A preliminary report by Sabar Institute has recorded 1,58,677 names removed during Manipur’s 2026 Special Intensive Revision (SIR), with 66.4% of the removals concentrated in the state’s 20 hill constituencies. Women account for 55.96% of the names and form the majority of removals in every one of Manipur’s 60 Assembly constituencies.
The report, dated September 27, 2026, was prepared by Sabir Ahamed, Souptik Halder and Ashin Chakraborty. It analyses the Absent, Shifted and Dead (ASD) lists published by the Chief Electoral Officer, Manipur, examining removals by constituency, district, polling booth, gender, age and recorded reason.
Its main finding is the difference between the hills and the Imphal Valley. The hill constituencies recorded an average of 79 removals per booth, compared with 31.2 in the valley. Churachandpur district alone accounted for 38,264 names, nearly one-fourth of the state total.
The authors do not conclude that any community was deliberately targeted. They identify places and categories requiring closer verification, particularly because the records do not distinguish people displaced since May 2023 from those who moved voluntarily.
What the report examined
The study covers all 60 constituencies and analyses records from 3,014 polling booths across Manipur’s 16 districts. The statewide average was 52.6 removals per booth, rounded to 53 in the report’s summary.
The authors checked the constituency files against the combined state file. They report that the records matched name for name and that each constituency’s total agreed with the sum of its available booth records.
However, 27 booth records in seven constituencies could not be downloaded from the election website. These included 15 records in Tamei, four each in Chingai and Tadubi, and one each in Phungyar, Ukhrul, Karong and Kangpokpi.
The figures for these seven constituencies are therefore lower bounds: they count the available records and may understate the full number. All seven constituencies are in the hill group.
The report also states that the authors did not have the complete electoral rolls, including voter totals by constituency, booth, gender or community. It consequently provides counts and averages per booth, rather than the percentage of registered voters removed. It makes no comparison with winning margins or turnout because the necessary electoral rolls and 2022 result sheets were unavailable to the study.
Hill constituencies account for 1.05 lakh names
The report divides Manipur into three geographical groups: 39 valley constituencies, 20 hill constituencies and Jiribam, which it treats separately.
The valley group covers Imphal East, Imphal West, Bishnupur, Thoubal and Kakching. The hill group covers Chandel, Churachandpur, Kamjong, Kangpokpi, Noney, Pherzawl, Senapati, Tamenglong, Tengnoupal and Ukhrul.
Nineteen of the 20 hill constituencies are reserved for Scheduled Tribes. Kangpokpi is a general constituency situated in a hill district. Together, the hill constituencies accounted for 1,05,294 removals across 1,332 booths—66.4% of the state total. The 39 valley constituencies accounted for 51,106 removals across 1,640 booths, or 32.2%. Jiribam recorded 2,277 removals across 42 booths, accounting for the remaining 1.4%. The average number removed per hill booth was about two-and-a-half times the valley average. Jiribam’s average was 54.2.
These figures describe the distribution of recorded removals. Without the total electorate in each group, they do not establish the comparative probability of an individual hill or valley voter being removed.
Churachandpur accounts for nearly one-fourth of removals
Churachandpur district recorded 38,264 removals across its four constituencies: Saikot, Churachandpur, Henglep and Singhat. Its average of 132.4 removals per booth was the highest among the 16 districts. The district’s total exceeded the combined 33,896 removals recorded in Imphal East and Imphal West, which together contain 23 constituencies.
Among the other hill districts, Tamenglong recorded 10,973 removals, Kangpokpi 10,425, Ukhrul 10,050 and Senapati 7,689. Pherzawl recorded 6,732, Chandel 6,672, Tengnoupal 5,228, Noney 4,759 and Kamjong 4,502. In the valley, Imphal West recorded 17,187 removals and Imphal East 16,709. Thoubal recorded 6,957, Bishnupur 6,279 and Kakching 3,974.
The district comparison shows variation within the hills as well. Noney averaged 91.5 removals per booth, followed by Pherzawl at 86.3 and Chandel at 83.4. Senapati’s average was substantially lower, at 34.2. Kakching had the lowest district average, at 26 removals per booth. Bishnupur and Thoubal followed at 26.3 and 26.6 respectively.
Saikot records the highest constituency total
Saikot had the largest number of removals among all 60 constituencies: 14,152 across 84 booths. It also had the highest average, at 168.5 removals per booth.
Churachandpur constituency followed with 9,381 removals, Henglep with 8,130, Tamenglong with 7,000, Chandel with 6,672 and Singhat with 6,601. Ukhrul recorded 6,382, Tengnoupal 5,228, Nungba 4,759 and Thanlon 4,696. The ten constituencies with the highest removal totals were all in the hill group.
The ranking changes when measured per booth. After Saikot, Henglep averaged 135.5 removals, Singhat 117.9, Thanlon 114.5, Churachandpur 105.4 and Tamenglong 104.5.
