Radical
Statistics

The Journal

The Subjects

The Books

News

Links

About

Home

‘Race’ Statistics: Their’s and Our’s

Ludi Simpson

Official statistics are part of the machinery of state and are challenged by movements that demand change in government policies. But statistics have also been used to expose the results of racism and to challenge the assumptions of government ministers. This contribution attempts to shed some light on the contradictory roles of statistics about race.

Statistical practice generally attempts to understand society through personal data about individuals, and leads attention away from direct study of institutions and the state. The development of statistics as a scientific discipline one hundred years ago has a direct link with racist ideology; the measurement of correlation between personal characteristics was developed in response to the Eugenic Society’s concerns in Britain that European genetic stock, assumed to be superior, was weakened by classes and races whose genetic composition was considered to be inferior ⁽¹⁾. Eugenicists interpret disadvantage as genetic inferiority and seek to deter breeding of the disadvantaged or to protect breeding of the most advantaged, in the same way as dogs and horses are bred for prowess in running. The eugenic perspective was weakened with the defeat of its most successful proponents: Nazi fascism. But it emerged again under the academic guise of socio-biology and within psychology by the assignment of a genetic factor to measures of intelligence ⁽²⁾.

The very measurement of racial stratification can be traced to racial slavery and to discriminatory legislation in the United States. Differences between measured ‘races’ justified, assumed and hid an economic relationship between classes. More recently and in less polemical fashion, academics from many disciplines use ‘ethnic group’, ‘ethnicity’, ‘race’ or ‘the ethnic minorities’ as terms of explanation for different experiences in health, employment, housing and living arrangements, migration and family building. Government frequently uses the results of this industrious effort not to address the causes of those differences, but to accommodate them within a multicultural framework.

It is therefore no surprise that anti-racists should not look to current statistical practice for solid support. The transformation of the Institute of Race Relations involved a rejection of the dispassionate academic statistical outlook. Its staff felt the urgent need to resolve "the conflicting aims of objective, unbiased research on the one hand and policy-oriented research on the other: shifting the focus of race research from black people to white institutions, from the done-to to the do-er, from the victim to the perpetrator" (Bourne, 2001).

But if statistical practice leaves something to be desired, can it provide support for change as well as for the status quo? If statistical practice, with its conceptual categories and instruments of measurement, is specific to the society it is found in, why not go on to ask what kind of statistics would arise in a different society? Or more practically, what kind of statistics should critical social movements demand now for their own purposes, or indeed collect and make themselves? Out of anti-racist politics, what demands come for statistics that are not yet collected, involving new conceptual categories and instruments?

This paper does not arrive at answers to all those questions, but it takes some steps along the way. It takes a look at, in turn, the political, historical and statistical views of official collection in the UK of data that are currently labelled ‘ethnic group’. The final section returns to measures and approaches that are most useful to anti-racist practice.

1.Government policy and UK official race statistics

(i) Anti-discrimination legislation

Race and ethnic group provide official measures of the results of discrimination and therefore of rights denied. Statistics are required and encouraged by government anti-discrimination legislation, which charges the Commission for Racial Equality to ensure that progress towards (or away from) equality is statistically monitored and gives it powers of investigation of suspected discrimination ⁽³⁾.

Some statistics tell a story that should inform public policy. The 91 deaths in the period Feb 2000 to April 2001 from racist attacks, or while in police custody, or to asylum-seekers each signal the end of a human life, the brutal cutting short of potential. There were successful prosecutions in only about half of these cases ⁽Athwal, 2001). Opinion surveys show a consistent feeling that racial discrimination in public services has worsened over time and is expected to continue to do so. Powerful descriptive statistics show that "Overall, people from minority ethnic communities are more likely than others to live in deprived areas and in unpopular and overcrowded housing. They are more likely to be poor and to be unemployed, regardless of their age, sex, qualifications and place of residence. As a group they are as well qualified as white people but some black and Asian groups do not do as well at school as others, and African-Caribbean pupils are disproportionately excluded from school." Such statistics undermine any claim that there is no problem for government to address ⁽⁴⁾.

