Census, and tricky issues in recording caste

Census: The caste question in the questionnaire is posing knotty problems

| Photo Credit:

Now that the house-listing operations for the much-delayed Population Census have been completed and the population enumeration is eagerly awaited, the announcement that the Census questionnaire is being pilot-tested has come as something of a surprise. Unfortunately, little is known about its proposed contents.More seriously, press reports have quoted senior officials as saying that caste will be recorded simply as reported by the respondent. This raises fundamental questions about both the feasibility of counting castes and the utility of the data so generated. Such a decision would be particularly disturbing when the disastrous experience of the Socio-Economic and Caste Census (SECC) 2011 is still fresh in our minds.Mammoth exerciseThe accuracy of any census count requires that the item being counted be clearly identifiable. In an exercise as vast as the Indian population Census, it is, therefore, essential that the items included in the questionnaire are standardised, readily understood and capable of being recorded without ambiguity. A sample survey offers greater flexibility because enumerators can probe individual responses in greater detail. A Census, by contrast, must depend on uniform concepts and standardised procedures across the country to be adopted by millions of enumerators.Normally, Census questionnaires are subjected to extensive technical scrutiny, giving users some indication of the concepts and definitions underlying the resulting data. When the 2021 Census was announced, for instance, there was no proposal to include a caste question in the Census schedule. There was, however, an announcement that the National Population Register would be updated alongside the Census, paving the way for the preparation of a National Register of Citizens. The chronology of these operations, as described by the Home Minister, generated considerable informed public debate.The operationalisation of a caste question poses a major technical challenge. The experience of the SECC 2011 is a stark warning. Its caste data could not be satisfactorily tabulated, largely because caste names were not standardised and there was no comprehensive list against which responses could be classified. The exercise reportedly generated millions of caste and community names, making meaningful aggregation at the State level — and still more at the national level — virtually impossible.The old Census reports provide ample evidence of the difficulties that arise when caste is left entirely to the respondent’s description. Recent caste surveys conducted by States such as Bihar and Telangana followed a different approach by providing a framework for classification. Yet these experiences cannot simply be transplanted to a national census. A State-level census can restrict its classification to castes found within that State. A national census cannot do so because of India’s substantial inter-State migration.States’ experiencesIn Bihar, reported caste names were matched against a State master list of 214 unique castes. In Telangana, caste and sub-caste names were recorded along with the caste category used by the government. In either approach, however, migrants belonging to communities recognised in another State but absent from the local list could become problematic. They risk being classified as “others”, losing the identity necessary for meaningful national aggregation.The problem is not merely hypothetical. Delhi has a substantial Scheduled Caste population but, according to the Census, no officially recognised Scheduled Tribe population, since the recognition of Scheduled Castes and Scheduled Tribes is State-specific. Yet the 2011-12 National Sample Survey estimated roughly half a million Scheduled Tribe persons in Delhi. Many were migrants from States where their tribal status is recognised. If a national caste census relies primarily on State-specific lists, such inter-State migrants could easily disappear into an amorphous “others” category. The problem would be particularly acute in States and cities experiencing large-scale immigration.The preparation of these lists is, however, the critical task. The caste lists used in the 1931 Census, official documentation of castes and communities undertaken in the 1980s and documentation of the Socially and Educationally Backward Classes provide useful starting points. But no existing list can simply be adopted mechanically. Extensive consultation will be necessary to resolve differences in nomenclature, synonyms, sub-castes and claims of inclusion or exclusion.This would take time — perhaps a few months — but such a preparation is indispensable if the caste census is to produce data of analytical and policy value. A technically rushed census may generate an enormous volume of information while failing to produce relevant statistics.Complex issuesThe historical experience of the Census illustrates why conceptual clarity matters. The 1931 Census, for instance, faced difficulties in defining the category of “Anglo-Indian”, then an influential community. The Travancore Census report noted that, in the 1921 Census, some English-speaking Christians who adopted European modes of dress had been included in the category, apparently without justification. A more precise definition was subsequently adopted based on ancestry in the male line. Understandably, the new definition resulted in considerable number of people being excluded. The episode illustrates a perennial census dilemma: a category cannot be completely open-ended and simultaneously remain statistically comparable.Caste Census today presents complex conceptual and administrative problems. What is to be recorded when a respondent gives a broad caste category, a sub-caste, a community name, a local synonym or a caste name used in another State? How are overlapping names to be classified? How should claims that one caste is equivalent to, or distinct from, another be treated? These questions need to be settled before enumeration begins, not after millions of responses have been collected.There is also an important question about the placement of the caste question in the questionnaire. Putting it at the beginning may influence responses to subsequent socio-economic questions, with respondents potentially interpreting their economic and social circumstances through the lens of caste. The ordering of questions, therefore, deserves careful consideration. Significantly, caste was placed at the end of the questionnaire in the 1931 Census.The repeated changes in the timing of the Population Census, uncertainty about its coverage and continuing ambiguity about the eventual use of its data have already undermined confidence in one of India’s most important statistical exercises. The Census should not become an exercise driven by political compromise and expediency at the expense of technical rigour.If the caste question is to be included, the opportunity should demonstrate precisely the opposite: that a politically sensitive exercise can be designed with the highest standards of statistical and administrative ingenuity.Mohanan is former Chairman, Kerala State Statistical Commission; Kundu is Senior Fellow at IORA Ecological SolutionsPublished on September 3, 2026