Wangkhei stood out in the valley. It recorded 3,756 removals across 50 booths, averaging 75.1 per booth—more than twice the valley average and the eleventh-highest constituency average statewide. At the other end, Singjamei recorded the smallest total, at 671 names. Hiyanglam recorded 692 and Sugnoo 848. Thongju had the lowest average per booth, at 19, followed by Hiyanglam at 19.2 and Moirang at 19.6.
“Permanently shifted” is the most common recorded reason
The ASD records classify removals under four reasons. “Permanently Shifted” accounted for 72,473 names, or 45.7% of the total. Death accounted for 43,000 names, or 27.1%. Another 35,810 names, or 22.6%, were classified as “Untraceable/Absent”. The remaining 7,394, or approximately 4.7%, were recorded as already enrolled elsewhere.
These are the reasons entered in the records. The study does not independently establish the circumstances of every person listed under them. The geographical differences were substantial. In the valley, death accounted for 43.2% of removals, permanent shifting for 35.8%, absence or inability to trace the voter for 16.6%, and enrolment elsewhere for 4.4%.
In the hills, permanent shifting accounted for 50.4%, absence or inability to trace the voter for 25.7%, death for 19.1%, and enrolment elsewhere for 4.9%. Taken together, shifting and absence accounted for about 76% of hill removals, compared with approximately 52% in the valley.
The authors also found that constituencies recording more removals per booth generally had a larger share attributed to shifting and a smaller share attributed to death. They explicitly state that this relationship does not explain why the removals occurred.
Recorded reasons differ between constituencies
Nungba had the highest proportion of removals attributed to permanent shifting: 3,359 of its 4,759 names, or 70.6%. Tamenglong followed at 69.5%, Thanlon at 67.9%, Ukhrul at 67.1% and Tamei at 63.7%.
Death accounted for the highest share in Lamsang, at 62.9%, followed by Konthoujam at 62.4%, Patsoi at 62.1% and Thongju at 61.8%. Thanga had the highest proportion recorded as untraceable or absent: 675 of 1,087 names, or 62.1%. The corresponding shares were 48.6% in Uripok, 47.7% in Wangkhei and 44.6% in Tengnoupal.
Saikot recorded 5,611 names under this category, representing 39.6% of its removals. Its other entries comprised 5,242 permanently shifted voters, 2,839 deaths and 460 cases of enrolment elsewhere. Moirang had the highest proportion recorded as already enrolled elsewhere, at 21.1%, involving 199 names. Heirok followed at 12.7%, Mao at 11.2% and Uripok at 10.1%.
The report cautions that a high percentage within a constituency’s removals does not necessarily mean a large number of people. Both the count and its share need to be read together.
Women form the majority in every constituency
The records contain 88,799 women, 69,871 men and seven persons recorded under the “Other” gender category. Women constituted 55.96% of all names counted, while men constituted 44.03%. Women formed the majority in every constituency. Their share ranged from 51.1% in Patsoi to 68.8% in Wabgai. Other constituencies with high female shares included Bishenpur at 68%, Khangabok at 66.7%, Lilong at 65.6% and Wangjing Tentha at 64.3%. Thoubal and Keirao each recorded 64.2%.
At district level, women accounted for 64.1% of removals in Kakching, 63.5% in Thoubal and 62.9% in Noney. The female share was 58.4% in the valley, 54.8% in the hills and 54.6% in Jiribam.
The authors make a specific qualification: because the study lacks the gender composition of the full electorate, it cannot establish whether women faced a higher removal rate than men. It establishes that more women than men appear in the removal records.
Women’s removals are more often linked to shifting
The recorded reasons also differ by gender. Permanent shifting accounted for 52.4% of women’s removals, compared with 37.1% of men’s removals. Viewed across the entire permanent-shifting category, women accounted for 64.2% of those listed.
Death showed the opposite pattern. It accounted for 35.3% of men’s removals and 20.6% of women’s removals. Absence or inability to trace the voter accounted for 22.3% of women’s removals and 22.9% of men’s. Enrolment elsewhere accounted for approximately 4.7% in both groups. The report further records that 46,786 of the 88,799 women—about 53%—were listed against a husband’s name rather than a father’s.
The authors identify marriage-related changes of residence as an issue requiring attention when a woman’s enrolment moves between two places. However, the records do not establish how many women were removed specifically because of marriage or how many subsequently failed to secure enrolment at their new address.
The report does not claim that these figures demonstrate different treatment of women by election officers.
People aged 30–44 form the largest age group
People aged 30–44 accounted for the largest number of removals, at 57,103. They were followed by those aged 45–59, at 38,408. The records also included 25,086 people aged 60–74, 19,600 aged 75 or above, and 18,480 aged 18–29.
The statewide median age was 46. In the valley it was 48, compared with 45 in the hills and 44 in Jiribam. People aged 60 or above accounted for approximately 34% of valley removals, compared with 26% in the hills. The report connects this difference with the higher proportion of death-related removals in the valley.
Its age charts show that removals among older people were more often attributed to death, while those among younger people were more often attributed to shifting.
