Singular fixation on ‘caste census’ in autocratic governance: A transgender position

Getting your Trinity Audio player ready...

There can be no doubt that data, in its presence and absence, is an essential tool for understanding the world we live in. However, the current unchecked focus on collecting or releasing census (particularly caste census) data to the exclusion of all other data in India can very well be seen as a pile of hogwash. As a rule, one might say that the moment those in power accept and enlarge focus on one particular demand or position coming from the people it has likely been co-opted by the government and thus lost its strategic and emancipatory potential. This has become abundantly clear as various states release ‘caste census’ data followed by all round panic particularly among the ‘backward classes’. Therefore, it becomes necessary for those working towards emancipation to rethink the perspective and strategy, without compromising on principles.

There are two fundamental problems with the census data juggernaut.
1. Methodological erasure and recasting of categories, and especially the transgender category. This combined with the nonsensical refrain of ‘reservation/participation to the extent of population’ means compulsory suppression of data on transgender persons across the board.
2. Incoherence and a lack of reasoned methodological evolution in collection of data, which in turn signifies that all data, apart from perhaps the total population numbers, suffer from either meaninglessness or uselessness taken individually or together. That is, it cannot reflect substantial demographic changes over time, even at the level of the fundamental unit used in Indian Census, i.e. family or household. Even though family and household are not interchangeable units, that is how it is used in data collection and that is already an intrinsic problem with the methodology.

I will look at public information available from the Telangana Caste Census (2024-25), the Karnataka Caste Census (2014-15), the Bihar Caste Census[1] (2022-23) and the All India SECC (2011). Reference will also be made to other surveys or data that may be more or less relevant, as the case may be, to establish the many contradictions.

Ambedkar’s position on census and population statistics

Before beginning, it is essential to reproduce the entirety of Appendix IV from States and Minorities, that clarifies Ambedkar’s position on census and how it impacts minorities.

Appendix IV

STATISTICS OF POPULATION

The population of the Scheduled Caste in 1941 is estimated at 48,793,180. Can this figure be accepted as accurate?

In coming to a definite conclusion on this issue the following points must be borne in mind: —

(i) The population of the Scheduled Castes in 1941 as compared with their population in 1931 shows a decline.

(ii) The population of all other communities during the same period shows an increase of 15 p. c.

The question is whether there is any special reason why the population of the Scheduled Castes should have declined.

Was the basis for computation of the population of the Scheduled Castes in 1931 the same as in 1941? The answer is in the affirmative. The figures given for 1931 are the result of recasting of the Census of 1931 in the light of the definition of “Untouchables” given by the Lothian Committee. The same basis was adopted in 1941. It cannot therefore be said that the decline in the population of the Scheduled Castes in 1941 was due to an over-estimate made in 1931.

It is true that the Census for 1941 does not give any figures for the Scheduled Castes for Ajmer-Merwara and Gwalior State. But even adding to the total for 1941 the figures for these two areas as they stood in 1931 the population comes to only 49,538,145 which still shows a comparative decline.

The want of any proper explanation for the decline of the Scheduled Caste population and an increase in the population of all other communities during the same decade only reinforces the impression which every honest student of Indian census has formed namely that the Census of India has over a number of decades ceased to be an operation in demography. It has become a political affair. Every community seems to be attempting to artificially augment its numbers at the cost of some other community for the sake of capturing greater and greater degree of political power in its own hands. The Scheduled Castes seem to have been made a common victim for the satisfaction of the combined greed of the other communities who through their propagandists or enumerators are able to control the operations and the results of the Census.

In the light of these circumstances it is fair to demand that an accurate figure for the population of the Scheduled Castes would be the Census figure as corrected by the inclusion of the population for Ajmer-Merwara and Gwalior State plus an increase of 15 per cent, to give them the benefit of the general rise in the population.

Therefore, at the outset: census is intended for the study of demography and census (population count) is not for deciding percentage reservation.

