In the 357 of Uttar Pradesh’s 403 assembly constituencies with published notice lists, 2.86 crore of the 11.06 crore electors on the draft roll (26%; over a fourth of draft roll enrolled voters) were sent a notice asking them to establish their eligibility.In these constituencies, the draft stage had already dropped 2.52 crore names (18.4% of their pre-revision roll). About 92 lakh notices went to electors who could not be mapped to the 2003 roll, the last intensive revision, and 1.95 crore to electors flagged by software for “logical discrepancies”: a name mismatch, an implausible age gap with a parent or grandparent, or six or more children listed against one parent.Younger electors carried most of the burden: 39% of those under 40 (1.96 crore of 4.98 crore) were noticed, against 15% of those aged 40 and over (91 lakh of 6.08 crore). Most of the difference comes from logical-discrepancy flags, above all name mismatches: 19.5% of electors under 40 were flagged for one, against 3.3% of those aged 40 and over. Younger electors were also more often unmapped (11.3% against 6.7%).Notices also ran in families. Among electors whose father was noticed, 54% of children were noticed themselves, against 23% when a father of the same age was not noticed. The pattern holds for wives (44% against 25%) and is sharpest for siblings (71% against 8%). Electors paired with an unrelated person of the relative’s sex and age from the same polling station were noticed at 30-37%, so a relative’s notice predicts one’s own far better than the polling station does.Table 1. Notices by age, 357 constituenciesUnder 4040 and overDraft electors4.98 crore6.08 croreNoticed1.96 crore (39.3%)91 lakh (14.9%)Unmapped to the 2003 roll11.3%6.7%Any logical discrepancy28.7%8.6%Name mismatch19.5%3.3%Six or more children5.1%1.4%Percentages are of draft electors. Unmapped and logical-discrepancy rates use polling stations whose lists show both types.Women and men were noticed at almost the same rate (26.0% and 25.7%), but for different reasons. Women were nearly three times as likely to be unmapped to the 2003 roll (13.7% against 4.7%), and among those under 40, four times as likely (19.8% against 4.9%), a pattern consistent with women who moved after marriage. Men were far more often flagged for a logical discrepancy (21.3% against 13.2%), above all a name mismatch (16.1% against 3.9%).Muslims were 18.2% of draft electors in these constituencies but received 19.5% of the notices (55.8 lakh). Muslim men were noticed more often than Hindu* men (about 30% against 25%). Comparing electors of the same age, Muslims were 4.1 percentage points more likely to be flagged for a logical discrepancy, mainly for six or more children (3.5 points) and name mismatches (1.7 points), but 3.5 points less likely to be unmapped to the 2003 roll. Muslim women were much less often unmapped than Hindu women (8.3% against 14.9%).Table 2. Notices by age and religion, 357 constituenciesHindu, under 40Hindu, 40 and overMuslim, under 40Muslim, 40 and overDraft electors3.99 crore5.06 crore99.2 lakh1.02 croreNoticed38.9%14.9%41.1%14.7%Unmapped to the 2003 roll12.5%7.0%6.7%5.3%Any logical discrepancy27.2%8.4%34.7%9.7%Name mismatch18.9%3.2%22.0%4.0%Six or more children3.8%1.2%10.4%2.2%Religion is estimated from names; “Hindu” here covers all electors not classified as Muslim. Percentages are of draft electors.Two further details stand out. The two kinds of notice almost never overlap: only 52 electors received both. Flags for six or more children fell mostly on younger electors, and especially on Muslims: 10.4% of Muslims under 40 were flagged for this reason, against 3.8% of Hindus under 40.Data and methodsWe digitised the notice lists published polling station by polling station for the 357 constituencies where they were available, taking each noticed elector’s constituency, polling station (part) number, serial number, voter ID (EPIC) and the reason for the notice from the text of the PDFs. We also digitised, by optical character recognition, the SIR draft roll and the lists of names removed at the draft stage. The pre-revision roll is the draft roll plus those removal lists. Each noticed elector was linked to their entry in the draft roll by constituency, part and serial number, checked against the voter ID. Ages are as of 1 January 2026. Family members are linked through the relative’s name and house number printed on the roll. All figures refer only to the 357 constituencies with published notice lists; figures on notices cover the polling stations in those constituencies whose lists were published, which hold 99% of their draft electors. The rolls do not record religion. We inferred it from names with a statistical classifier, so religion figures are estimates and are reported only in aggregate, never for individuals (Chaturvedi & Chaturvedi, 2024). Small differences from official totals can arise from recognition errors and from constituencies whose notice lists were not published.*’Hindu’ here covers all electors not classified as Muslim.For our full coverage on the SIR and the issues it has brought to citizens, click here.