Policies for equal employment demand reasonable targets, based on demographic analysis. Home Office statistics for 2001 show West Yorkshire police with 3% Black and Asian officers, and 8% Black and Asian population, highlighting that police recruitment does not reflect the policed population; the pattern is repeated in each of the major conurbations of Britain. In this case, the targets demanded by Anti-discriminatory policy involve short-term projections of the number of Black and Asian adults of working age. It also implies monitoring not just whether the targets are met, but how. Greater recruitment, perhaps from overseas as suggested by the London Metropolitan Police, won’t itself change the police behaviour observed by potential recruits and deterring their applications⁽⁵⁾ .

Statistical modelling take further the descriptive differences between groups in the population. It attempts to measure the ‘causal path’ of inequality by associating those differences with other variables. The high unemployment rate of black people in Britain has remained throughout the last 40 years; half of it is associated with characteristics such as place of residence and qualifications, so the rest must be mainly due to discriminatory practises (Twomey, 2001; Leslie et al, 2001). This statistical ‘explanation’ of differences by reference to other variables is common but should not be taken to diminish the importance of the ‘explained’ half of the inequality. People’s residence and qualifications and social class are also achieved after discriminatory processes in education, immigration, housing and employment, entrenched within Britain’s institutions⁽⁶⁾ .

To avoid these conceptual difficulties of statistical modelling, discrimination in the job market has been measured with experimental statistics (7). The results of experiments are powerful, but they are not so easy to achieve. No-one likes to be randomly allocated a treatment which they don’t like.

In the US where there is a longer history of anti-discriminatory legislation and associated statistics, community campaigners have defended those statistics when they have come under threat, as a means to improve services to those who had been excluded (Wyley, submitted 2001).

Thus it seems that race and ethnicity categories are useful to monitor and to expose oppression and its results. However, they are not so essential to plans to eradicate the causes of inequality. Statistics are not needed to implement policies to remove institutional racism, if the power and the will are there. But evidence of racism and its results helps to neutralise resistance to that power.

For any policy that demands change, the observed behaviour is intended not to be observed in the future. Successful anti-discriminatory policies may need race and ethnic group to be establish themselves, but demand change so that they need not be collected in the future.

(ii) Diversity and the protection of cultural space

Government policies for ‘multi-cultural Britain’ need to categorise people into separate cultural identities in order to understand and accommodate differences. Roy Jenkins’ formulation of racial integration in 1966 as ‘equal opportunity accompanied by cultural diversity in an atmosphere of mutual tolerance’ was acted on in the 1980s through a barrage of race awareness training and development of cultural polices in education, social services and other public arenas.

Under this government policy, statistics are permanently needed. Government resources are specified to ensure that each culture can express itself and be appreciated within regulatory limits, including extended support for faith schools. Cultural and religious organisations have supported and indeed led new arguments for measurement of both ethnic group and religion as official categories, because such recognition allows resources to be claimed. ‘Cultural background’ is the main label for the current categories of ‘ethnic group’ in the 2001 Census and in national surveys, which are discussed later on.

Multicultural policy expects that we all have a cultural background, a different priority from the focus on discriminated social groups, or the immigration focus on culturally non-European immigrants. It readily extends to Welsh and Scottish, to Jewish and to Latvian. Organisations that grab resources on the basis of measured community size would be only too pleased to welcome you into their category. In this theory, it is quite OK to have more than one cultural background, though in statistical practice it is rare to measure this richness of cultural origins.

(iii) Immigration policy

This third arm of government policy was subtly addressed by the Observer, whose headline declared ‘The last days of a white world’. Whites would be a minority in Britain at the end of this century, quoting a demographer who did not want to be named for fear of recrimination. Thus they couched their projection as a terrible hidden truth that brave people (and brave newspapers) must tell (Browne, 2000).

No such projection has ever been documented but the story parallels Enoch Powell’s political fear of rivers of blood that would flow from conflict between white and black workers in Britain, Thatcher’s fear of a swamped English culture, Blunkett’s schools swamped by refugee children, and the current government’s keenness to limit immigration and asylum in spite of the projected lack of labour within the UK economy.

This third context for race statistics is entirely to do with limiting immigration, though the aim is often phrased as good race relations. "Controlling immigration was necessary to improve race relations. The fewer the blacks, the easier their integration. No blacks, no problem: it was all a matter of numbers. Hitler said the same sort of thing" (Sivanandan, 1990). In this ‘numbers game’ the one rule is that the smallest number is best.