It is important to note that, in their criticisms of recent state-wise ‘caste census’ data, representatives of Backward Classes are often seen to be implying (on similar lines to what Ambedkar has stated above) that Backward Classes numbers are being suppressed for political reasons[2]. They also make inter-state comparisons on Backward Classes data[3]. These criticisms cannot be taken at face value. For two reasons: one, the definition question. As Ambedkar has pointed out above, the definition/basis of computation for one category i.e. Scheduled Castes was the same in 1931 and 1941. Members of the motley group of Backward Classes, can make no such claim because the scientific basis of definition/classification and inclusions/exclusions within Backward Classes is itself under scrutiny. Two, leading from one, the Backward Classes is not a uniform nationally identified and fixed category, leaving the States to play fast and loose with the categories therein. Further, State boundaries have also been redrawn in the past few decades, with Telangana being officially formed only a decade back. Therefore, while the state-wise share of SC/ST populations across states might reflect a degree of parity with the national percentages, it is not meaningful or logically sound to make inter-state comparisons of Backward Classes’ share, growth or decrease in population etc.

Census: Part constitution, part French delusion, zero democratic evolution

The broad list of questions[4] that have been part of the Census from 1872-2011 is available in public. Questions that are to be part of 2025 census have not been made available here. Looking at what is available, the following becomes clear:

From 1872-1941 the data collected was to reflect biological sex, marital status, level of literacy (including degree of spread of English education), occupation (dependency and employability), language, type of disability (called infirmities), place of birth, religion and ‘caste, race or tribe’.
From 1872-1941 the caste and religion questions underwent some changes reflecting the way in which the colonial government was looking to deal with the society: First it appeared as “religion” and “caste or class”, then it went through the following changes every decade: “Caste, if Hindu, sect if other religion” –> “Main religion, sect of religion” AND “Caste or race – main, subdivision of caste or race” –> “Caste of Hindus & Jains, Tribe or race of others” –> “Religion & sect of Christians” AND “Caste of Hindus & Jains, Tribe or race of others” –> “Religion” AND “Caste, Tribe or Race” (from 1921-41). That is, by 1921, Caste, Tribe and Race had become largely interchangeable terms for the colonial government.
From 1951-2011 the central unit clearly became ‘household-family’, and the data collected largely continued the indicators from pre-independence. A key difference was that fertility of family became an essential question. From 1961-1991 disability (or ‘infirmities’) was not part of the questionnaire.
From 1951-2011 the question of identity was organized in this order: nationality (including place of birth, residence and eventually migration history), religion and ‘special groups’. The term ‘special groups’ was dropped from 1961 when it clearly became the enumeration of SC/ST population with specific caste or tribe within that category enumerated from 1981-2011. The word ‘indigenous’ appears only once, in 1951 as an optional state-level query to assess land ownership.
Notably, from 1951-1991, the word ‘occupation’ was not part of the questionnaire. Even though the National Classification of Occupations has been done from 1946-2015. (That classification, in light of statutory ban on manual scavenging and criminalization of sex work, demands a separate study not within the scope of this work).

At present, the counting system is that the oppressors are counting only one part of the most oppressed. This is different from the British colonial system where the colonizer did count everyone, albeit with errors, including ‘his’ people who were residing in the colonies.

The reason for collecting the data on SC/ST population comes from the Constitutional mandate. The irrational reason for not collecting Caste, Tribe or Race data as a universal data set is similar to the French (and German) position[5] on enumeration of race: i.e. ‘we are an equal (casteless) society’. In 1978, the French Constitutional Law officially banned the State from collecting data on race, ethnicity and religion. On the contrary, both UK and USA collect census data on race and/or ethnicity (though not interchangeable) largely on the principle of self-identification/declaration. In Africa, South Africa is known to have completed over a century of ‘enumeration of race’ from Apartheid to post-Apartheid[6] period. In South Asia, only Nepal[7] is known to have conducted two national-level caste censuses.

These comparisons are relevant for at least two reasons. First: today caste classification is being seen as the foundation or predecessor for race classification. Whether that is fully true except at the experiential realm or simply at the level of general oppressor psyche is not a question I am dealing with here. But, as per the current understanding of history, caste classification pre-dates race classification. Therefore, even though a ‘later’ development, how the race classification has been dealt with in ‘modern’ enumeration becomes important. Second: to understand methodology and evolution of process, if any.

Undemocratic, untenable classifications

The combination of the constitutional mandate with the French method of denial has created a continuing situation of governments making and unmaking classifications at their will. A specific character of the pre-independence census is that it clearly separated ‘religion’ from ‘caste, race or tribe’. And both were universal categories. Intermittently, religious sub-sect and sub-caste data is said to have been included (whether appropriately or not). In the post-independence period, however, this changes entirely to religion as a universal category, and only SC/ST as a singled-out caste and tribe category. In the political interim is occurring the contested OBC classifications of all Muslims in certain states, certain dominant Hindu castes, and the classification of converted SCs and STs. So, one can only know the overall change in population figures of religious communities (without sub-sect data) or the overall change in population figures of (presumably Hindu) SCs and STs. Every other permutation is largely a mystery.

