Maharashtra’s Special Intensive Revision (SIR) has left 20,688,421 names off the SIR-2026 draft electoral roll, according to a preliminary report by the Sabar Institute, a public policy research organisation. This amounts to 21.1 per cent of the state’s 97,854,049 electors on the roll before the revision — effectively one in every five voters. In sum this means that the proportionate deletions of voters in Maharashtra (21.1 per cent) is second only to Arunachal Pradesh which ultimately saw 22 % of its electorate deleted. After Maharashtra, is Karnataka with 99.3 lakh electors vanishing from the electorate which amounts to 18 per cent of the state’s voters previous to the SIR.
The report, which examines deletions across all 288 Assembly constituencies and 36 districts, also reveals that the overwhelming majority of those whose names were placed on the ASDD — absent, shifted, dead and duplicate — lists were not recorded as dead.
It also finds substantial variations between Mumbai and the rest of Maharashtra, between urban and less urban districts, and among different categories of electors. The report cautions that it is preliminary and concerns the names left off the draft roll. Some affected persons may have reapplied and could subsequently appear on the final electoral roll. It also stresses that “correlation is not causation.”
It is on the Diwali weekend that the Sabar Institute made these public on X.
“Maharashtra’s Special Intensive Revision has left 2,06,88,421 names off the draft electoral roll. That is 1 in 5 electors (21.1 per cent). We analysed every single name, across all 288 constituencies,” the institute said in a post on X. Analysing the ECI’s constituency-wise data, the report found that deaths accounted for only 16.8% of the deletions. The remaining 83% were recorded under other categories, including “untraceable/absent” (37.5%), “permanently shifted” (37.1%) and “already enrolled” (8.5%).
Most of those removed are alive
One of the report’s central findings is that deaths accounted for only 16.8 per cent of the total deletions. According to the report, 17,206,190 people — or 83 per cent of all deletions — were marked as permanently shifted, untraceable or absent, or already enrolled elsewhere.
The report notes that 46 per cent of these living voters were below the age of 40. State-wide, the largest recorded category was “untraceable / absent” at 37.5 per cent, closely followed by “permanently shifted” at 37.1 per cent. Another 8.5 per cent were marked as “already enrolled”. The report notes that death becomes the dominant recorded reason primarily among older age groups, while among voters below 40, the overwhelming majority were placed in non-death categories.
Mumbai and its satellite cities carry the heaviest losses
The report identifies Mumbai and the surrounding urban belt as the area with the highest concentration of deletions.
According to the findings, Mumbai’s 36 constituencies account for 11 per cent of Maharashtra’s electors but 17 per cent of all deletions.
The report states that Mumbai lost 34.7 per cent of its electorate, compared with 19.5 per cent elsewhere in the state.
Thane district recorded particularly high losses. The report says 35.4 per cent of its electorate was affected by non-death deletions, placing it at the top of the district-level list. Other districts with high levels of non-death deletions included Mumbai City, Mumbai Suburban, Pune, Nagpur and Palghar.
At the other end of the scale, Hingoli and Gadchiroli recorded the lowest levels of non-death deletions, at about 6.6 per cent, according to the report.
Colaba saw more deletions than votes cast in 2024
The report highlights the scale of deletions in some individual Assembly constituencies by comparing them with voting figures from the 2024 Maharashtra Assembly election. Its most striking example is Colaba, where 119,606 names were deleted, while 118,499 votes had been cast in the constituency in the 2024 election. The report describes Colaba as the only seat where deletions exceeded the total number of votes cast in 2024, with deletions amounting to 1.01 times the number of votes cast.
It further says that 22 other seats in Mumbai, Thane, Palghar and Pune recorded deletions equivalent to between 75 and 100 per cent of the votes cast in 2024.
The stakes are electoral
The report also compares non-death deletions with the winning margins in the 2024 Assembly election. It finds that non-death deletions exceeded the 2024 winning margin in 190 of Maharashtra’s 288 Assembly constituencies.
In 95 constituencies, the report says, the “untraceable / absent” category alone exceeded the winning margin. The report does not claim that these deletions changed election results. Rather, the comparison is used to show the scale of the deletions relative to the margins by which seats were won.
The report’s summary states: “In two out of three seats, more living voters were removed than the number of votes that decided the 2024 winner.”
No evidence of an alliance pattern after controls
One of the report’s political findings is that it found no statistically significant alliance pattern once other factors were taken into account. At a simple state-wide level, seats won by the Maha Vikas Aghadi (MVA) recorded higher non-death deletions than those won by the Mahayuti. However, the report says this difference was substantially linked to the geographical concentration of MVA seats in Mumbai.
Outside Mumbai, the pattern changed. After controlling for Mumbai, region, reservation status, urbanisation and electorate size, the report found that the difference between MVA-held and Mahayuti-held seats was −0.27 percentage points, with a p-value of 0.82.
The report therefore concludes: “No alliance pattern survives Mumbai and urbanisation.”
Where turnout was lower, deletions were higher
The report identifies low voter turnout in the 2024 Assembly election as the strongest predictor of higher non-death deletions. It reports a Spearman correlation of −0.81 state-wide, and −0.74 outside Mumbai According to the report, constituencies in the lowest fifth of turnout recorded 34.9 per cent non-death deletions, while those in the highest fifth recorded 7.7 per cent.
The report says each percentage point of lower turnout was associated with 0.83 percentage points more non-death deletions. It interprets this relationship as being consistent with the possibility of stale electoral rolls, while presenting it as an association rather than proof of causation.
