Asking the Right Questions: Castes and India’s 2021 Census

If not for the COVID-19 pandemic, India’s national census survey would have been now in progress. Once it takes off, it will include a new and interesting aspect. The enumerators will approach a part of the interviewees with lists of prepared caste and tribe names, and people will have to pick which social group they belong to.

The Indian caste system remains an unmeasured and astonishingly complex social phenomenon, a challenge multiplied by the size of the country and its population. Until today, nobody has even effectively counted how many castes there are in India. During the 2011 census, the New Delhi government attempted to do this for the first time since independence. The results were bewildering. As the questions about castes were open – each respondent could name their caste – the result was 4,673,034 different caste names. 

In 2015, a government commission was established – with Arvind Panagariya at its head – to filter these into a coherent list, by erasing millions of (assumed and real) errors and consulting with state governments to find out how many responses were simply regional names of the same community. But the commission apparently has not brought its work to an end; its head quit and moved on to work at an American university; and the government did not publish the full statistics on castes.

With hindsight, the ambitious attempt was perhaps doomed to fail. To illustrate the scale of the challenge one can merely look at history. It took the Anthropological Survey of India seven years (1985-1992) to conclude its research on the Scheduled Castes of India. Six hundred scholars worked on a select number of communities (450) and with a limited number of respondents. The result was contained in 120 manuscript volumes (and on 257 floppy discs). Of course, the survey carried much more information than just the caste names, as it sought to establish the general customs of each community, but even that work showed that the same castes may be called by a few different names and that one caste is often divided into various subgroups.

Still, the 2011 census experiment reminded us of important questions on the (self)-identifications of castes. It was assumed that much of the 2011 census data on castes had to be mistakes, such as spelling errors (and that’s what multiplied the results) and that many replies were just different names of the same castes. Respondents could have also given their subcaste, their clan or their family name, or another category as their caste name (indeed, a surname may be the same as a caste name, but very often it is not).

Such intriguing intellectual challenges are not limited to India and the issues of caste – they often appear where statistics-based tools are applied to the realm of identity. For the enumerators, the caste was a point in the survey, a countable category. For the respondents, it was a part of their identity, but intertwined with many other aspects – language, ethnic group, tribe, clan, and so on. Where officials wanted to divide these aspects into distinct compartments, for people these probably were more like concentric ripples in water, overlapping each other and splashing across the dividing lines. Similarly, in one of the past censuses, when asked about their mother tongue, a small group of respondents replied with the name of their profession – “Kisan” (agriculturalist).

And yet, while certainly many of the entries had to be different names of the same caste, or simply different spellings of the same name, an attempt to synthesize the data should be done carefully. Even such an innocent aspect as different spelling may not be that obvious in India. For instance, in the state of Uttar Pradesh, we find the Jatav caste and the Yadav caste – and while in some instances the “j” and “y” sounds may appear interchangeably (as in Yamuna/Jamuna), in this case they are two separate communities.

Much more importantly, the idea of various castes being just regional versions of the same caste is a preconceived, generalized notion in of itself. Social scientists long ago realized that there is no such thing as an overall, all-Indian caste hierarchy. In other words, as some scholars pointed out, we should not speak of a “caste system” as such, but rather “caste systems” – each system to be understood in its regional settings. By the same logic, can we speak of all-Indian castes at all? 

In some cases, we can clearly conclude that the same caste spans across various regions and states, and that its representatives identify themselves as one group (by forming caste associations, for instance). But it is often not that easy. Even when we take two communities from two different regions that follow the same professions and are placed at the same level of respective local hierarchies, we cannot be sure that they are the same caste, and that they would identify themselves as such, without resorting to deeper ethnographic study. Can we truly rely only on a single question in the census to answer that? 

For the next census, 2021 – the work on which will start this year – the government will thus not even attempt to ask all citizens about their castes. The census will reportedly “only” count those groups that belong to the Scheduled Castes (SC) and Scheduled Tribes (ST) categories. Moreover, the government will only let respondents pick from a digitally prepared set of replies, as reported by the Indian journalist Vasudha Venugopal.

Scheduled Castes and Scheduled Tribes are government categories, not customary social terms, however. As per the 2011 census, 16.2 percent of Indians were identified as belonging to the SC, and 8.2 percent as belonging to the ST. In the first years of the country’s independence, it had been decided that various backward groups should be subject to affirmative action, called “reservation” in India. The Dalit (untouchable) castes, which were to receive this government privilege, had been termed Scheduled Castes, and the tribes – Scheduled Tribes. Thus, it will be much easier for the administration to count these communities, because it had already been decided which groups belonged to these two categories. What we do not know is how many people belong to each of the castes and tribes within these categories at present.

This enumeration will carry not only sociological, but economic and political significance. Ever since reservation began, there have been various claims about which groups should be included in the two categories, and various changes were made over the years. Members of SCs and STs have demanded various changes in how the reservation system works as well. For the parties, knowing the exact numbers of members of a given caste is an important political variable, as it lets one assume, for instance, which demand raised during the campaign has a bigger chance of attracting a numerous electorate coming from a particular caste.

The New Delhi government has thus faced a dilemma: Should it let the people name their caste – or force them to choose from a set list of answers? It ambitiously tried the first option in 2011. It turned out that the second option, while more imposing, will more likely lead to concrete and publishable results, rather than another puzzle to solve.