Original ResearchThe utility and validity for public health of ethnicity categorization in the 1991, 2001 and 2011 British Censuses
Introduction
A question on ethnic group was asked for the first time in the 1991 Census in Great Britain. Revised ethnic group questions that were similar in structure were asked in the 2001 and 2011 Censuses for England and Wales, and Scotland. Data from the 1991 and 2001 Censuses have transformed our knowledge of minority ethnic health and health care by providing, for example, population denominator data for rates/ratios, information on the determinants of health and health inequalities, a template for collecting ethnic data in administrative systems and surveys, and, in Scotland, a means of populating ethnic group in administrative data that lacked this information.1 Additionally, in 2001, a question on general health was added to that on limiting long-term illness, and a third more detailed health question was added in the 2011 Census in Scotland. These gains for population health have been widely acknowledged.
However, there has been less focus on the utility and validity of census ethnicity classifications and categories for public health, and on census agencies’ and respondents’ understandings of the concept of ethnicity. Census ethnicity categories are now widely used in public health for the stratification of data, such as that on health-related behaviour and the use of health services, and in population profiling for public health risk assessment. Most debate has focused on the meaning of ethnicity and race, and the reliability of these social constructions as proxies for other biological, social and biosocial variables. There is general agreement that these concepts are crucial for assessing the risk of discrimination and disadvantage along the lines of race and ethnicity; moreover, the view that they should only be used as proxies for other variables that cannot be measured and when they are the most reliable proxies available is judicious.2
A wide range of opinion exists on whether ethnicity can be reliably measured, including whether it can accurately reflect the most salient categories of group identity,3 but with some consensus on the fact that it should be self-assessed.4 Some investigators point to the limitations of single, mutually exclusive categories for measuring a multidimensional concept, and favour breaking down ethnicity into elements which can be measured separately, and analysed jointly or separately. Such elements might include parentage or ancestry, national identity, language, religion, country of birth, and patterns of behaviour, friendship and association.5 Others eschew this multidimensional approach that unpacks ethnicity as too simplistic, arguing that a global measure is needed as ethnicity is a context-driven social and psychological concept.6, 7
At the interface between those asking ethnicity questions (census agencies) and those answering them, the diversity of ways in which ethnicity is understood by these actors is poorly documented. UK census research into respondents’ understandings of the terms ‘ethnic’, ‘culture’, ‘nationality’ and ‘race’ has shown that these concepts are generally quite separately defined but are sometimes used interchangeably.8 Most respondents also distinguished between ‘ethnic origin’ and ‘ethnic group’, recognizing the importance of the former as referring to an individual’s parental background while not necessarily associating themselves with a particular group. Parentage was much more important in determining ethnic group than where a person was born or language and religion in surveys of patient populations.9 Again, amongst mixed race respondents, the contribution of parental race/ethnicity to ethnic identity was far more salient than social factors (such as the perceptions of wider society, feelings of group allegiance, and identification by friends and peers).10
The importance of the different dimensions of ethnicity to self-identity may vary across groups. For example, national identity (being British) was shown to be especially significant for Black groups in the free-text responses of the 1991 and 2001 Censuses,11 while other research indicates that religion is prioritized amongst South Asian groups.12 Although non-response rates for the ethnic group question in censuses and surveys are now very low, utilization of ‘other’ write-in categories is high, and it is clear that many respondents need the cues of the question’s context, including list of categories, to answer it successfully. In one interview survey of unprompted ethnic group, almost one-quarter of respondents replied that they did not know what an ethnic group was or were unsure which group they belonged to, while almost all selected a category when prompted with the census classification.9
Census ethnicity questions therefore need to be robustly designed and tested to ensure acceptability and understanding by those answering them. The Office for National Statistics (ONS) acknowledges that ethnic identification is a multifaceted and changing phenomenon which may include aspects of ancestry, country of birth, nationality, language spoken at home, religion, culture, skin colour and national/geographic origin. However, in placing emphasis on the subjective nature of ethnicity, it argues that respondents will draw on these dimensions in ways that are relevant to them, while acknowledging that the ethnic group options presented to the respondent are not completely those of self-identity.13 Processes of group identification and social categorization are not, however, separate but are mutually implicated and feedback to each other.14
Moreover, when the data collection instrument is a national, decennial census, there are additional issues to consider. The categorization cannot always incorporate groups which may be numerically important at a local level as opposed to a national level. Moreover, taking part in the census is compulsory, so priority is accorded to matters such as acceptability, parsimony imposed by the limitations of questionnaire length, respondent burden, sensitivity with respect to personal information (the question on religion is voluntary), and optimal capture (the 2011 Census in England was printed in 56 other languages).
