The Census: Counting a Vast Nation

Reference library · 1123 words

Counting more than a billion people, in a country with over a hundred languages and a population that includes nomadic herders, pavement dwellers and residents of Himalayan valleys cut off by snow for half the year, is one of the largest peacetime administrative operations attempted anywhere on Earth. India has done it every ten years since 1881, without interruption through two world wars, famine, Partition and multiple wars with neighbours. The chain broke only recently: the census due in 2021 was postponed as the COVID-19 pandemic took hold, and the delay has now stretched for years, leaving the country running on figures collected in 2011.

The institution is older than the Republic. A first attempt at an all India enumeration was made in 1872 under the colonial administration, but it was not synchronous, meaning different areas were counted at different times, which produces double counting and gaps when people move. From 1881 the count was synchronous, and the decennial rhythm has held ever since. After independence the operation was placed on a statutory footing by the Census Act of 1948, which is still the governing law. The work is carried out by the Office of the Registrar General and Census Commissioner of India, which sits within the Ministry of Home Affairs. The Act does two important things: it obliges the public to answer, and it forbids the disclosure of individual records, so census returns cannot be used against a particular person by any other arm of government. That confidentiality is the foundation of the whole exercise, because a census that people fear will be used to tax, deport or prosecute them is a census that will be lied to.

How the count is actually made

The Indian census is conducted in two phases roughly a year apart. The first, house listing and housing, records every structure and whether it has electricity, a latrine, a drinking water source, a bank account, a bicycle, a mobile phone. This phase is the single richest source of data on Indian living standards. The second phase, population enumeration, counts people and records age, sex, marital status, religion, language, literacy, education, occupation, migration and disability, and is completed within a tight window of a few weeks, with a short revisional round afterwards to catch births, deaths and movements against a fixed reference moment. For the 2011 census the enumeration ran through February 2011, and areas that are snowbound were counted several months earlier on a separate reference date.

The fieldwork is done not by professional statisticians but by an enormous temporary army, numbering in the millions, drawn largely from government schoolteachers who are deputed for the purpose and trained through a cascade of instruction from national trainers down to district and block level. This has obvious virtues: teachers are literate, locally resident and already on the public payroll. It also has costs, since schooling is disrupted and enumerators have limited time to master a complex schedule. Quality is checked afterwards through an independent post enumeration survey that re-counts a sample and estimates the level of omission.

The results are used for far more than curiosity. Constituency boundaries, reserved constituencies for Scheduled Castes and Scheduled Tribes, the share of central tax revenue each state receives on the recommendation of the Finance Commission, and the coverage of national welfare schemes all rest on census figures. That explains why the delay matters in practice. The National Food Security Act sets coverage as a proportion of rural and urban population, and because the shares are applied to 2011 figures, critics argue that many people who would now qualify are excluded simply because no newer count exists.

What the census does and does not ask

The 2011 count returned a population of roughly 1.21 billion, an overall sex ratio of 943 females per 1,000 males, a literacy rate around 74 per cent, and an urban share close to a third. Two of those numbers carried heavy political weight. The child sex ratio in the zero to six age group had fallen further, which was read as evidence that sex selective abortion had spread rather than receded despite the law prohibiting prenatal sex determination. And the urban figure understated urbanisation, because India's definition of a town requires a minimum population, a minimum density and a threshold share of male workers in non agricultural work, so many dense settlements are still classified as villages.

The language tables illustrate how the census constructs categories rather than simply recording them. Enumerators write down whatever mother tongue a respondent names, which produces raw returns numbering in the tens of thousands. Linguists at the census office then group these into rationalised mother tongues and finally into a much shorter list of languages, with a cut off that excludes tongues with fewer than ten thousand speakers from separate publication. The grouping decisions are technical, consequential and disputed, since a community's claim to be a distinct language rather than a dialect of Hindi has implications for schooling and recognition.

The most politically loaded omission concerns caste. The census has enumerated Scheduled Castes and Scheduled Tribes throughout the independence era, because those categories carry constitutional entitlements. But a general caste enumeration, covering all other communities, has not been part of the census since 1931. A separate Socio Economic and Caste Census was conducted in 2011 outside the regular census framework, and its caste data were never fully released, with officials citing quality problems in the returns. Demands for caste to be counted in the census proper have come from parties across the spectrum, on the argument that reservation policy cannot be evaluated without knowing group sizes. Opponents argue that formally counting caste entrenches it. The government announced in 2025 that caste would be enumerated in the coming census, which is scheduled to be conducted with a reference date in 2027 and an earlier date for snowbound areas. The exercise is expected to be substantially digital, with enumerators using an app and a self enumeration option offered to households.

Two further controversies shadow the exercise. The house listing phase has been linked in policy documents to the National Population Register, which in turn has been associated in public debate with a proposed national register of citizens. That association triggered widespread protest in 2019 and 2020 and made some communities apprehensive about answering, which is precisely the kind of fear a statistical agency cannot afford. Separately, statisticians have argued that the long gap has degraded every dataset that uses the census as a sampling frame, including the surveys that produce India's employment, consumption and health estimates. The census is not just one dataset among many. It is the scaffolding on which the others are built, and the scaffolding is now more than fifteen years old.

Back to the library