Zero-Eighteen-Zero: the story of "ethnic groups"


  • Author: Chris Smaje
  • Date: 28 Feb 2013
  • Copyright: Image appears courtesy of iStock Photo.

Here’s a trivia question for social statisticians: what were there none of in 1981, nine of in 1991, sixteen of in 2001, and eighteen of in 2011?

The answer is official ethnic groups in the UK, as recognised in the country’s decennial population census. In 1991, the predecessor of the UK’s Office For National Statistics introduced an ethnic question into its census for the first time. The question asked respondents to indicate their ‘Ethnic Group’ from a list of seven predefined categories, and two additional write-in ones. How the nation responded is indicated in the table. How a social statistician might respond to the question itself and to the subsequent history of ethnic group questions in the UK census is the subject of this article.

thumbnail image: Zero-Eighteen-Zero: the story of

Ninety four per cent of the population deemed themselves to be ‘White’ in 1991, and of the remaining six per cent, over a fifth allocated themselves to one of the ‘Other’ categories – including nearly ten per cent in the residual ‘Other – Other’ group. On the face of it, a social statistician might baulk at employing a classification that allocates virtually everybody to one category and then fails to capture adequately a large proportion of the remainder within another five categories. But of course the geographic distribution of ethnic groups is highly skewed. Inasmuch as census data is mainly used by local authorities in relation to their own populations, the census categories may be more revealing at the local level. Or to put it another way, sensitivity to what is sometimes called the ‘ecological’ character of the data is required – what’s true at a higher level of aggregation may not be true at a more disaggregated level. Even so, the proportion of people self-identifying as ‘White’ drops below fifty per cent in only a handful of inner city boroughs nationally.

In the two subsequent censuses of 2001 and 2011, the ethnic question was further refined, as shown in the table. Additional ‘White’ categories were added – including ‘Irish’ and ‘Gypsy or Irish Traveller’ – and a whole series of ‘Mixed’ categories were introduced. By 2011 the ethnic question was looking quite unwieldy in comparison to its slimline counterpart of 1991, with thirteen predefined categories and an additional five write-in ones, plus completely separate questions on ‘National identity’ and religion. Yet the proportion of ‘non-white’ respondents allocating themselves to ‘other’ categories still exceeded a fifth, and the residual ‘Other – Other’ group still accounted for over four per cent. Looking across the twenty years from 1991-2011, it’s difficult to untease the effects of demographic change from the artefactual effects of the different ethnic classifications, but perhaps one might conclude that ethnicity resists enumeration.

Taking a step back from the numbers to look at ethnic definitions themselves, the survey statistician must always be thinking about the question of validity – does my survey question actually measure what it’s purporting to measure? With much social science, and above all in the field of ethnic surveys, there is an awkward circularity to negotiate. Ethnic distinctions, conditioned by various histories of racism, nationalism and colonial or postcolonial labour relations, emerge as categories of everyday social description, which the social scientist or the policymaker must then more or less unwillingly adopt in order to try to characterise people’s experience as the objects of these categorisations.

Table 1: 'Ethnic Group' in England and Wales, Censuses 1991-2011

1991 (% total)

2001 (% total)

2011 (% total)

White (94)

White British (87.5)

White English/Welsh/Scottish/Northern Irish (80.5)

White Irish (1.2)

White Irish (0.9)

Other White (2.6)

Gypsy or Irish Traveller (0.1)

Other White (4.4)

Indian (1.7)

Indian (2)

Indian (2.5)

Pakistani (0.9)

Pakistani (1.4)

Pakistani (2)

Bangladeshi (0.3)

Bangladeshi (0.5)

Bangladeshi (0.8)

Chinese (0.3)

Chinese (0.4)

Chinese (0.7)

Other Asian¹ (0.4)

Other Asian (0.5)

Other Asian (1.5)

Black Caribbean (1)

Black Caribbean (1.1)

Black Caribbean (1.1)

Black African (0.4)

Black African (0.9)

Black African (1.8)

Other Black (0.4)

Other Black (0.2)

Other Black (0.5)

Mixed White & Black Caribbean (0.5)

Mixed White & Black Caribbean (0.8)

Mixed White & Black African (0.2)

Mixed White & Black African (0.3)

Mixed White & Asian (0.4)

Mixed White & Asian (0.6)

Other Mixed (0.3)

Other Mixed (0.5)

Arab (0.4)

Other - Other (0.6)

Other - Other (0.4)

Other - Other (0.6)