Section snippets
Methods
A narrative review of published literature accessed via Medline, Embase, King’s Fund Database, DH Data and Web of Knowledge; reports of the ONS and General Register Office (Scotland) Census Development Programmes (CDPs); and other policy material was undertaken to address the research question: What is the utility and validity of ethnic categorization in the 1991, 2001 and 2011 Censuses for public health purposes? The assessment of a comprehensive body of knowledge on contested concepts,
Results
The results of the literature synthesis are presented with respect to specified criteria of utility and validity for public health. Amongst utility criteria, capture of the complex ethnic diversity of the country is needed to facilitate studies on the health of populations. However, respondents’ understandings and burden, and the need for data that does not incur the ‘small numbers’ and related confidentiality problems set limits to increased granularity. The categories should be measurable in
Discussion
The census continues to be an important source for public health specialists engaged in work on health protection, health improvement and health inequalities issues for a number of reasons. The categorization and classifications are the outcome of robust regimes of testing, resulting in their rapid adoption as the gold standard measure across most official ethnicity data collection (such as that by hospitals, general practices and the social care sector) and monitoring. This facilitates
Ethical approval
None sought.
Funding
None declared.
Competing interests
None declared.
References (60)
The conceptual basis of ethnic group terminology and classifications
Soc Sci Med
(1997)- et al.
Ethnicity and health in Scotland: can we fill the information gap?
(2006) ‘Population profiling’ and public health risk: when and how should we use race/ethnicity?
Crit Public Health
(2005)- et al.
Census categories of ethnic group are limited
BMJ
(1995) Describing ethnicity in health research
Ethn Health
(2003)- et al.
Measuring ethnicity: challenges and opportunities for survey research
Ethn Racial Stud
(2010) The medicalization of race: scientific legitimization of a flawed concept
Ann Intern Med
(1996)Ethnicity in psychiatric epidemiology
Br J Psychiatry
(1998)- et al.
Ethnic group question: findings from focus group discussions
(1996) - et al.
Practicality of recording patient ethnicity data in general practice: descriptive intervention study and attitude survey
BMJ
(1996)
A guide to comparing 1991 and 2001 Census ethnic group data
Changing ethnic identities
Ethnic group statistics: a guide for the collection and classification of ethnicity data
Social identity
Who are the Whites? Imposed Census categories and the racial demography of Brazil
Soc Forces
Fed Regist
International migration and the United Kingdom. Report of the United Kingdom SOPEMI correspondent to the OECD, 2008
Inter-ethnic unions. Census update 2005/12
‘Mixed race’, ‘mixed origins’ or what? Generic terminology for the multiple racial/ethnic group population
Anthropol Today
Operationalising the collection of ethnicity data in studies of the sociology of heath and illness
Sociol Health Illness
Superdiversity and its implications
Ethn Racial Stud
Information paper. Deciding which tick-boxes to add to the ethnic group question in the England and Wales Census
1991 census. Definitions. Great Britain
2011 census special populations advisory group. Paper SPAG(05)02. Taking forward the 2001 census community liaison programme: draft proposals and timetable
Testing methods of collecting racial and ethnic information: results of the current population survey Supplement on race and ethnicity. Statistical Note No. 40
Capital assets: a community research intervention by the African Forum in Redbridge and Waltham Forest
People, Families and communities survey
Cited by (32)
Improved prediction equations for estimating height in adults from ethnically diverse backgrounds
2020, Clinical NutritionCitation Excerpt :Allowing participants to self-identify their ethnic group is important but constrained by the options that are offered to them and is a sensitive area. A degree of categorisation is necessary to enable data to be analysed but the challenges associated with heterogeneous data and lack of clear boundaries is recognised [37,38]. This is exemplified by the modelling identifying the measurements of Indian participants of the current study as different from those of other Asian participants but heterogeneity of data was noted even within the Indian group.
Ethnic differences in self-assessed health in Scotland: The role of socio-economic status and migrant generation
2022, Population, Space and PlaceDiscourses on im/migrants, ethnic minorities, and infectious disease: Fifty years of tuberculosis reporting in the United Kingdom
2022, History of the Human Sciences