Two young women marked ‘shifted’ for every young man outside Mumbai
The report identifies a significant gender pattern among young voters marked as “permanently shifted”. Outside Mumbai, among electors aged 18 to 29, 822,798 women and 379,991 men were recorded as permanently shifted.
This amounts to approximately 2.2 women for every man. The report says this pattern is consistent with women moving to their husband’s residence after marriage and subsequently needing to enrol at a new address. Mumbai showed a different pattern, with young male deletions exceeding young female deletions. The report says this is consistent with the role of migration and male migrant workers in the city.
Overall, however, women and men were deleted at nearly identical rates — 21.4 per cent of women and 21.2 per cent of men on the 2024 roll. The difference became more visible after deaths were excluded. Outside Mumbai, 51.2 per cent of living voters removed were women, compared with 43.4 per cent in Mumbai.
More than half of third-gender electors were deleted
The report identifies third-gender electors as the group with the highest deletion rate.
According to its findings, 3,250 of Maharashtra’s 6,101 third-gender electors on the 2024 roll were placed on the ASD list. This represents 53.3 per cent of the third-gender electorate — more than two and a half times the overall state deletion rate of 21.1 per cent. The report says that 54.9 per cent of third-gender deletions were recorded as “untraceable / absent”, compared with 37.5 per cent among women and men.
Only 3.7 per cent of third-gender deletions were recorded as deaths. The report also notes that 75 per cent of the third-gender electors removed were below the age of 40. At the constituency level, it records particularly high deletion shares in several seats. In Bhiwandi West, 151 of 158 third-gender electors were deleted, while in Malad West, 313 of 345 were removed, according to the report.
Muslim-named electors account for 14.6% of deletions
The report estimates religious patterns through a name-based classification system, and explicitly states that this is a statistical estimate rather than an Election Commission record of an elector’s religion. According to this classification, 3,020,173 Muslim-named electors accounted for 14.6 per cent of all deletions.
The report compares this with a Census 2011 expectation of approximately 12.1 per cent, while also noting that the composition of deletions varies substantially across districts. The report says the Muslim-named share of deletions was 21.3 per cent in Mumbai and 13.2 per cent elsewhere.
It also emphasises the limits of name-based classification and says that its classifier was re-tested specifically on Maharashtra names.
Already enrolled” gap remains after comparing voters of the same age and seat
The report finds a difference in the recorded reasons for deletion when comparing Muslim-named electors with other electors. According to the findings, 12.3 per cent of Muslim-named deletions were marked “already enrolled”, compared with 7.9 per cent of other deletions.
Because deaths are more common among older populations, the report says it conducted comparisons between people of the same age group and in the same constituency.
After this adjustment, it found that the odds of a Muslim-named elector being recorded as “already enrolled” rather than dead were 2.07 times those of other electors of the same age in the same seat. For the “untraceable / absent” category, the corresponding odds ratio was 1.40, while the odds ratio for “permanently shifted” was 1.13.
However, the report expressly cautions against treating these findings as proof that any group was deliberately targeted.
It states: “None of this proves that any group was targeted. It shows where an audit of individual duplicate deletions would be best placed.”
Urbanisation, rather than community composition, explains much of the district pattern
At the district level, the report says that urbanisation is the strongest factor associated with higher levels of non-death deletions.
Outside Mumbai, it finds that districts with a larger urban population tended to have substantially higher deletion rates.
The report says that the relationship between urbanisation and non-death deletions was stronger than the relationship with the district’s Scheduled Caste, Muslim or Buddhist population shares. It cautions that these are area-level associations and should not be used to infer that individual members of a particular community were necessarily more or less likely to be deleted.
The report’s conclusion on this issue is that “urbanisation, not community share, drives the district pattern.”
The report’s broader warning: consequences may extend beyond voting
In its conclusion, the Sabar Institute says the implications of electoral-roll exclusion may extend beyond the right to vote.
The report notes that electoral-roll status can become connected with access to documentation and government services. It refers to difficulties involving passport renewal, the Public Distribution System, domicile and residence certificates, education, land transactions and vehicle registration.
It says prolonged uncertainty over electoral status can also create financial stress, psychological distress and disruption to family life. The report further raises concerns about delays in adjudication and appeals, arguing that a formal right to appeal may have limited value where people cannot easily obtain legal assistance, understand procedures, secure documents or afford repeated visits to administrative offices.
It identifies migrant workers, women changing residence after marriage, students, elderly persons, persons with disabilities, third-gender electors and people without continuous documentary records as groups that may face greater difficulty in documentation-heavy verification systems.
Demand for transparency, review and a countrywide study
The report calls for greater transparency around the use of algorithms, automated tools and logical discrepancy filters, saying these require independent examination and meaningful human review.
It also calls for sustained media coverage, civil society engagement and greater democratic oversight, including discussion in Parliament. Finally, the report says an independent countrywide study of the SIR is urgently required. Such a study, it says, should examine deletions and restorations, appeals and adjudication delays, demographic and regional patterns, documentary requirements, the use of algorithms and the social and economic consequences of exclusion.
Its concluding position is that maintaining accurate electoral rolls is necessary, but that “accuracy must be accompanied by transparency, accountability, accessible appeals, legal aid and protection against wrongful exclusion.”
The full report can be accessed from here.
Related
SIR: ECI “removes Form 6 declaration” but serious questions on legality of procedure remain
“It Is Not Just Numbers”: Prakash Raj on the Exclusion of 13.6 Crore Indians | SIR | Teesta Setalvad