as identical statistics. And no reliable caste- or religion-by-sex breakdown of deletions exists — claims about which women were hit hardest run ahead of the data.
- The turnout–deletion link is moderate, not mechanical. The correlation across constituencies is about 0.45 — a real tendency, but many seats break it and three saw their gender ratio rise. Read the scatter as a pattern, not a rule that predicts any single seat, and never as evidence of intent.
- The charts use different baselines and units. The constituency scatter measures the drop from the 2024 election roll to the final SIR roll (907 → 892); the seat bars measure the change between the January and September 2025 rolls (914 → 892). They also span three geographies — 40 parliamentary constituencies, 243 assembly seats, 38 districts — so the figures are not interchangeable.
Read carefully, the data supports a specific, defensible claim: women were removed from the rolls in greater numbers than men in both states where sex-disaggregated figures exist, and the verification rules relied on documents women are measurably less likely to possess. It does not, by itself, establish why — and the missing "why" is the number worth demanding next.
- Election Commission of India — SIR draft and final rolls; deletion figures as reported by Patna Press.
- The Hindu data team, "The unexplained drop in gender ratio following Bihar's SIR" (3 Nov 2025) — the constituency turnout-vs-drop dataset and the 243-seat gender-ratio change.
- The Wire, on the sex-ratio decline and its regression.
- BehanBox — West Bengal deletions by sex.
- ADR — the accepted-documents list and the Aadhaar exclusion.
- National Family Health Survey-5 (2019-21), Bihar report — schooling and literacy by sex; Census 2011 — marriage as the reason for female migration.
ImpactMojo Data (2026). "The women dropping off the voter list." ImpactMojo Data Dives. Retrieved from https://impactmojo.in/DataDives/women-and-the-sir.html
Have a dataset worth digging into?
Data Dives are independent investigations built from public data. If you know a number that deserves a closer look, tell us.
Pitch a Data Dive →