Does the All India SECC (2011 Caste Census)[8] address any of these concerns? No. In fact, the SECC data is confounding, because the published data doesn’t provide information on religion. It then makes a separation between rural and urban such that SC, ST and ‘other’ caste data appears for rural households but not for urban households.

In the Telangana Caste Census 2024-25, as per the instruction manual available online[9], the classifications open up questions on the caste-religion-conversion-self-determination conundrum. This survey requires enumerators to place all SCs and STs converted to Christianity (and their progeny) in ‘Backward Class (C)’. This is relevant for two reasons: One, the KG Balakrishnan Inquiry Commission’s report pertaining to reservation rights of converted SCs and STs is yet to be tabled and the matter is also pending in the Supreme Court since the past two decades. Two, this significantly weakens some of the criticism about falling numbers among Backward Classes, as this unjust classification would actually falsely enlarge the total Backward Classes count and reduce the total SC, ST figures.

Similarly, ‘Backward Class (E)’, in the Telangana Census includes, among others, oppressed caste Muslims. For example, all washer and barber castes converted to Islam, who face varying degrees of untouchability and/or social segregation appear alongside other caste groups including Shaiks[10]. In the OC category, there is a mix of religion, caste and community. So, Buddhists, Jains, Sikhs, non-BC(C) Christians, and non-BC(E) Muslims are placed alongside Brahmins, Lingayats, Reddys and Kammas.

In 2016, the National Commission for DNT/NT/SNT communities released a provisional list for state-wise re-categorization of DNT/NT/SNT communities within existing SC, ST and OBC categories. Thereby potentially erasing their specific identities, structural issues and statutorily criminalized histories that compounded the disabilities of their caste or communal status. The same ‘re-classification’ has been followed in the Telangana Census.

The methodological question here is this: If the purpose of the socio-economic caste census is to gain a complete view of various populations as they stand in the present, in order to subsequently inform policy decisions, why has the existing/contested policy categorizations been codified and superimposed prior to or during the course of enumeration? Why has religion and caste been subsumed, replaced and conflated with each other? Why is it not that each caste, tribe (whether scheduled, notified, backward or otherwise), religion was codified and counted separately as per the self-declaration of those enumerated? That the codification and classification of categories and the responses elicited are at odds with each other is clearly seen in the questionnaire which seeks the individual religious and caste identities of family members. How did the enumerators resolve these contradictions of organizing the data, when it is clear to any observer of a caste society that religion as an axis for identity would essentially be a horizontal one, insofar as it can be an identity of birth and/or self-determination?

Even though the Karnataka Census (2014-15) is currently under discussion and details are yet to be made available, a similar situation may be said to exist. As per various news reports, the data from this census has already led to the creation of a new category, i.e. Category 1 B, for Most Backward among the Backward Classes. The basis of this new categorization is unclear and there is another hurried proposition of extending ‘creamy layer’ to all backward class categories.

In the Bihar Census (2022-23), though data for religion is said to have been enumerated separately, the outcome of such enumeration and whether appropriate internal classifications have been done remains unclear.

What of mixed or inter-caste identities?

A census is supposed to show demographic changes in the society where the census is conducted. It is a fact that inter-caste identities are not recognized as such in India. One can either belong to no caste or one specific verified caste. As per the G.O dated 3.10.2019[11] the Ministry of Social Justice and Empowerment has only taken a position pertaining to children of single/divorced/separated women belonging to Scheduled categories that such children may be granted the same certificate as their mothers. However, the protected status of inter-caste children in all other cases, where either parent belongs to a protected category and the caste certificate has not come from the father, continues to be decided in courts. In doing so, the stigma, allegations of ‘being an imposter’ (from within and outside the protected caste communities), as well as the fundamentally transformative identity of being inter-caste in a ‘caste-as-bloodline’ world that must all be faced by said children are not considerations in the courts. For, ultimately the courts are only concerned with classifying the child into one or other caste. The policy failure is also seen in that governments offer fiduciary protection to inter-caste marriages, but in a society that commits honour killings of inter-caste children, no such protections exist for the children of inter-caste marriages.

