Intended for healthcare professionals


Current census categories are not a good match for identity

BMJ 1999; 318 doi: (Published 19 June 1999) Cite this as: BMJ 1999;318:1696
  1. Judith Rankin, Senior research associate (J.M.Rankin{at},
  2. Raj Bhopal, Bruce and John Usher professor of public health (rbhopal{at}
  1. Department of Epidemiology and Public Health, School of Health Sciences, Medical School, University of Newcastle, Newcastle upon Tyne NE2 4HH
  2. Department of Public Health Sciences, University of Edinburgh Medical School, Edinburgh EH8 9AG

    EDITOR—Pfeffer discusses the complexities of theories of race, ethnicity, and culture.1 Surveys in the United Kingdom (UK) rely on the census question that first appeared in the 1991 census (incorporating colour and country of origin2) to define ethnic groups. The Office of Management and Budget's classification is dominant in the United States.3 Does the menu of terms given to people included in such classifications offer a good choice? Using data from the south Tyneside heart study,4 we compared respondents' identification of their ethnicity using the census question, a description in an open question, country of birth, and country of family origin.

    We recruited participants using the snowball sampling technique.4 Community workers provided the names and addresses of people aged 16-74 of South Asian origin who had been resident in South Tyneside for at least a year. (The term South Asian is used to refer to those individuals whose ancestral origin lies in the Indian subcontinent—here, India, Pakistan, and Bangladesh—and includes those born in the UK and others migrating to the UK via a third country (for example, Kenya).)

    Participants (n=334) were interviewed by a trained interviewer using a structured questionnaire in the preferred language, usually in the participant's home. The respondents first chose one of the categories from the census question (white, black Caribbean, black African, black other, Indian, Pakistani, Bangladeshi, Chinese, and other), then provided a description of their ethnicity. They were also asked where they and their mother and father were born.

    Nineteen self descriptions were given. The most striking observation was the rarity of the term Asian and the absence of the term South Asian—both commonly used in the UK to describe people originating from the Indian subcontinent. These labels do not capture ethnic identity. From the census categories, 39% respondents chose Indian, 38% Bangladeshi, 13% Pakistani, 6% other, and 1 black other (table 1). Only 81 (62%) of those who were Indian on the census question described themselves as Indian when given the open choice.

    Respondents' descriptions of their ethnic origin by 1991 census category (15 missing values). Figures are numbers (percentages)

    View this table:

    There was less variation among those who chose Bangladeshi from the census categories. Altogether 74% of 27 respondents born in Pakistan described themselves as Pakistani, whereas only 56% of 93 born in India described themselves as Indian. Of 68 respondents born in the UK, 38% described themselves as Indian, 13% as British/English/Anglo Indian, and 10% as Sikh (table w1 on Self defined ethnic origin and country of family origin were highly related when the country of family origin was Bangladesh but less so when it was India or Pakistan (table w2 on

    Using census categories is insufficient to capture self identification. If we had not asked the census question first there might have been even less agreement between modes of self identification. Too few categories are offered to reflect the true heterogeneity of ethnic groups. Similar issues also apply to other labels—for example, “black” and “white.”5 Our analysis emphasises the need for fresh thinking if identity and self identification are to be the basis of ethnic grouping.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.