Race and ethnic group are now synonyms not for oppression nor culture, but for immigration, and it does not matter that the move was by the person concerned or by their parents or by earlier ancestors. Academics whose work stems from immigration policy talk of "immigrant populations" and sometimes "a permanent foreign population." These are meaningless terms given the mixed and migratory origins of us all when taken to the distant past. The terms and those who use them tend to use ‘foreign’ to mean foreign colour, rather than foreign nationality. But "foreign" is also a pernicious term to use for the 6% black and asian citizens of the UK, carrying a sense of permanent limitation of rights and settlement, which is contrary to the law of the land.

In this context it is reasonable to fear the use of personal records that include race or ethnic group directly for discriminatory purposes by a future government. Such records and censuses were used in apartheid South Africa and to identify areas for pogroms against Jews in Nazi Germany.

(iv) Community cohesion

Most recently, a series of Home Office reports have taken a swipe at multicultural policy (Cantle, 2001; Denham, 2001; Kundnani, 2002). Rather than celebrating and protecting cultural space, they complain of self-segregation of communities, label faith schools and all schools dominated by one cultural group as divisive, and demand greater integration. In the same way, the government has demanded an oath of allegiance to Britain and a sufficiency of English in the Immigration and Asylum Bill (The Home Office, 2002:schedule 1).

The government accepted the Home Office recommendation that all local authorities should monitor community cohesion, without prescribing how it might be measured. There is an implication of interaction, of sharing the same work-places, play places, and living spaces. Social cohesion in government statements also means lack of friction on the streets, but does not require either lack of racism or social equality. Social cohesion statistics of perception and integration are therefore very different from the statistics of discrimination. Integration as a goal flies in the face of multicultural space to self-organise. Nonetheless, as all are aspects of government policy, local authorities and government are expected to measure them all.

2.The history of categories of race and ethnic group in Britain

Official categories are represented in practice in government surveys and its ten-yearly census. Up to 1971, an individual’s own place of birth was the only means of understanding the mix of origins in Britain. It is categorised finely to reveal those born in each country of the world, and published tables from UK censuses divide the world into more than 20 countries and regions. But the low number of residents born outside Britain and their geographical dispersion within Britain meant that before 1981 tables of country of birth were not cross-tabulated with social conditions. Interest in the numbers of racial groups gained strength in the 1960s and 1970s as immigration became a strong political issue.

At the same time, black populations made their organised contribution in many city areas, and racism was resisted strongly by black organisations, stimulated by political traditions new to Britain including the Black Power movements of the Caribbean and the USA. Already all the policies discussed above were evident to some extent, but only immigration made it onto the government statistical agenda. Immigration policy and anti-racists alike focused on the Black population as a whole, and this was not to change until multi-cultural policies came to the fore in the 1980s.

In 1966, the government Chief Statistician John Boreham firmly rejected a question on ethnicity or race as impossible in a census:

It is extremely difficult to define ‘coloured’ precisely at all; and it is impossible to define it precisely in a way that will work in a census. We did not (and will not) attempt it, but used the unequivocal and objective concept of birthplace instead. This leaves you with the awkward ‘white Indians’ and ‘black Englishman’ but there is no practicable way of identifying them in a census ⁽⁸⁾.

Instead, the Census of 1971 extended the question on country of birth to each person’s parents, and this was the basis of government projections of the ‘New Commonwealth and Pakistan’ population: those born in or with parents born in the New Commonwealth or Pakistan, which was not a member of the Commonwealth in 1971 or 1981. The census data gave numbers and age structures in each part of the country, that corresponded very well with Britain’s black population. The government’s aim was simply a demographic profile of the black population, just as the political debate about the statistics continued to focus on whether the immigrant population had been exaggerated or under-counted, and informed the politics of black immigration (e.g. Peach, 1974).

The Race Relations Act of 1976 was much stronger than those of the 1960s, and came after street skirmishes and resistance of young Black and Asian people in the early 1970s. The Race Relations Act didn’t stave off the first significant race riots, in Liverpool and London in 1981, but by focusing on the rights and conditions of people in Britain it gave a purpose other than immigration policy for statistically monitoring the conditions of Britain’s black population or what were by then called ethnic minorities.