Insofar as no census since 1951 has included a codification for mixed, inter-caste, inter-tribe and inter-faith identities, we have no way of knowing the true extent to which caste, tribe and religious identities have coalesced, separated or disintegrated through the systems of inter-marriage and reproduction. Nor how many individual persons claim mixed, inter-caste identities, how many among them have been able to access any state protections, and how many among them have faced a history of abandonment. This is particularly intriguing given that during the same period data on migration has supposedly been collected. It is difficult to presume that growing inter-state migration has not affected inter-caste marital and sexual relations. One may argue that we ‘already know’ through sample surveys and anecdotal data that inter-caste marriages are woefully limited. But can such data replace the actual count of crores of people? Or, is it possible that an actual count of mixed, inter-caste offsprings might challenge the conventional wisdom of a more-or-less fixed communal boundaries over the past seven decades?

The partial data released around the Karnataka census has created some hyperbolic concerns of ‘dying’ populations among backward classes and ‘extinction’ of traditional occupations[12]. In the absence of comparable data as to the historical population of those communities it would be premature to claim ‘extinction’. And in the context of ‘Indianisation’ and forced extinguishing of scientific thinking, the push for ‘protecting traditional occupations’ on the basis of unverified data must be approached with caution.

None of the caste censuses conducted by the states have included a category for declaring mixed, inter-caste origin or identity. The Telangana Census questionnaire, however, did include a penultimate question on whether any individual in the family has had an inter-caste marriage.

At this point, it becomes important to make a few comparisons.

It is presumed that in a free and independent country census data is collected based on self-declaration, with the prior and complete knowledge of methodology, questions, system of classifications, time-line, space for revisions of methodology, appraisal of grievances etc given to the public that is consenting to be counted. The All-India Census is bound by the Census Act 1948 (and Rules) as per which at least the schedules and questionnaires are to be published in the Gazette. However, no person has the right to inspect books, registers or records of the Census. Nor is there any clear provision for independent monitoring of data collection in politically sensitive regions or situations. Insofar as there is no constitutional or legislative framework binding the state-level censuses, it is unsurprising that the sudden politically-charged release of data has raised posthumous methodological questions.

A first comparison here would be to census in Apartheid South Africa. Khafani & Zuberi (2001)[13] speak of the four broad racial classifications (African, Asian, Coloured, European/White) and the various sub-classifications that were defined and re-defined over time. In the Apartheid period, the enumerators were required to use crass physical anthropology to conduct the classifications, which, given the fundamental irrationality of a purely biological/physical race classification and the reality of miscegenation, was “confounding” and “complicated” for the enumerators. Through the period, various races were wrongly counted or poorly classified, and often merged with their ethnic identity. From 1918-1941, much like what has happened in ‘independent’ India, census was mostly done for European population. Most notably, in certain years the State was also known to reclassify the data after enumerator’s classification.

To quote Khafani & Zuberi (2001), specifically about the 1936 Census in South Africa: “Here, the state gave populations the illusion of self-classification, but did so only to appease those involved. In the end, the state imposed its own racialised ideas on the enumeration process and rejected the ideas of the enumerated. Adding to the potential confusion in the classification of these individuals is the stated fact that someone other than the enumerator who may be once or twice removed from the process and who definitely had no direct contact with the enumerated might complete this change in classification.”

During this period in Apartheid South Africa, one must remember that the Population Registration Act was also operational – which might be seen as dangerously close to how the National Population Register and its follow-up National Register of Citizens may come to be used through the 2025 Census in India. Therefore, it is surprising that those who participated in the anti-CAA, NPR, NRC protests are today coming together seeking a caste census from the same governmental apparatus.

Ultimately, in the first post-Apartheid census, even though the four broad classifications were retained from the Apartheid period, the underlying classifications and definitions are said to have been made more dynamic, with almost 93% of the population being enumerated.

In neighbouring Nepal, only two caste censuses have been undertaken since 2011. Here, one of the reported concerns appears that over self-declaration, the popular understanding of surnames played a significant role in identification of castes leading to concerns of miscounting and undercounting of castes.

What of census data coming from colonizer countries like the UK and US? The Census website of the USA (that may well disappear under the current government) provides a clear historical picture of racial and ethnic classifications and changes in the enumeration method and questions from 1790 to 2010[14]. This data shows how, depending on the statutory position of each racial group, the early ‘biological’ definition of race, taxed/non-taxed status of indigenous/aboriginal/native groups, migration patterns of various ethnic groups, change in terminology for race groups, growing sub-classifications and so on changes were made in the classification and data collection method. Notably, in 1960, enumerator reporting was replaced with self-declaration, and from 2000, it became possible to record more than one race to declare mixed/bi racial identities.

