Scotland's New Ethnicity Classification for Scottish Official Statistics and Recommended for Scotland's 2011 Census

Annex C Key Findings of Each of the Main Evidence Strands

1. Key findings from the evidence collected by the review team are summarised below. Links to full, published findings are provided, where available. Links cannot be provided for analysis containing disclosive confidential data (with very small numbers) or where information was provided in confidence. Some findings are part of a separate ongoing project, e.g. the second wave of cognitive testing, and have not yet been published. The review also used some evidence generated by ONS (as detailed on page 20), however this is not referenced here as ONS's review is ongoing and will not be published until Autumn 2008. Scotland's evidence is presented under three broad category headings; research, consultation and questions testing.

RESEARCH

Ethnic Identity and the Census (2004-05)

2. Research was commissioned to explore how people wish to classify their ethnicity and what information data users require. The research used in-depth interviews and focus groups.

3. Stakeholders (11) wanted skin colour to be unlinked from ethnicity and nationality and saw the dichotomy of 'White'/`Black' in the existing classification as unacceptable. A separate question on colour was suggested.

4. Data users (12) wanted more information about white minority groups, a longer list of ethnic categories and a review of both the 'other' ethnic group category and the relationship between ethnicity, colour, nationality and religion.

5. Data providers (39 ) gave opposing views on the use of colour terms as descriptors of their ethnicity; it was central to some people's ethnic identity whilst others strongly objected to it.

Links: Ethnic Identity and the Census - Full Research Report (June 2005)

Ethnic Identity and the Census - Summary Report (June 2005)

Linking 2001 Census and the 2006 Census Test ethnicity responses (2007)

6. GROS linked responses from the 2001 Census ethnic group question (which includes references to national identity and colour) to the same person's response on the 2006 Census Test ethnic group question, (which excluded references to national identity and colour); for the same person. The effect of question changes on response patterns was assessed. Approximately 35,000 records were linked. Numbers of linked records for 'non-Europeans' were small and these findings are subject to error.

7. Tick boxes for each UK country (2006 question) encouraged responses away from the 'Other British' tick box (2001 question), suggesting they better reflect UK ethnicities. Over half of those identifying as 'British' in 2006 identified as 'Scottish' in 2001, suggesting that many Scots wish to identify as 'British' if given the option.

8. The 'European' tick boxes in 2006 drew people who identified as 'Mixed' or 'Asian' in 2001. This may be because they were born or had lived for a number of years in Scotland or elsewhere in Europe.

9. There was movement from 'White' in 2001 to 'African' categories in 2006, suggesting that White 'Africans' identified more with the 'African' than 'European' category on the 2006 question.

Scottish focus groups on use of colour as an ethnicity descriptor ( 2007)

10. SG commissioned ONS to undertake 12 focus groups (with 96 participants in all) with members of the public self-identifying as 'African', 'Caribbean', `Black' or 'Mixed' ethnicity in Aberdeen, Edinburgh and Glasgow to explore the acceptability of using colour to classify ethnicity. Due to recruitment difficulties, only the Glasgow groups went ahead (with 30 participants).

11. Ethnicity was seen as complex and multi-faceted including roots, origins, background, parents birth place, belonging, culture, beliefs, values, nationality, race, language and physiology. It was seen as fluid over time and wider than just race or skin colour.

12. A majority of participants felt comfortable describing themselves as 'Black' and of these many were proud of it. Those who preferred to be described by country of origin disliked the negative connotations of the term 'Black' and did not want this label imposed on them.

13. The meaning of the term 'Black' was seen as context specific and could denote skin colour, music, culture, history etc. Therefore, the acceptability of the term 'Black' was situation specific for some people.

Appraisal of test questions (national identity and ethnic group) by Glasgow City Council and The City of Edinburgh Council (2007-08)

14. The review team provided Glasgow City Council and The City of Edinburgh Council with a test version of classification (as at November 2007) for their appraisal, as data users and service providers. Local authorities have a statutory requirement, under race relations legislation, to promote equality and tackle discrimination on the grounds of race and they collect ethnicity statistics as part of this.