The government proposed a direct question on ethnic origin in the white Paper on the 1981 Census, not much more than a decade after the chief statistician had said it was impossible. "Authoritative and reliable information about the main ethnic minorities" was now needed.

In order to help in carrying out their responsibilities under the Race Relations Act, and in developing effective social policies, the Government and local authorities need to know how the family structure, housing, education, employment and unemployment of the ethnic minorities compare with the conditions in the population as a whole (HM Government, 1978).

So, it was now acknowledged that Black workers and their families were ‘Here to stay, here to fight’. And government, while turning the screws to restrict immigration also wanted to prepare the mechanisms of social control that would monitor and reduce the impact of the racist institutional practices of the British state and commerce. At this distance and accustomed to ethnic group data from the censuses of 1991 and 2001, researchers and statisticians might well assume that an ethnic group question was a welcome advance. In fact the tests for an ethnic origin question in the census were not successful, in particular the 1979 test in Haringey. They were held at a time when the ‘sus’ laws allowed police to pick up and pick on anyone in the street on suspicion that they might commit a crime, and they did so predominantly on black youth, and in a media atmosphere which gave credence to the repatriation politics of the far right. Many of the black respondents in tests of the census objected, and in spite of support from the newly established Commission for Racial Equality, the proposal did not go ahead for fear of jeopardising co-operation with the rest of the Census questions (9).

The British Society for Social Responsibility in Science, to which the Radical Statistics Group was affiliated, had published a broadsheet arguing very strongly against the asking of the question. It said that in the atmosphere of fear on immigration, repatriation and racist policing, with little evidence of commitment to prosecute racists, there was little purpose for the question. To help its readers empathise with the situation that blacks in Britain faced with a census question on race, it hypothesised a public debate on euthanasia at 65 and asked whether elderly people would be right to worry if there was a new census question on age (BSSRS, 1981).

So the 1981 Census did not include a question on ethnic origin. Nonetheless, in 1981 the social conditions of the black population were for the first time measured and tabulated in the census, and for the first time those data were also readily available on computers for each small area in Britain. The question on parent’s country of birth was dropped, but in a tabulation of over-crowding, shared bathroom and other housing facilities, and car ownership was given for all those in ‘New Commonwealth or Pakistani headed households’ those whose head of household was born in those countries (Table 36). Very few children of immigrants were yet old enough to have left home to head their own household, so the category ‘Head of household born in NCWP’ fitted the black population very well in the same way as parental country of birth had done in 1971. Each household decided for itself which of its members was to be recorded as the head of household.

The 1981 Census showed clearly that the black groups on average had considerably worse social and housing conditions than the population as a whole. Thus began the flourishing of an industry of ethnic and racial quantitative social research studies. At that time "NCWP" was still presented as one group, and in a separate table. Statistics of the whole population were given in other tables, and statistics of specific origins were not explicitly tabulated. Focus continued to be on the non-white population, not on the multicultural ethnicity that we all might have.

However the 1980s brought multi-cultural policies following the Scarman report in 1981 with resources for schemes and programmes directed towards black populations needs and employment. Now there was an advantage to measuring ethnic group, especially to those organisations and individuals who advanced within the schemes.

The 1991 Census and 2001 Census have both used a direct question on ethnic group. Though there are differences between these two censuses and the questions they used, we turn to the most recent census to see how the UK government now measures ‘ethnic group’.

3.The 2001 Census policy concepts in statistical practice

With three decades of research into question wording, with sociological and governmental studies into ethnicity and race over forty years, one would hope for some clarity of categorisation. But given the mixture of contradictory policies informing government’s need for the statistics, there is evident confusion in the ethnic group question for the 2001 census, reproduced in the illustration.

This is by far the longest of the 35 questions in the 2001 Census for each individual, taking a full A4 column for a single response to ‘What is your ethnic group?’

There are no explanatory notes. It is assumed that each person in Britain will agree sufficiently what is intended by ‘cultural background’ for the answers to be meaningful. The available categories are hierarchically structured, with a first level marked by five bold headings that equate to colour or race:

‘White’, ‘Mixed’, ‘Asian or Asian British’, ‘Black or Black British’, and ‘Chinese or other ethnic group’.