In the UK, data has been collected under the framework only of ‘ethnicity’[15] and only since the 1970s. This is directly related to the nature of inflow of ‘foreigners’ from its colonies: i.e. White, Asian and African colonies. In reality however, it continues to be a mix of race and ethnicity data (but not caste). From 1960s to 1980s, the data was collected informally and as per the government’s own admission through a deeply assumptive process left almost entirely to the enumerators. From 1991, they introduced a formal query for self-declaration of ethnicity (including sub-categories) with the option of declaring mixed ancestry.

In sum, in its intent, method and evolution at the central and state-levels the census exercises post-independence in India can be closely equated to whatever was happening in apartheid South Africa. A stagnant and unmonitored process that has failed the test of democratic evolution and transparent study of demography.

This finding can be easily established by a simple comparison of the publicly available Wikipedia ‘demographics’ data for each of the countries referred to here with that of India. While all other countries from a particular point in time can show (with degrees of error) demographic changes across race and ethnicity (recognized as their primary axis of discrimination), India shows data on decadal religious composition, and partial caste data for the singular year of 1983, citing the Mandal Commission Report.

Disability data

A popular opinion will be that since the state-level caste censuses have been undertaken with the reasoning of ‘uplift’ of disadvantaged populations it may be allowed that existing communal classifications operating at policy levels are used as is. This is not at all a viable opinion. When it becomes difficult to firmly make out the state’s intent or position, change the primary axis of disadvantage and see if it suffers all the faults there too. It will help to move focus away from communal classification to a less politically fraught but equally dated and unscientific classification system, i.e. disability. Here too, we see the absence of the right to informed self-reporting.

It is not surprising that the ‘intersectional’ debate is habituated to finding one human body upon whom to mark every social and physical disability known to us and present that person as an exemplar till exhaustion. As such, the rights of those disabled persons who do not clearly fall into the statutorily notified physical and mental disabilities (narrowly defined by specific percentages of disability), and their access to education and employment are also being decided in the courts for the past decade or so.

The patchy collection of data on disability in the national census has already been noted above. As per the 2011 All India SECC, we are required to believe that 98.46% of the rural population have no disability, and 98.92% of the urban population have no disability. In the 2011 SECC, there was an additional question on chronic illness (sub-classified only as Cancer, TB, Leprosy, Other) for the urban population, according to which 99.25% of Indian urban population have no chronic illness.

The Telangana Caste Census poses only one question on whether or not a member of the household has a disability. Details on chronic diseases have been sought only as part of occupational diseases. The data under both these heads is not part of what has so far been released to the public.

The Bihar Caste Census (2022-23) does not appear to have collected any data on disability. The same may be true for Karnataka as well.

As with everything, it is not always necessary that disability data be secured only through the census. Most countries collect health and disability data from multiple sources, only one of which is the census. However, having not maintained a sustained mechanism for disability reporting in the census, India has now reflected a desire to erase disability data from other sources going forward. The National Family Health Survey 5 (2019-21) became the first and last national health survey to specifically include questions on disability. The full set of data on disability has apparently not been released, and no data on disability will be collected from NFHS 6[16].

Counting transgender people

Globally, transgender persons have remained outside the purview of census counts, health and family surveys and disability records. In 2021, Canada[17] became the first country to attempt thorough counting of trans and non-binary persons in its census. This was done by reforming the census questionnaire through public consultation to separate sex (at birth) and gender identity. In the same year, UK’s census too sought information on trans identities in their census, but after significant criticisms as well as the political turn towards accepting only binary gender, in 2024 the UK’s Office for Statistics Regulation declared that the 2021 statistics on gender are no longer accredited statistics[18]. The 2022 South African census was criticized[19] for discriminating against trans, intersex and queer persons.

As with everything else in India, transgender persons are counted and not counted. The entire classification system is meaningless to our cause.

As per the All India SECC (2011), the rural household data reveals 75008 transgender households. It is unclear, from the manner in which the data is presented, whether this refers to transgender individuals or entire households. For the urban count, the data claims there are only 30314 transgender persons.

