1. Background & Methodology
1.1 Background
As part of the preparations for the 2011 Census, the Scottish Government ( SG) commissioned Ipsos MORI to undertake cognitive question testing on a provisional set of questions on national identity and ethnic group. This forms part of a wider review of ethnicity classifications being conducted by the Scottish Government, in partnership with the General Register Office for Scotland ( GROS). This review has been ongoing for several years and involves an extensive programme of research and consultation 1. It has considered the views and requirements of people from different ethnic groups and people who use census ethnicity data, as well as the methodological issues involved in developing survey questions which will enable the collection of accurate, robust and useful data on the Scottish population.
Following the 2001 Census, strong objections were raised about the use of the term Black in the ethnic group category relating to people of African or Caribbean ethnicity. More broadly, concerns were expressed that the 2001 question used a combination of ways to categorise ethnicity, including nationality, skin colour/race and geography (and did so inconsistently across ethnic group categories), and that it failed to capture the complexity, fluidity and multi-faceted nature of contemporary ethnic identity.
As part of the ongoing ethnicity review, the Scottish Government developed a revised ethnic group question, as well as a national identity question, which were included in GROS' 2006 Census Test 2. Following the Census Test and the Census Test Follow Up Survey 3, further research and consultation has been conducted 4, and the test questions have been incrementally developed based on the findings of this work.
Whilst revisions have been made to the proposed ethnicity classification to address the concerns which have been raised, it is also possible that some of the changes could have unintended or unhelpful consequences. It is against this backdrop that the Scottish Government commissioned Ipsos MORI to conduct detailed cognitive testing of the proposed ethnic group and national identity questions, in order to understand how respondents might interpret and answer revised questions, and thereby get an understanding of how the proposed questions work in practice.
1.2 Ethnicity and the Scottish Census
Ethnicity has been recorded in the Scottish Census only since 1991. The 1971 and 1981 censuses provided a proxy for ethnicity derived from each person's country of birth, but by 1991 this was no longer a useful basis for ethnic classification. Between 1991 and 2001 the categories for recording ethnic group changed with the introduction of over-arching categories of White/Black/Asian etc, the inclusion of 'mixed' categories, and the identification of Scottish and Irish as ethnic categories alongside British.
The rationale for the introduction of, and changes to, the ethnic group categories are interesting but secondary to the fact that they demonstrate the fluidity of ethnic classifications: the emergence of new information requirements reflects a combination of increasing awareness of ethnic difference and identity as social and political phenomena and the increasing complexity of ethnic classification in a multi-racial society. In addition, unlike many of the other questions addressed by the census, ethnicity is not an entirely objective fact but a personal and social construct reflecting "a real or putative shared identity based on one or more symbolic elements such as culture, language, religion, kinship, shared territory or physical appearance". 5 More specifically, research by Akinwale 6 comparing responses from the same people to the 1991 and 2001 Censuses as part of the ONS Longitudinal Study demonstrates that the ethnic group reported by individuals changes independently of social characteristics and question wording - people change their perception of their own ethnicity over time.
That ethnicity might change should be expected from the nature of the topic. It is, as Reber 7 notes, "a politicised social construction of reality" and we should expect debates around ethnicity (and religion, which Akinwale demonstrates as strongly linked to ethnic identity) to influence how people view themselves. This means that each person's ethnic identity is a mix of:
- physical (essentially racial) characteristics - skin colour and other physical features
- nationality and ancestral nationality - many people born overseas may see themselves as British by having taken British nationality or by being British by birth and many Americans claim to be Scottish on the basis of distant ancestral connections
- group connections and adherence to cultural aspects associated with particular ethnic groups such as religion
- political events and debates - classic examples include the radicalisation of black Americans, the emergence of greater number of native Americans in US Censuses as people reappraised their own identity in changing political contexts. In the UK, the emergence of an Islamic identity among (and to refer to) young Asians in Britain is seen as replacing weaker connections with countries of ancestral origin and reflecting world politics.
In addition, the identity people report will reflect their perceptions of themselves, their perceptions of the questioner - how the person or body formulating the question expects them to respond to it - and their estimation of how their response might be interpreted. For example, in the Scottish Household Survey, which uses 2001 Census categories, respondents frequently do not choose the official categories for White people - Scottish, Irish, Other British, etc - and prefer to classify themselves as British (not 'other British', simply British).
The Census faces a peculiar challenge in collecting data. It is the definitive statement of social and demographic facts. In its own right it is the most important (and often the only) source of information at small area level, and it sets the benchmark for other datasets which adopt the classifications used in the Census and compare their outputs with Census findings. While the Census needs to be as accurate as possible, since it is a self-completion exercise, it is also limited in the burden it can place on respondents and in the complexity of the task it asks them to perform. Census questions are necessarily a compromise. But because of its role in providing data and shaping other types of data collection, the Census not only records social phenomena, it is instrumental in shaping those phenomena, or at least in the period between Censuses, it helps define the way in which issues are discussed. It helps to create the language used to discuss and analyse ethnicity and it helps to form the boundaries of the discussion: ethnic groups are more likely to be analysed and discussed if they are officially recognised and categorised in the Census.
Proposals to amend the ethnicity question for the 2011 Census need to be viewed in the light of these different features of ethnicity and the Census. Three things flow from this discussion:
- there is no objective fact of ethnicity that remains fixed. Instead, people change their identification of themselves in response to social changes;
- the Census will necessarily be limited in its capacity to record the full range of variation in ethnic groupings. It is an exercise in finding the best fit between the need to collect robust and useful data, the range of ethnicities possible and the limited space within which to record them; and
- since it is a self-completion exercise, people have to rely solely on the information on the form to identify themselves among the categories. They will therefore attach meaning to categories and might adapt their responses to reflect their interpretation of the categories given to them.
The methods used in this research reflect these issues and draw on the need for clarity and simplicity in collecting data and people's responses to survey questions. It also anticipates where response problems might arise and the implications of these problems for data collection and data processing.