Collected, then called unusable: the caste data India paid for and never saw
The last full caste count in a census was in 1931. The 2011 caste data was gathered and then buried. Census 2027 promises a count, and the notification that launched it forgot to mention caste. Easy mistake.
Primary source: The Leaflet: Centre tells SC it did not make results of 2011 caste census public because it is flawed ()
India decides who gets reserved seats, how large each quota is, and whether a group is backward, without a current national count of caste. That is a choice successive governments have made, not an accident.
What Bihar showed
Bihar’s survey reported that Extremely Backward Classes made up 36.01% of the state’s population and Backward Classes about 27.1%, together more than 63%. Scheduled Castes were 19.65%. Whatever one thinks of the policy the state then built on it, the survey demonstrated that a government can count caste and publish the result.
If 98.87% of the data was error-free in 2016, how did it become unusable by 2021?
The Centre has not published the error audit that would answer this. Maharashtra argued exactly that contradiction in court. The data was not released, and no independent body has been allowed to test the government’s claim.
What to watch in Census 2027
- Whether the caste question appears on the final schedule.
- What list of categories enumerators use, and whether respondents can answer freely.
- Whether caste tables are released on a fixed date, or left to political discretion.
A census that counts caste and then delays the tables would repeat 2011 under a new name.
The two statements that cannot both be true
In 2016, the government told a parliamentary committee that 98.87% of the individual caste and religion data from the 2011 SECC was error-free. In 2021, it told the Supreme Court that the caste data was not usable for policy. Officials have offered explanations: the free-text entries ran to millions of distinct caste names, many misspelled or overlapping; classification was never completed. Those explanations may be true. The public has never been shown the data to judge.
A dataset can be messy and still be published with its error audit, so that researchers, courts and the public can decide what it supports. Choosing not to publish at all is a decision, and it has a beneficiary: whoever would prefer the argument to remain unsettled.
Citations
- Union of India’s affidavit in the Supreme Court, September 2021, in Maharashtra’s petition for SECC 2011 caste data: called the caste data unusable for official purposes. The court dismissed the petition on 15 December 2021.
- Cabinet Committee on Political Affairs decision of 30 April 2025: caste to be enumerated in the next census. Document
- Census Act, 1948, and the notification of 16 June 2025 under its section 3 fixing 1 March 2027 as the reference date for most of India (1 October 2026 for snow-bound areas); the notification itself did not mention caste. Document
- Commission to examine sub-categorisation of OBCs (Justice G. Rohini), set up under Article 340 on 2 October 2017; report submitted to the President on 31 July 2023 after 14 extensions; not published.
Drafted with AI assistance from the cited public sources and translated into Hindi with AI assistance. Labelled source-linked: every claim is tied to a linked document, but this record has not yet been verified line by line by an editor. Corrections are logged in public.