There is no doubt that the data suffers from significant undercounting and miscounting. There are contradicting views on how data was collected in 2011. Some say transgender data was counted under the male category and some say that enumerators were instructed to collect data under the ‘other’ column. What was followed in the Census and what in the SECC, it is not known nor clarified from their questionnaires. It may be presumed that transgender persons were informally counted based predominantly on appearances and then placed in the ‘other’ column. The 2011 Census (not the SECC) is supposed to have counted transgender population at 4,87,803. Quite bizarrely that census also counted transgender children (0-6 age), which would in fact be reflecting intersex children. In any case, with the 2011 method as a base, it has become almost universally repeated across various data sets that transgender persons make up less than 1% of the population, with some state-wise variations.

It would be fair to conclude that the Karnataka Census has not collected any data specifically on transgender persons seeing as the data collection was done in the year 2014-15 and within the mandate of Backward Classes Commission.

As per the data released in media for Telangana Census, the gender break-up of data is 50.51% (male) and 49.45% (female); leaving transgender persons with 0.04%.

For the Bihar Census, 82836 is the total count of persons under the ‘transgender etc’ column. This makes up only ~0.06% of the total population counted (13,07,25,310).

But that is an incomplete picture. Insofar as caste census is being discussed, the Central and State lists include jogtin as caste group (including both SCs and OBCs) in the OBC category. The jogti/jogta caste group includes gender non-conforming (by force or choice) transfeminine persons sexually subjugated and forcibly dedicated to temple deities. So, should we say that transgender persons have been counted or not in the caste census? This is definitely not the kind of question for which ‘yes’ and ‘no’ can be correct answers.

It is not that a caste census will strengthen or weaken caste divisions in the country. It is that a census without scientific and democratic organization of classifications will worsen the irrationality, fear and forced disappearance of many complex identities and peoples. And ‘counting’ is very different from appropriate classification, which in turn is vastly different from scientific classifications tempered with understanding.

Transgender persons, in the Indian subcontinent would have to be counted separately as transgender households and transgender individuals. This is different from the conventional wisdom on how households and within that gender is counted, where certain households are marked as women-headed households. For a composite picture of transgender communities in India (and a significant shift in method and policy itself) transgender households would have to be recognized. Such households would have to be counted for internal variations in caste, faith, education, employment, disability and migration, among other things. At the same time, individual self-declaration of trans identities across all households must also continue. However, in the absence of privacy of individual respondents the data will likely be recorded per account of key respondent in the family. The classification of household/individual would have to go hand-in-hand with sub-categorizations of gender identity. None of this can be expected to happen in the near future.

To press the point, it will be helpful to look at certain other types of data released by the Ministry of Statistics and Programme Implementation (MoSPI), the NFHS and the All-India Survey on Higher Education (AISHE)[20].

The NSSO releases data on monthly per capita consumption expenditure (since 1950) of households, which is categorized across social groups (broadly as SC, ST, OBC, other), geography (rural, urban), employment type and types of consumption (food, non-food). This data is essential to understand the consumption burdens, change in consumption patterns and access to nutrition across time and social groups. The latest survey released data for 2023-24[21]. As per this report, the questionnaire duly provides the option of recording ‘transgender’ in the gender column. But this data is nowhere to be found in the final record. This could be because it was not recorded or the final numbers were seen as inconsequential to inform the otherwise large data sets used for analysis. Or it was impossible to utilize data on consumption patterns of transgender persons without altering the existing organization of data.

The NFHS 5 (2019-21) offers data on social groups under various chapter heads, and it also includes the option of ‘transgender’ in its household survey questionnaire. And yet, the final report does not provide a significant view of the health and well-being of transgender individuals. This is because the NFHS surveys are conducted through four questionnaires: the household survey, the woman’s survey, the man’s survey and the biomarker questionnaire. Therefore, the only tabulation clearly reflecting data on transgender persons is shown in the gender disaggregation of disability data collected through the household survey. In the context of disability data the grossly unequal spread of the base numbers surveyed becomes clear. Where the base numbers for total number of women and men reporting disability stood in ten+ lakhs, the base number for transgender persons reporting disability was just 153.

The latest AISHE report is for the period 2021-22. This analyses data provided by higher education institutions. It provides broad strokes information on social group and male-female break-up of student enrolment, teaching and non-teaching staff. Obviously, there is no data on transgender participation in higher education.

The MoSPI also releases periodic data and indicators on unemployment, the labour force and something called ‘women and men in India’. These data sets and indicators are entirely organized around the rural/urban and man/woman matrix – erasing social groups and of course transgender identities.

