A preliminary study by the Sabar Institute has found that 2.068 crore names were left off Maharashtra’s 2026 draft electoral roll following the Special Intensive Revision (SIR), accounting for 21.1 per cent of the 9.785 crore electors registered before the exercise.
Only 16.8 per cent of the deleted entries were recorded as belonging to people who had died. The remaining 1.72 crore entries, or about 83 per cent, were classified as permanently shifted, untraceable or absent, or already enrolled elsewhere.
The report, titled SIR Maharashtra, was prepared by Sabir Ahamed, Ashin Chakraborty and Souptik Halder. It draws on Election Commission of India (ECI) deletion lists, pre-revision electoral figures, the 2024 Assembly election results and Census 2011 data. The analysis covers all 288 Assembly constituencies across Maharashtra’s 36 districts.
Mumbai Records Higher Deletion Rate
The study found that deletions were particularly high in Mumbai and surrounding areas. The city’s 36 Assembly constituencies account for 11 per cent of Maharashtra’s electorate but 17 per cent of all deleted entries.
The overall deletion rate in these constituencies was 34.7 per cent, compared with 19.5 per cent in the rest of the state. Excluding deaths, the rate stood at 30.1 per cent in Mumbai and 16.1 per cent elsewhere. Thane recorded a 35.4 per cent non-death deletion rate, while Pune and Nagpur also reported high figures.
The researchers compared deletions with the results of the 2024 Assembly elections. They found that non-death deletions exceeded the winning margin in 190 of the state’s 288 constituencies. In 95 constituencies, the number of electors marked absent or untraceable alone was higher than the winning margin.
In Colaba, 1.19 lakh names were removed, compared with 1.18 lakh votes cast in the 2024 Assembly election. The report also identified 22 constituencies across Mumbai, Thane, Palghar and Pune where deletions amounted to between 75 and 100 per cent of the votes polled in 2024.
The study found no clear alliance-based pattern after accounting for factors such as region, urbanisation, reservation status and electorate size. The adjusted difference in non-death deletion rates between constituencies held by the Maha Vikas Aghadi (MVA) and the ruling Mahayuti alliance was just 0.27 percentage points.
The strongest association identified was between lower voter turnout in the 2024 election and higher non-death deletion rates. The researchers recorded a Spearman correlation of minus 0.81, indicating a strong relationship between the two.
Muslim-Named Electors and Duplicate Classifications
The report found that Muslim-named electors accounted for 30.2 lakh deletions, or 14.6 per cent of the total. This was higher than the 12.1 per cent share estimated by the researchers using district-level Census 2011 population figures.
The proportion varied across districts. Muslim-named electors accounted for 27.3 per cent of deletions in Aurangabad, 26.4 per cent in Parbhani and 24.6 per cent in Akola. Their share was 22.6 per cent in Mumbai.
The study also identified differences in the reasons recorded for deletion. Among Muslim-named electors, 12.3 per cent of deletions were classified as already enrolled elsewhere, compared with 7.9 per cent among other electors.
After comparing electors within the same constituency and ten-year age group, researchers found that Muslim-named electors had 2.07 times the odds of being classified as already enrolled elsewhere. Their odds of being marked untraceable or absent were 1.40 times higher, while the odds of being classified as permanently shifted were 1.18 times higher.
The pattern remained almost unchanged when the researchers applied a stricter classification requiring both the elector’s name and the name of a parent or spouse to be identified as Muslim. The higher proportion of Muslim-named electors classified as already enrolled elsewhere appeared in 285 of the state’s 288 constituencies.
The report raises the possibility that name-based matching may have contributed to the higher rate of duplicate classification. It points to common surnames such as Shaikh, Khan, Sayyed and Pathan. However, the ECI’s lists do not identify the other registration against which an elector was matched, preventing researchers from directly testing this explanation.
The researchers caution that their findings do not establish that Muslim voters were deliberately targeted. They also note that religious identity was estimated through a name-based classification system developed by the institute, not through official ECI records.
Shaikh was the most common surname among deleted entries, accounting for 753,978 deletions, or 3.64 per cent of the total. The 50 most frequent surnames together accounted for 30.1 per cent of named deletions. The report cautions that these figures cannot be treated as community-specific deletion rates.
Women and Third-Gender Electors
The study found broadly similar deletion rates among women and men. About 21.4 per cent of women and 21.2 per cent of men registered before the revision were removed from the draft roll.
However, women accounted for 49.8 per cent of deletions among living electors across the state. Outside Mumbai, their share rose to 51.2 per cent.
A significant difference emerged among younger electors marked as permanently shifted outside Mumbai. In the 18 to 29 age group, 822,798 women and 379,991 men were listed in this category, a ratio of 2.2 women for every man.
The researchers said this pattern was consistent with women moving to their husbands’ homes after marriage and being removed from electoral rolls at their parental addresses. They noted that the final electoral roll would help establish how many had registered at their new addresses.
Third-gender electors recorded the highest deletion rate among the groups examined. Of the 6,101 third-gender electors on the 2024 roll, 3,250, or 53.3 per cent, appeared on the deletion list. Only 3.7 per cent of their deletions were attributed to death, while 54.9 per cent were classified as untraceable or absent.
The impact was particularly high in Malad West, where 313 of 345 third-gender electors were included in the deletion list. Bhiwandi West recorded 151 deletions among 158 third-gender electors.
The report also found a strong association between urbanisation and non-death deletions. The Pearson correlation between the proportion of urban households in a district and its non-death deletion rate was 0.94. The researchers said urbanisation explained district-level variation more closely than religious or caste composition, while cautioning that area-level comparisons cannot establish the likelihood of deletion for individual voters.
Researchers Call Findings Preliminary
The Sabar Institute describes the report as a preliminary analysis of names missing from Maharashtra’s SIR-2026 draft electoral roll. Some electors whose names were removed may have reapplied and could appear on the final list. The researchers said the findings would need to be reassessed once the final roll is published.
The report also clarifies that removal from an electoral roll does not, by itself, cancel a person’s citizenship or legal right to public services. However, electoral documents are sometimes used as proof of identity, residence or domicile, and deletion can create difficulties when people access administrative services.