15. Feedback was given on how well the classification (including a national identity and ethnic group question) would enable them to meet their statutory obligations under race relations legislation for the purposes of service provision, resource allocation and promoting equality. Key equality and service delivery personnel participated and a household survey was also used by Glasgow.

16. Overall, national identity was seen as an aid to self-expression of increasingly complex identities. However, it would not help local authorities to fulfil their statutory obligations, service delivery, resource allocation or tackling discrimination in any substantive way.

17. Changes to the ethnic group question were generally welcomed as improvements that would help effective service delivery to communities. 'Roma' or 'Romany' people were identified as an emerging ethnic group facing potential discrimination. The need to maintain comparability with the 2001 Census was highlighted in several cases.

18. Glasgow City Council included the test version of the classification (see ethnic group question 3, page 27 and national identity question 3 at Annex F) in its annual Household Survey. A representative quota sample of 1,020 residents was interviewed by Mruk, with 10 interviews carried out in each of 102 Census Output Areas. Interviews were conducted using Computer Assisted Personal Interviewing, face-to-face in residents' homes between November 2007 and January 2008.

19. No operational difficulties were caused by introducing a new classification into this survey. No respondents raised objections to the classification (categories or tick boxes) with the interviewers. On the national identity question, the majority of respondents identified as 'Scottish' (89%), followed by 'British' (9%). Three per cent of people identified a non- UK national identity. Two per cent said they did not understand the meaning of national identity.

20. On the ethnic group question, the largest groups were 'Scottish' (90%), 'British' (4%) and 'Pakistani' (3%). Numbers for other groups were very small, therefore findings should be interpreted with caution. However, in relative terms 'Polish' was a large group at 0.5 per cent of respondents (the same proportion as identified as 'English'). No respondents identified as 'Black, Black Scottish or Black British' or as 'Gypsy/Traveller'. More respondents identified using the 'Other [African, Caribbean or Black] write-in box (0.3%) than the 'African, African Scottish or African British' tick box (0.1%). More respondents identified as 'Indian' (0.6%) than 'Sikh' (0.1%).

21. Only 0.1 per cent refused to answer the ethnicity question. Three per cent said they did not understand the meaning of an ethnic group and only 1 per cent did not understand that they could only give a single response. Nobody said they were unable to find an appropriate response option or said they objected to the colour labels 'Black' and 'White'. Nobody objected to the 'Sikh' tick box or reported that having to make a choice between 'Indian' and 'Sikh' was confusing. Only 0.4 per cent objected to the category labels 'A' to 'E'.

22. Following its main Household Survey, Glasgow City Council carried out a further 200 interviews with residents from 'non-White' ethnic groups. Again, the test version of the classification was included (see ethnic group question 3, page 27 and national identity question 3 at Annex F). Two results were particularly interesting. Twenty five per cent of respondents in this group did not understand the meaning of 'national identity' and 13 per cent refused to answer the ethnic group question.

Appraisal of test questions (national identity and ethnic group) by NHSScotland (2007-08)

23. In November 2007, ISDNHSScotland helped the review team to appraise the test classification (see ethnic group question 3 on page 27 and national identity question 3 on page 86). It was provided to key equality officials in a cross section of Health Boards. These included Greater Glasgow & Clyde, Highland, Lothian, Orkney and Tayside. This was to find out if the classification would meet their needs ( i.e. those of a large organisation) in terms of equality legislation, service provision, patient care, tackling discrimination and promoting equality.

24. As with the Local Authorities, the national identity question was seen as an aid to self-expression, particularly for people who were born in the UK (and identify strongly as 'Scottish') but whose parents or grandparents were born outside the UK. However, it is not seen as useful for delivery of patient care; questions on language and religion were seen as more important to know.

25. The revised ethnic group question was received positively as providing a more comprehensive ethnic profile of the population. Again, complementary information on language and religion were seen as important for effective delivery of patient care to different ethnic groups.