Under each of these five headings are sub-categories which indicate a country or region of origin:

British, Irish, Indian, Pakistani, Bangladeshi, Caribbean, African, Chinese, and opportunity to write in an answer should these labelled categories not be acceptable.

The mixed purposes and policies that motivate the question and its general acceptance are evident in several ways (10).

1. ‘British’ is used as part of the Asian and Black main headings, but a final category of ‘British’ is only allowed under the main heading ‘White’. On purely technical grounds, this is a mistake, confirmed by enumerators reporting that some Asian respondents had ticked ‘British’ having seen it as the first box and wishing to confirm their British identity and nationality (11). In response to the demand in Wales for a ‘Welsh’ category, government and the Commission for Racial Equality now recommend use of the 2001 Census question expanded to include ‘Welsh, ‘Scottish’ and ‘English’ in addition to ‘British’ and ‘Irish’, but again only under the ‘White’ heading. The use of cultural background with ‘British’, a term that most will clearly relate to nationality, is a weak operational approach (CRE, 2002) (12).
2. The lack of notes emphasises the government’s view that "the ethnic group which each person chooses as his or her own is intrinsically the ethnic group of self-identity, rather than being ascribed by anyone else" (Haskey, 2000, p.37). But since the categories are pre-defined it is not self-identity at all except in the trivial sense of self-completion of the form. The question elicits the individual’s response to an official set of categories, the official cognitive system which compels respondents to allocate themselves to a particular group whether they like it or not.
3. The focus on detailed differences among categories other than White continues the idea of a foreign population, from immigration politics. It is reinforced when the academic who convened discussions leading up to the 2001 Census classification) writes about "people from ethnic groups" (Aspinall, Ethnicity.., 2000). Which people are not from an ethnic group?
4. Culture is mixed with descent. Peter Aspinall has useful reviews of these issues, pointing out for example that "The option for the Irish-descended population in which ‘Irish’ is set against ‘British’ will not enable the full size of the ‘Irish’ cultural background group to be captured". He expects that those with Northern Ireland protestant backgrounds may especially choose British over Irish (Aspinall, Policy, 2000). His idea of Irish includes both Protestants and Catholics, but worries that some of these Irish may not agree with him! For other writers the terms ‘descent’ and ‘origin’ intend to capture a more constant, less socially constructed demographic variable than cultural background (Berthoud, 1998). But ‘descent’ is a difficult concept to measure for anyone whose known origins cannot be traced to a single country, and that includes most people who care to think about it. The British Household Panel Survey like many others mixes its labels to acknowledge the mix of concepts in its measurement, using ‘RACE’ as its name for ethnic group categories from the Census.
5. The category of ‘Mixed’, new for the 2001 Census, is intended to reflect demand from those who wish to identify in this way, although little evidence of the demand has been documented. The three choices offered hardly reflect the multiple sources that define people’s mixed identity. The declared intention is distracted with a demographic concern to reflect the mixing of categories through the offspring of mixed unions. There is a pre-occupation, if not a fixation, with mixed marriages. They are seen partly as a positive sign of the most intimate form of integration, and partly as a threat to the standard categories of identity seen as important to the world (13).
6. The new use of ‘mixed’ categories was justified from the multicultural camp as reflecting new cultural identities. It has a seriously negative impact on the effectiveness of employment discrimination statistics, by reducing the number in the discriminated population (eg. ‘Black’) without demanding that the employing organisations re-survey their workforce to exclude ‘Mixed’ from the ‘Black’ employees and applicants (14).

Although the Census leads official statistical categorisation by providing baseline population figures for other studies, away from the Census there is an opportunity to plan the instruments of data collection. Most writers conclude that no measure of ‘ethnicity’ applies in the wide variety of contexts where it is needed. They recommend that the means of measurement should be related to the purpose of the research, while "If it is unknown which of ethnicity, race or culture is the most important influence then an attempt should be made to measure all of them. A range of information is best collected: genetic differences, self assigned ethnicity, observer assigned ethnicity, country or area of birth, years in country of residence, religion" and diet and language too where relevant (British Medical Journal, 1996) ⁽¹⁵⁾. At the same time the same writers add the warning that ethnicity cannot be unpacked into component parts, nor can it be re-constructed from measures of possible component parts (Aspinall, 2001).