Total N = 50.9 million

Total N = 52.0 million

Total N = 56.1 million

The critical sociologist can have a lot of sport with the unhappy compromises that result. Is the census’s ‘ethnic group’ a matter of skin colour or ‘race’ (‘White – Black’), nationality (‘Indian’), global region (‘Arab’/‘African’), or some notion of inherited purity (‘Mixed’)? Is Jewish an ethnic group? Are Turkish people white, black or Asian? What is the ethnic difference between Northern Irish and Irish? Does religion constitute ethnicity in Ireland? Or anywhere? Must ‘Travellers’ actually be travelling, an issue that has notoriously arisen in court cases where the ethnic status of gypsies is questioned on the grounds that they no longer have nomadic life habits. And so on. All of these may be important questions in particular contexts, but the hard pressed public sector manager may consider them to be rather academic ones when all s/he really needs to track is the ethnic patterning in service use, such as whether the local Pakistani women are getting adequate access to family planning services. In this instance, s/he just wants the relevant number from local census data to slot into the denominator on the spreadsheet.

But people can have finely graded and contextual senses of personal identity that quantitative surveys can never hope to capture. The same woman at the clinic may consider herself to be Pakistani, or Asian, or British, or English, or Black, or Punjabi, or Jat, or Muslim, or ‘Mixed’, at different times or in interaction with different people, or perhaps all of them simultaneously. Or she might resist such designations altogether. The trajectory of the census questions from 1991 to 2011 seems increasingly to attempt to capture something of those finely graded, self-ascribed identities, but tick box surveys will always be chasing after shadows in this area.

Perhaps we must also question what, ultimately, we mean by ‘ethnic patterning’. The public sector manager could test a multivariate model of service use, looking at age, gender, socioeconomic status and other such variables as well as ‘ethnic group’, however defined. She might find that those other factors are associated with most of the variation in outcomes, with her ethnic variable returning as non-significant in her model. Does that mean that ‘ethnic group’ is not a relevant dimension of social experience in contemporary Britain? That depends on whether you think policymakers should be addressing the patterning of outcomes by the more or less well understood categories of everyday experience such as ethnic group, or dissolving them away in theories of socio-historical causation.

There is one final twist in the tale of bureaucratic ethnic categories worth exploring. In ‘The hyperreal Indian’, a wonderful paper published nearly twenty years ago, Alcida Ramos examined the way that such categories can actively come to organise social behaviour, rather than just passively recording the categories of the world ‘out there’². Through an admirable concern for the wellbeing of indigenous Indians, Brazilian NGOs necessarily engage in definitional exercises to identify the ‘real Indians’ who are the focus of their work. But of course many people have some more or less plausible claim to Indian origins – inasmuch as bureaucratic categorisation sorts out the ‘reality’ of such claims it can incentivise people to self-define and change their behaviour accordingly, while imparting an ultimately indefensible notion of ethnic purity into local social identification. Bureaucracies need to know whether you’re an Indian or not. They don’t deal well with the lived world that all of us actually inhabit, where we might be inclined to answer not sure, maybe, that depends or who’s asking?

This kind of administrative constitution of social identities is quite widespread globally – the debate over the ‘Other Backward Castes’ in India is another example. It’s probably unavoidable in modern state bureaucracies, but it fuels all sorts of ultimately quite absurd local conflicts and jockeying for position. Perhaps ethnic categorisation in the UK doesn’t furnish us with any examples of bureaucratic ethnogenesis that are quite so stark (though the ‘mixed’ categories of 2001-11 raise a lot of questions), but we’re all increasingly inured to partitioning ourselves out in endless surveys and official forms enquiring into our gender, sexuality, ethnicity, disability and so on. The motives are sound, but perhaps there’s a danger that official ethnic categories and their cognates will become an example of what the anthropologist Daniel Miller calls ‘virtualism’ – reified and routinised products of past social identifications that increasingly come to condition the manner in which we can shape new ones in the formal worlds of work and public life.

There’s a chance that there won’t be any more national censuses in the UK, and personally I could be persuaded that that is either a cause for celebration or regret. At least it would make for a pleasing circularity in trivia questions – what started from nothing in 1981, grew to eighteen by 2011, and then dwindled to nothing once more by 2021?

1. Calculated from the 'Other' category.
2. Ramos, A. (1994) ‘The hyperreal Indian’, Critique of Anthropology, 14, 2: 153-71.

Related Topics

Related Publications

Related Content


This website is provided by John Wiley & Sons Limited, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ (Company No: 00641132, VAT No: 376766987)

Published features on are checked for statistical accuracy by a panel from the European Network for Business and Industrial Statistics (ENBIS)   to whom Wiley and express their gratitude. This panel are: Ron Kenett, David Steinberg, Shirley Coleman, Irena Ograjenšek, Fabrizio Ruggeri, Rainer Göb, Philippe Castagliola, Xavier Tort-Martorell, Bart De Ketelaere, Antonio Pievatolo, Martina Vandebroek and Lance Mitchell.