26. A need for information on A8 nationals other than 'Polish', was highlighted, particularly 'Slovakians' and a need for more information on the other white minority ethnic groups. 'Sikh' and 'Jewish' were regarded as religion groups rather than ethnic groups and their inclusion in the ethnicity question caused confusion.

Desk Research (2005-2008)

27. The review team conducted various pieces of desk research on ethnicity and its classification. This included analysis of the 2001 Census and other existing research in the field of national identity and ethnicity. Some of this analysis has been published. Key references and findings are outlined below.

'Ethnic group statistics - A guide for the collection and classification of ethnicity data' (2003)

28. In 2003, ONS published official guidelines for the collection and classification of information on ethnic group. Guidance is given for two methods: a two-question and a single-question method.

29. It was recommended that, generally speaking, a two-question classification should be used i.e. a national identity question followed by an ethnic group question. A single ethnic group question was recommended when only one question can be asked or direct comparison with 2001 Census data is important, response rates will not be damaged by omission of a national identity question or national identity is not important to measure. A national identity question and response categories were recommended for England, Scotland and Wales.

Link: Ethnic group statistics - A guide to the collection and classification of ethnicity data (2003)

'Who are the 'Mixed' ethnic group?' (2006)

30. In 2006 ONS undertook a demographic/socio-economic analysis of people responding to the 'Mixed Ethnic Group' category in the 2001 England & Wales census.

31. The 'Mixed' ethnic group category was first included in the UK censuses in 2001. Four 'Mixed' tick boxes were included in the England & Wales census and Scotland used a single write-in box. In England & Wales, the majority (but not all) of people with a 'Mixed' ethnic identity had one 'White' parent and one 'African', 'Caribbean' or 'Asian' parent and were born in Britain. People of 'Mixed' ethnic identities had the youngest age profile of any ethnic group in GB. Half were under 16 years of age in 2001. Four-fifths of people with 'Mixed' ethnic identities were born in the United Kingdom.

Link: Who are the 'Mixed ethnic group?' (May 2006)

Mixed Race in Britain: A Survey of the Preferences of Mixed Race People for Terminology and Classification: Interim Report (2006)

32. In 2006, the University of Kent ran a pilot study to find out how people with a 'Mixed' ethnic identity describe and classify their ethnicity. Participants completed three ethnic group questions: (version 1) three tick boxes 'White and Black Caribbean', 'White and Black African' and 'White and Asian' and an 'Any Other Mixed background' write-in box (version 2) a single tick box 'Any Mixed Background' and a write-in box and (version 3) an option to multi-tick all applicable ethnic groups. Preferences were then obtained.

33. Three quarters of respondents chose to write-in their ethnic identity, most with two short terms. Respondents most preferred the term 'Mixed race', followed by 'Mixed heritage' and 'Mixed parentage'. A majority found version 1 easiest to complete and version 3 most difficult. Almost equal numbers thought that versions 1 and 2 best enabled them to describe their racial/ethnic identity. Only a fifth said this about version 3, the multi-tick option. Versions 1 and 2 scored best on respondents' ability to understand the question. The aspects of version 1 most disliked were that all the mixes included 'White' and, in this way, were too limited.

Link: Mixed Race in Britain: A Survey of the Preferences of Mixed Race People for Terminology and Classifications: Interim Report (July 2006)

Ethnic classification of people from Central and Eastern Europe - What do we know? (2007)

34. In 2007, the review team carried out an analysis of what is known about the information needs of data users on recent migrants from Central and Eastern Europe and how such individuals classify their ethnicity. Data sources used were: Accession Monitoring Reports produced by the Home Office, analysis of responses given on the 2001 Census and 2006 census test, consultation with census data users and cognitive question testing.

35. Home Office statistics show that since May 2004, around 70 per cent of A8 migrants to Scotland were 'Polish' nationals. Responses to the 2006 census test show that people from Central and Eastern Europe use national identity, nationality or country of birth to describe their ethnicity i.e. 'Poland'/'Polish', and not the term 'Eastern European'.