Their conclusion seems to be that asking about ethnicity is at one and the same time necessary and untenable. A census question or questions are needed, but cannot measure what is needed. A range of questions are needed, but cannot be used to reconstruct ethnicity. Adding a religion question in the census has compounded the conceptual and motivational confusion ⁽¹⁶⁾. Things were certainly easier before ethnicity came along. "Ethnicity was a tool to blunt the edge of black struggle, return ‘black’ to its constituent parts and at the same time, allow the nascent black bourgeoisie, petit-bourgeoisie really, to move up in the system," according to Sivanandan writing in 1983 (Sivanandan, 1983).

To add to the conceptual difficulties, planning must involve prediction (as plans are about the future), and therefore some constancy of categories over time. But this is not possible with measurement of cultural affiliation. We each have many cultural affiliations, which we express differently in different contexts; we change our affiliation over time too, in response to our own development and understanding, as well as the environment within which our declarations of difference take place. What we are willing to identify within official enquiries depends not only on what is acceptable to ourselves but what we understand is socially acceptable and to our advantage, and will differ according to whom we feel the enquiry will be used for. Changes in affiliation are observed according to the political context and who fills in the form (some examples are given in Simpson, 1997).

All this leads to contradictory and self-defeating attempts to define ethnicity in quantitative social science that keep academics busy and probably has no effect on racism and social change.

4. Emerging from the mire

How can we emerge from this mire? Roger Ballard provides a helpful analysis of how race and ethnicity can be distinguished and seen to interact. He attacks the mainstream of social analysts who making ethnicity and race synonymous. He rejects ‘ethnic minorities’ as a misnomer, and rejects ‘self-identity’ as a reason for measuring ethnicity. Instead he highlights the rich diversity of ethnicity, which creates different cultural forms, including a variety of forms of resistance to racism (Ballard, 1998). This is helpful to our understanding but it doesn’t clarify much what measures might be useful, even if we focus on the one purpose of anti-racist action.

One might conclude that no measures of race or ethnicity can help tackle racism. Measure nothing. The collection of race data for whatever purposes reinforces the racial thinking that constitutes racial prejudice. Government neighbourhood regeneration programmes which target areas markedly different in racial composition, or which specifically target race or ethnicity as a reason for funding, are at best responding only to the surface impact of racism. Such programmes encourage competition and tensions between areas and social groups rather than unite them against causes of inequality.

But there are plenty of examples where statistical patterns are so clear that precise definitions are less important than clarity of interpretation. Statistics of inequality have kept anti-racism on the government agenda and are an element of sound education about society. Demographic statistics with an ethnic group dimension derived from the Census have successfully shown the need for continued support for children coming to primary schools without English, and for taking seriously the different cultural preferences of pupils’ families (Simpson, 1997). Clarity of interpretation demands firstly that racial causation is completely avoided in discussing the association between ethnic group and social conditions or behaviour. It also demands prominence for the social and political contexts for the story that statistics tell, with justification for publishing the statistics within those contexts.

Taking ‘research’ generally, it is easy to point to examples of good anti-racist practice, whether from investigative research of the kind promoted in the Searchlight journal, or from the political, cultural and legislative investigations supplied in Race and Class. Anti-racist statistical practice is more limited. It could be encouraged in three ways with which this article finishes: myth busting, exposure of inequality, and statistical modelling of institutional racism.

Statistical myth busting is putting legends under the magnifying glass of a statistical review. For example, current legends about self-segregation (usually pejoratively against Muslims and sympathetically about Whites as in ‘White flight’) are passed from official enquiry to official inquiry, from news item to news item, and from government minister’s mouth to all who are unquestioning enough to pass them on. Such legends are based on assumptions that can be tested with demographic data and laid to rest or amended with greater confidence about what represents truth. The legends to focus upon are those that have most political currency self-segregation, international marriage, polarised school composition, council estate and inner city conditions, immigration and receipt of state benefits.

