The data derived from the national census is one of the mainstays of discussions of integration. The distribution of citizens by residence, profession and income according to ethnicity and religion provides easily accessible metrics and prompts calls for government activity where these metrics suggest discrimination.
In its famous 1997 report on Islamophobia, the Runnymede Trust also argued that census categories ‘give many people a sense that an important part of their identity is being recognized and respected’.
However, I think we also need to recognize that the census does not only measure categories, it also constitutes them. By including asking a citizen body if they are black or white, or Hindu, Jewish or Muslim, it implies that these are the forms of identity that it will recognize, and that discrimination and fair-dealing will be measured in these terms. Thus the recent disclosure of the salaries of highly paid BBC presenters was discussed in terms of gender and ethnicity, but not in terms of the social origins of their parents, their place of birth or their education, which might be other ways of considering representation and inclusion.
The division of the population according ethnicity and religion influences the allocation of resources by the state. For instance, the agitation by some British Kashmiris in the 1990s for recognition on the census was based on the presumption that they were more discriminated against than (other) Pakistanis. They argued that the identification of large Pakistani populations was ‘rewarded’ with Urdu language services and multicultural education that discussed Pakistan.
Instead they hoped for a reclassification of many of these ‘Pakistanis’ to re-direct resources towards teaching Pahari and the history of Kashmir. Of course, if these efforts had been successful then further advantages would have accrued to Kashmiri activists, to represent an ethnic minority to the British state; to stimulate and direct the charitable donations of Kashmiris and possibly to influence British foreign policy as well.
The logic that we allocate resources where identity groups are concentrated has several concealed dangers. One is that it encourages segregation: if the only way to ensure that my children are taught ‘my religion’ and ‘my culture’ is for me to live alongside enough like-minded people to constitute a significant group then this will influence my choice of residence. An example of this is the religious education syllabus, which is determined at a local level dependent on local demographics (seen through the categories provided by the census).
Another danger is that the census encourages the homogenizing of identity. If group rights depend on numbers, this encourages group members to take steps to prevent other members from leaving the group or expressing forms of membership that undermine the identity of the majority.
For instance, the sustained focus on ‘Islam’ as the key definer of the social identity of Muslims has often ossified religious culture and impoverished the other parts of the cultural heritage of citizens of Muslim heritage. If rights are given to those with the greatest numbers, who speak with a united voice, then this means that people who might have emphasized their identities as Bengalis or socialists in the past are encouraged (by the government and by their peers) to present themselves as Muslims.
It also means that when spokesmen are acknowledged for ‘Muslims’ as a whole these are likely to be clerical leaders of majority traditions: the singular categories of the census offers no space for Sufi or Shia voices among a Sunni majority, let alone for cultural Muslims who are only offered a choice between ‘Muslim’ and ‘No religion’.
We should remember that the census has a long history as a tool of colonial governance in British India. While there was no religious category in the nineteenth century domestic census, it was considered an appropriate instrument of government for the colonies. India’s diverse and overlapping religious traditions were reduced to mutually exclusive blocs, whose numerical strength in a locality was used to determine quotas at government colleges, language policy and personal law.
The categories of the census were used to identify Indians as Muslims or Hindus whenever they registered to vote, attended hospital or paid taxes. When they started to be used these categories were not accurate descriptors of the population (several villages in rural Punjab did not know whether they were Hindu or Muslim when asked by British administrators). But they did set up expectations that eventually form reality, through the provision of different forms of personal law and by presenting clear signs from the state that religious identity trumped other forms of identity.
In British India, the census served as a tool for severing bonds between Indians, and making them more suspicious of one another than of their rulers. This made it an effective tool for diverting Indians from recognizing shared class interests or criticizing government. I presume that the Runnymede trust did not have this history at the forefront of its mind when it invoked India as a positive example of the inclusion of a religious question on the census.
A prime danger of censuses is that they will automatically be compared to previous censuses. Implicitly, this comparison provides a history of demographic competition, in which size and concentration leads to a claim for resources. It is easy to see how such histories unwittingly provide the basis for national anxieties about an enemy within. There is little room for a narrative that allows for the assimilation or integration of outsiders, no way of measuring the changes in trust or compromises in custom.
Instead the census is used to map a zero-sum game for cultural and economic resources, in which, a scaremonger might argue, the ‘white British’ are doomed to lose. As Eric Kaufmann has suggested, one key intervention that the integration debate requires is a shift in narrative, which allows a ‘majority’ to understand minorities as populations open to cultural change.
Of course, the data provided by the census to social scientists would be hard to replace. But it is worth noting that researchers who do without national ethnic and religious census categories are often forced to frame their questions better than those who do have this apparent glut of data.
One important example is the work of Michèle Tribalat, who works on French demographics and has no ethnic or religious census data to work with. Her own interpretative categories dig rather deeper than simply asking about religion, for instance. Instead she asks about adherence to dietary taboos, prayer, language use and marriage practice. This allows her to make much tighter analytical connections: instead of being content to use ‘Muslim’ as an analytical category, she is able to identify specific practices that correlate to specific social outcomes.
It is this kind of fine grain that is often missing from the British integration debates: the easy availability of the census data may have meant that we have been framing questions more poorly, or confusing correlation with causation, as well as adversely affecting the process of integration itself.