36. There was a significant need among data users for information on recent migrants from A8 countries and they felt this could be best captured using several census questions, including ethnic group. Their information needs tended to be around language and housing provision and to monitor discrimination. They found it difficult to find robust, alternative information sources of information.

CONSULTATION

Seminar with key stakeholders to consider ethnicity research findings (2005)

37. In 2005, the review team held a seminar with key stakeholders to present interim findings of its 'Ethnic Identity and the Census' research. Three workshops were held on 1) Consulting with communities, 2) Monitoring ethnicity data in a changing Scotland and 3) Provision and use of ethnicity data. Forty-one delegates attended from a wide variety of public and community bodies.

38. Workshop 1. Barriers to community engagement included suspicion, consultation fatigue, perception of institutional racism and language ability. Some solutions included using community gatekeepers, targeting the right communities and feeding back results.

39. Workshop 2. Migrant workers and refugees are changing Scotland's ethnic profile, with mixed ethnicities becoming more diverse and difficult to categorise. Ethnicity is fluid; it differs across generations, is context specific and may be affected by devolution.

40 Workshop 3. Some object to providing ethnicity data for historical reasons, confusion or objection to the classification used or limited understanding of why data are collected. This may be eased by explaining the purpose of collecting ethnicity statistics and how data are used.

Link: Scottish Executive Review of Census Ethnicity Classification Seminar - Note (April 2005)

Review of Census Ethnicity Classifications Consultation and Analysis - (2005)

41. The SG's consultation explored separating out some different aspects of ethnic identity: geographic background, ethnic affiliation, national identity, visibility (including colour), religion, language etc. The consultation set out the background and asked a total of 11 questions focused on four topic areas. It was distributed to around 450 organisations, individuals, community groups and was issued publicly.

42. Ninety-three responses were received. The majority (82%) of responses were on behalf of organisations or groups, and 18 per cent responded as individuals. The largest number of organisational responses (32) was from local authorities, representing 42 per cent of total organisational responses and 34% of all responses.

43. The majority of respondents (70) agreed that several questions should be used to capture information on ethnic identity instead of a single question. Key reasons for this were that a single question limits the scope of information and detail that can be collected on ethnic identity, it is important to enable people to identify in their preferred way, or that ethnicity is too complex to be captured using a single question.

44. Despite support for new ethnicity question(s), the need to maintain comparability over time and across data sources was highlighted. Some respondents asked for the number of new questions to be balanced against the information need of data users.

45. The issue of how to capture information on colour prompted the most varied response. That said, a total of 30 respondents favoured the use of the term 'visibility' compared to 11 favouring use of the term 'colour'. None of the minority group organisations supported the use of the term 'colour'.

Links: Review of Census Ethnicity Classifications Consultation (June 2005)

Analysis of Response to the Review of Census Ethnicity Classifications Consultation (December 2005)

2011 Census Question Consultation and Analysis (2004-2006)

46. Between Autumn 2004 and January 2006 GROS ran an online consultation to gather views on the 2001 census questions, asking which questions should be retained, dropped or modified for the 2011 Census. Eighty-six responses were received; over half from individuals and a fifth from Local Authorities. Respondents were asked about the ethnic group question used on the 2001 Census.

47. A majority thought that the ethnic group question should be modified. Most agreed that it is inappropriate to mix colour terms and national identity but some felt that this is the only way to fully capture ethnicity. Some Local Authorities stressed that any new ethnic group question must allow comparison with questions asked on previous censuses and surveys. Some respondents felt that the question should be reduced in size.

Links: Autumn 2004 Census Consultation

Report of the Autumn 2004 Census Consultation

Spring 2007 Census Consultation and Analysis (2007)

48. In 2007, GROS conducted a formal consultation with census users, asking for views about the content of the questionnaire, general conduct of the census and its outputs. As part of this, specific questions were asked about data needs for national identity and ethnicity, preferences for the national identity and ethnicity questions in the 2006 census test as compared with the 2001 Census and the impact of question changes if full historic and UK comparability were not achievable. Responses totalled 128 and a majority were received from organisations or groups.