The final question we would have to ask here is this: do the caste census details emerging from various quarters provide any startlingly new information that is not already established through the various national surveys and commission data on employment, health and education? And where startling information appears, is it backed by verifiable data?

Monitoring administrative classifications

It is clear, therefore, that this belief that a ‘caste census’ will provide a solution for all the policy problems and instantly reform the teetering reservation system in the country is grossly misplaced. It is important to remember that provision under Article 340 of the Indian Constitution is not towards conducting census but towards independent inquiry commissions away from political clutches. While the fuel of caste census keeps burning, it becomes important to look at how the Centre has been silently organizing or reorganizing administrative data, data storage/access and classification systems. Alongside the Data Protection Act, the Centre has in 2024 formulated a National Metadata Structure (NMDS)[22] which will likely govern how and what part of essential statistics (metadata) will be accessible. In 2021, the MoSPI released another weirdly titled paper “Administrative Data: Issues, Concerns and Prospects – An Indian Perspective”[23]. The entire paper is strewn with positions on handling administrative data from Western writers. The concerning portion of this paper appears at the end, where they are hinting at the possibility of ‘alternative sources of data for official statistics.’ Without a clear definition and explanation, the phrase ‘alternative sources of data’ could come to mean anything.

Therefore, it would do us all well to shift focus ever so urgently away from the census to studying the existing classification systems and mechanisms and making them just, rational and scientific. For, what we are suffering at present is a result of decade-upon-decade of criminally irrational classifications. Per example, on the one hand the caste and class reservation classification have been widely debated and yet we do not a historical view of the definition/basis for each classification, the inclusion and exclusion of various caste groups with recorded reason, collated and updated with time. On the other hand, by 2024 the MoSPI has published its ‘List of National Statistical Classification for usage in official statistics’[24] which includes classification systems operating since independence, and does not make reference to socio-economic classification systems including caste, tribe class, gender under the broad head of Demographic Statistics. Bringing us a full circle to part constitution, part French delusion and zero democratic evolution.

Some of the transgender political positions world over have always taken an anarchist view and said no to ever being counted – ‘we are ungovernable, outside all classifications’ being the position. While gaining psychological strength from that position, it remains extremely important to govern those who are governing and demand one’s share in the resources of the State.

[1] The Bihar report is entirely in Hindi and as such inaccessible to me. I thank Atul Anand for responding to my queries and providing me information and data from that report. Any errors in understanding are mine.

[2] See: https://thewire.in/politics/telangana-caste-survey-what-it-says-what-critics-have-said

[3] See: https://www.thenewsminute.com/telangana/telangana-caste-census-raises-serious-questions-is-bc-population-being-undercounted

[4] https://censusindia.gov.in/census.website/CENSUS_ques

[5] See: https://www.theguardian.com/world/2020/jun/16/france-and-germany-urged-to-rethink-reluctance-to-gather-ethnicity-data

[6] See: Khalfani, AK and Tukufu Zuberi. Racial classification and the modern census in South Africa, 1911-1996. Race & Society 4 (2001), pp. 161-76

[7] See: https://www.himalmag.com/politics/caste-census-india-hindutva-dalit-south-asia

[8] https://secc.gov.in/homepage.htm

[9] https://seeepcsurvey.cgg.gov.in/

[10] For a brief discussion on Caste and Muslim classifications in Telangana see: https://www.siasat.com/caste-differences-among-muslims-in-telangana-and-their-status-2365207/

[11] MOSJE Government Order dt 3.10.2019. https://socialjustice.gov.in/public/ckeditor/upload/98661672891374.pdf

[12] See: https://www.deccanherald.com/india/karnataka/caste-census-finds-21-communities-with-fewer-than-100-members-in-karnataka-3493748

[13] Full citation at footnote 5

[14] https://www.census.gov/data-tools/demo/race/MREAD_1790_2010.html

[15] https://history.blog.gov.uk/2019/03/07/50-years-of-collecting-ethnicity-data/

[16] See: https://idronline.org/article/diversity-inclusion/dropping-disability-questions-from-nfhs-6-is-a-mistake/

[17] See: https://www150.statcan.gc.ca/n1/daily-quotidien/220427/dq220427b-eng.htm

[18] See: https://statistics.ukdataservice.ac.uk/dataset/england-and-wales-census-2021-gender-identity-by-age-and-sex-4-categories