One-tenth of booths account for 36% of removals
The removals were concentrated in a relatively small number of booths. The most affected tenth—roughly 300 booths—accounted for 36% of all names counted. The most affected quarter accounted for 62%. The median booth recorded 32 removals, below the statewide average of 52.6. A total of 97 booths recorded at least 200 removals each.
Under the new 2026 booth numbering, Leijangphai in Thanlon recorded the highest count, at 426 names. Permanent shifting accounted for 96% of its removals. Hamleikhong Hundung in Ukhrul followed with 412, of which 84% were attributed to shifting. In Saikot, Tuibong (S.W.-II) recorded 396 removals and Khopuibung 386; untraceable or absent voters accounted for 84% and 75% respectively.
Wangkhei’s booth 12, listed as Kabo Leikai Nongpok in the report, recorded 369 removals, with 81% classified as untraceable or absent. Sadim in Karong also recorded 369, with 86% under that category. Kaimai in Tamei recorded 366 removals, 92% classified as untraceable or absent. Ukhrul (E-2) recorded 335 and Azuram in Tamei 324, with every removal at both attributed to permanent shifting.
Changed booth numbers affect comparisons
The report warns that booth numbers changed in 29 constituencies. Researchers cannot assume that a booth carrying a particular number before revision corresponds to the booth with the same number afterwards.
Across the records, approximately 62% retained their old booth number. The report therefore presents separate tables using the old and new numbering.
This distinction changes the highest booth total. Under the old numbering, Saikot’s part 49 recorded 446 removals—the largest count in that table. Its records are associated with new parts 53 and 54, and 78% were classified as untraceable.
Similarly, old part 78 in Churachandpur recorded 373 removals distributed across new parts 86 and 87.
The authors include both systems so that verification can identify the original polling area and avoid misleading comparisons caused by renumbering or the division of records.
Displacement cannot be separated from ordinary migration
The report places particular emphasis on displacement since May 2023. A person staying in a relief camp or another district may be recorded as shifted or untraceable at the original address, even when the departure was involuntary.
The source data does not identify displacement as a separate category. The study therefore cannot determine how many entries marked permanently shifted or untraceable concern displaced people.
This limitation is central to its findings. The concentration of removals in hill constituencies, together with their high proportion of shifting and absence entries, identifies where verification should begin. It does not, by itself, prove that those entries were incorrect or that a community was deliberately excluded.
The authors recommend checking these cases booth by booth, with particular attention to families unable to return to their recorded addresses.
Names cannot establish community identity
The report also examines the final word in people’s names, while expressly warning against treating it as a reliable community identifier.
Manipur’s electoral records contain different naming practices. A final word may be a clan name, village name, father’s name or title, rather than a surname.
Across 1,58,671 records used for this analysis, the authors counted 50,647 distinct final words. The ten most common accounted for 16.8% of those records.
The difference between the valley and hills was substantial. The ten most common final words accounted for 37.5% of the valley records examined, compared with 9.3% in the hills. The study counted 9,771 distinct final words in the valley and 42,871 in the hills.
Statewide, the most frequent were Devi, appearing 10,303 times, Singh, appearing 6,406 times, and Haokip, appearing 3,725 times.
The authors draw no conclusion about community-wise removals from these counts. They instead raise a practical concern automated methods that match or classify people by names may perform differently across Manipur’s naming systems and require careful examination.
Concerns extend beyond voting
The report discusses how uncertainty over electoral status can create difficulties in administrative processes where electoral documents are used to establish identity or residence.
It refers to cases examined by Sabar Institute involving passport renewal, Public Distribution System access, domicile and residence certificates, education and other services. These observations form part of its broader discussion of SIR; they are not findings established by the Manipur ASD dataset.
It nevertheless argues that demands for documents, repeated visits to offices and unresolved appeals can place additional burdens on affected people. It identifies displaced families, migrant workers, women changing residence after marriage, students, older people, persons with disabilities, third-gender electors and people with incomplete documentary records as groups requiring particular attention.
Verification, accessible appeals and publication of voter totals
The report recommends beginning verification with booths recording the largest concentrations of removals, especially cases classified as permanently shifted or untraceable.
It calls for publication of constituency-level voter totals so that researchers can calculate removal rates, rather than relying only on counts and averages per booth.
It also recommends simple objection procedures, accessible appeals and free legal assistance for people unable to manage the paperwork or afford repeated visits to administrative offices. Displaced families and women changing residence after marriage require particular attention, the authors state.
Its wider recommendations include transparency and human review where automated tools or discrepancy filters are used, documentation of individual cases by researchers and civil society groups, parliamentary scrutiny, and an independent nationwide study covering deletions, restorations, appeals and adjudication delays.
For Manipur, the report identifies a specific task that verify the concentrated removals in the hill constituencies and establish whether people recorded as shifted or absent remain eligible for enrolment, including those living away from home because of displacement.
The report may be read here.
Related:
Bringing together fact-finding reports on Manipur, a commitment to conflict-ridden Manipur
Just days before polling, two killed in fresh violence in Manipur