49. National identity was often misunderstood as citizenship, nationality or country of birth and therefore misunderstood by some as a measure of immigration. Around a third of respondents had little or no use for such data though some believed it aided self-expression and could provide information on integration and affiliation.

50. A majority of respondents use ethnicity data for policy development, followed by service provision, tackling discrimination, promoting equal opportunities and workforce monitoring, often to meet statutory requirements under race relations legislation.

51. A majority of respondents require ethnicity data at Scotland level but a sizeable proportion need information for GB or the UK. Some respondents (notably business and commercial users) consider UK comparability crucial. For others, developing a meaningful question is as important as GB and UK comparability.

52. A loss of comparability with ethnicity data derived from the 2001 Census question is problematic for monitoring change over time and since a new question(s) would require operating and monitoring systems to be changed this would be a significant burden on data users.

53. A majority of respondents preferred the ethnicity question on the 2006 census test (excluding colour and national identity and based on geography) over the question used on the 2001 Census (based on colour, national identity and geography). Reasons for 2006 preferences included more detailed breakdown of categories and more acceptable terminology i.e. removal of colour labels. Reasons for 2001 preferences included comparability of data and ability to monitor discrimination on the basis of colour.

Links: Spring 2007 Census Consultation

Analysis of Responses to the Spring 2007 Census Consultation

Scottish stakeholder meeting on 'African, Caribbean and Black' ethnicities (2007)

54. In September 2007, the review team held a stakeholder meeting to provide an update on the latest position regarding the development of Scotland's ethnicity classification and to discuss this. The focus was on classification of 'African', 'Caribbean' and 'Black' ethnicities.

55. A small number of stakeholders, who had contributed to consultation earlier in 2007, were selected on the basis that each person represented a different facet of the complex and polarised debate on the acceptability of colour terminology. Some stakeholders were individuals, others represented organisations and the meeting was mediated by an external facilitator. No decisions were taken at the meeting.

56. The review team highlighted that ethnicity data from the census are used for resource allocation, monitoring discrimination, service provision and policy development and that for communities to be visible in the statistics it is vital that they complete the next census. It was emphasised that the census is a statistical tool used to measure Scotland's demographic profile and is not a political vehicle.

57. Some stakeholders were in favour of using colour to describe ethnicity, whilst others believed that ethnicity should not be equated to skin colour. Some felt that the inconsistencies within the ethnicity classification being developed would make it difficult to monitor inequality, whereas others felt this was not the case.

58. Alternative forms of classification were discussed including: removal of colour, using geographic categories and replacing category headings with an alphabetical list of all ethnic groups. The review team explained some of the constraints on question design, the rationale for category and tick box ordering and the evidence on 'Sikh' and 'Jewish' as ethnic groups.

QUESTION TESTING

Scotland's 2006 census test (2006)

59. In April 2006, GROS ran a census test to evaluate possible new questions and assess different ways of completing the census. Around 52,000 households in various Scottish regions (including parts of Glasgow City Council, West Dunbartonshire, Highland, Stirling, Perth & Kinross and Argyll & Bute City Council areas) were included. These areas were purposely chosen for the test because each presented particular enumeration challenges. Therefore the resultant frequency counts are not necessarily representative of Scotland as a whole and results have been interpreted with caution. A new national identity question was tested, alongside a revised ethnic group question based on geographic categories and with references to colour and national identity removed (see ethnic group question 2 on page 26 and national identity 2 at Annex F).

Results from the ethnic group question

60. Frequency counts of responses to the ethnic group question are shown below. Under the 'European' category, 'Scottish' and 'British' tick boxes elicited the highest response (34,154 and 4,150 respectively) - suggesting that a fair proportion of Scottish residents may identify ethnically as 'British' rather than as 'Scottish', 'English', 'Welsh', or 'Northern Irish'.