Statistical exposure of inequality measures the end product of discrimination, without necessarily identifying the sources of that discrimination. It is similar to myth-busting in that it helps to demystify and oppose ruling ideas and stereotypes. Myth-busting and exposure of inequality are statistics’ contribution to anti-racist education. Measuring the extent of inequality is also necessary to plan and to implement serious anti-racist public initiatives, though its role is subsidiary to the will and the power to carry through those initiatives.

Statistical modelling of institutional racism attempts to explain inequality through the causes of discrimination, but is virtually absent from academic literature. Typologies of the processes of racism and institutional racism are needed that can be operationally implemented, attached to institutions and to the consequences suffered by individuals who are dealt with by those institutions. Statistical studies tend to measure non-process variables and claim that differences in social conditions that they cannot explain may be due to processes including racism, a weak and negative method of explanation as already discussed. Experiments are more powerful, but not easily set up with different institutional practises applied in the same type of institution. Longitudinal data are more suited to examining the effects of processes but are very much more expensive. Clearly, this item of the agenda will take time to develop.

The proof of anti-racist statistics is in their use. It is through liaison with anti-racist organisations on products of use to anti-racist struggle, that anti-racist statistics should have some credence.

5. Conclusions

The measurement of ‘race’ assumes and justifies a stratification of society while hiding its class relations and imperialist development. To a large extent statistics can be seen to monitor rather than reduce inequalities, divide rather than unite communities, and delay rather than encourage solutions. In the UK, the motivation for government measurement of ‘ethnic group’ can be traced to aims of anti-discrimination, multi-culturalism, controlling immigration, and social cohesion, each of which demands different categories and analyses. The leading instruments of government measurement, the Census and national surveys, have not developed categories of ‘ethnic group’ in a refined instrument representing a single theory, but contain the conceptual confusion of conflicting theories, and will continue to do so.

Statistical methods developed in response to historical problems; racist eugenic concerns were a high priority among the concerns of many of the early statistical theorists. Anti-racist priorities take statistics in the directions of campaigning myth-busting, highlighting inequalities and the full reasons for them, and the measurement of institutional racism.

I am willing to work with other readers of Radical Statistics who wish to pursue these aims in a collective fashion.

Notes

1. The link between statistical development and eugenics is explored for example by MacKenzie (1999) and by Zuberi (2001). Eugenic concerns focused on the supposed genetic impact of mixing between classes and races, and of the lower birth rates of the favoured classes.
2. Gould (1996) provides a fine account of the attempts to measure intelligence as innate and genetic qualities.
3. Now dropping the quote marks around these terms, on the understanding that they are all constructions that help to represent a theory about society, and not that they define something immutable about the person they are attached to.
4. The quote is from a report published by the Cabinet Office (2000). The opinion survey is reported by the Home Office (2001).
5. Black and Asian members of the public put discriminatory behaviour by the police ahead of any other service, according to the opinion survey reported by the Home Office (2001).
6. The standardisation or explanation of differences between groups is justified and necessary when it involves variables that are not social or cannot be the result of discrimination, as age is in some cases. Thus people in black and asian categories have low crude illness rates when expressed as the number of people ill as a proportion of the group’s population; but this is mainly due to their young age structures, where few people have reached the ages where illness is much more common. Age-standardised ratios show relatively high illness among black and asian groups, and low take-up of caring services among Asian groups who are much more likely to care for their elderly and sick within the community and without state support.
7. Job-seekers for hospital jobs were twice as likely to get an interview without an Asian-sounding name as with one, when using the same CV (Esmail, 1993).
8. Institute of Race Relations, Newsletter, March 1966, p.11, quoted in Leech, 1989.
9. Leech (1989) provides a full account of the course of porposals in the 1970s and 1980s for a question on ethnic group in the national census of Britain.
10. There was no campaign against the question prior to its use in 2001, except in Wales to write in ‘Welsh’. Early returns showed that the question achieved response rates over 95%, similar to questions on health and qualifications, and better than questions on employment (www.statistics.gov.uk/census2001).
11. Experiences of taking the Census in 2001 are surveyed in Simpson (2001).
12. The Office for National Statistics now recommends that surveys include a further question of ‘National group’ to elicit one response from ‘English/Scottish/Welsh/Irish/British/Other/Not stated’ and that further work on ethnic group ‘calls for an approach that transcends statistics and embraces sociological and political considerations as well’ (ethnic classification, 2002). The Department for Education and Skills’ Annual Schools Census has already taken a different line, including ‘English/Scottish/Welsh’ as sub-categories of White-British (www.dfes.gov.uk Ethnic categories key list and covering letter.doc, 2002).
13. Rocheron (1997) takes apart this pre-occupation within French demography. In the British government-sponsored feasibility study for population statistics, three chapters focus on the mixed group as requiring major technical innovation (Haskey, 2002).
14. In the U.S.A., those who tick ‘Multi-racial’ on the census to declare their identity, know which box to tick for affirmative action employers, but may not realise that the first decision may restrict the effectiveness of the latter (Wright, 1994).
15. Very similar advice is given by Aspinall (2001), Berthoud (1998) and Modood (1992). In this quote, presumably ‘genetic differences’ are presumably intended to refer to differences in skin colour, a throwback to the early eugenicist perspective that has been scientifically dismissed for many decades: while skin-colour is indeed inherited, there is far more genetic variation between population groups than within them.
16. See for instance articles by Boag, Reynolds and Southworth in Radical Statistics, Issue 78, Autumn 2001.

References

Aspinall, P. (2000) ‘The new 2001 Census question set on cultural characteristics: is it useful for the monitoring fo the health status of people from Ethnic Groups in Britain?’, Ethnicity & Health, 5(1), pp.33-40.

Aspinall, P. (2000) ‘The Challenges of Measuring the Ethno-cultural Diversity of Britain in the New Millennium’, Policy & Politics, 1, pp. 109-118.

Aspinall, P. (2001) ‘operationalising the Collection of Ethnicity Data in Studies of Sociology of Health and Illness’, Sociology of Health and Illness, 23(6), pp. 829-862.

Athwal, H. (2001) ‘The Racism that Kills’, Race & Class, Vol. 43, No. 2.

Ballard, R. (1998) ‘Asking Ethnic Questions: Some Hows, Whys and Wherefores’, Patterns of Prejudice, 32(2), pp. 15-37.

Berthoud, R. (1998) ‘Defining Ethnic Groups: Origin or Identity’, Patterns of Prejudice, 32(2), pp 53-63.

Bourne, J. (2001) ‘The Life and Times of Institutional Racism’, Race & Class, Vol. 43, No. 20.

British Medical Journal (1996) ‘Ethnicity, Race and Culture: Guidelines for Research, Audit, and Publication’, British Medical Journal, 312:1094.

British Society for Social Responsibility in Science (1981) Race: The Case Against Including any Question on Race or Ethnic Origin in the 1981 Census, London: BSSRS.

Browne, A. (2000) ‘The Last Days of a White World’ The Observer, September 3^rd.

Cabinet Office (2000) ‘Minority Ethnic Issues in Social Exclusion and Neigbourhood Renewal’, Report from the Social Exclusion Unit.

Cantle, T. (2001) ‘Community Cohesion: A Report of the Independent Review Team’, London: Home Office.

Commision for Racial Equality [CRE] (2002) http://www.cre.gov.uk/gdpract/em_cat_ew.html.

Denham, J. (2001) ‘Building Cohesive Communities: A Report of the Ministerial Group on Public Order and Community Cohesion’ London: Home Office.

Esmail, A., Everington, S. (1993) ‘Racial-discrimination Against Doctors from Ethnic Minorities’, British Medical Journal, Vol. 306.

Gould, S. (1996) Mismeasure of Man, New York: Norton.

Haskey, J. (2000) ‘Projections of the Population by Ethnic Group: A Sufficiently Interesting or a Definitely Necessary Exercise to Undertake?’, Population Trends, No. 102, pp. 34-40.

Haskey, J. (ed.) (2002) Population Projections by Ethnic Group, a Feasibility Study, Studies in Medical and Population Subjects, No.67, The Stationery Office.

HM Government (1978) 1981 Census of Population (Census White Paper), Cm 7146.

The Home Office (2001) ‘Race Equality in Public Services’.

The Home Office (2002) Nationality, Immigration and Asylum Bill, 12^th April 2002. Bill 119.