61. 'Asian: Sikh' elicited 75 responses suggesting that some 'Asian' people tested may identify as ethnically 'Sikh' in a census proper rather than as 'Asian: Indian' or another ethnicity.

62. Twenty-one people identified as 'Jewish', suggesting that some people would identify ethnically in this way were they to be given this option on the census proper.

63. Ninety-one people gave responses using the new 'Arab' tick boxes. The relatively small number of responses to 'Arab: Other' (12) and 'North African' (5) suggests that the 'Arab' tick boxes provided were adequate response options for those completing the test.

64. The relatively small number of responses to 'African - Other' (2) suggests that the African tick boxes were adequate for those tested. Similarly, the relatively small number of responses to 'Other ethnic group: Other' (26), at the end of the question, suggests that most respondents in the test found, and used, a suitable category (and tick box or write-in box therein) to describe their ethnicity. The relatively high number of responses to the 'European: Other' (486) and 'Asian: Other' (107) tick boxes suggests that the tick boxes listed under these categories may not have been fully adequate or have functioned as intended, for those tested.

65. Five per cent of respondents did not answer the question. Three per cent gave multiple responses (only single responses are permitted on this question). Relative to all other questions asked on the test, multiple responses were high on the ethnic group question.

Ethnic Group Question - Frequency Counts
Response	Frequency	%
European - Scottish	34,514	73.5
European - English	1,562	3.4
European - Northern Irish	181	0.4
European - Welsh	79	0.2
European - British	4,150	8.9
European - Irish	549	1.2
European - Other (write-in)	486	1.0
Multiple ethnic groups (tick and write-in)	100	0.2
Asian - Pakistani	1,026	2.2
Asian - Chinese	115	0.2
Asian - Indian	110	0.2
Asian - Bangladeshi	7	0.0
Asian - Sikh	75	0.2
Asian - Other (write-in)	107	0.2
Arab - Middle East	58	0.1
Arab - North African	21	0.0
Arab - Other (write-in)	12	0.0
African/Caribbean - North African	5	0.0
African/Caribbean - East African	99	0.2
African/Caribbean - Southern African	52	0.1
African/Caribbean - West African	92	0.2
African/Caribbean - Central African	43	0.1
African/Caribbean - Caribbean	8	0.0
African/Caribbean - Other (write-in)	2	0.0
Other - Gypsy/Traveller	21	0.0
Other - Jewish	21	0.0
Other - Other (write-in)	26	0.1
No Response	2,099	4.5
Error	1,234	2.7

Results from the national identity question

66. Frequency counts of responses to the national identity question are shown below. The majority of test respondents identified as 'Scottish' (76.3%) or 'British' (20.0%). The proportion of respondents identifying as 'British' on the national identity question (20.0%) was higher than those who identified as 'European: British' (8.9%) on the ethnic group question. This suggests that the national identity quesiton was working as intended for some in that it allowed them to identify their 'British-ness' first and then go on to identify a 'non-European' ethnic origin or heritage. Four per cent of respondents did not answer the national identity question.

National Identity Question - Frequency Counts1
Response	Frequency	%
Scottish	35,487	76.3
English	1,335	2.9
Northern Irish	161	0.3
Welsh	88	0.2
British	9,276	20.0
Irish	486	1.0
Other (write in)	980	2.1
No response	1,927	4.1

Note: 1 Percentages add to more than 100 per cent because some respondents answered this question giving more than one national identity. The percentages shown here include multiple responses.

67. The national identity question allowed respondents to provide multiple responses. The number of single and multiple responses given by respondents is shown below. Nearly 90% gave single responses and 7% gave two or more responses.

National Identity Question - Single and Multiple Responses - Frequency Counts
Number of ticks	Frequency	%
0	1,927	4.1
1	41,424	89.1
2	3,049	6.6
3	88	0.2
4	5	0.0
5	0	0.0
6	0	0.0
7	1	0.0

68. Respondents were asked if they were unhappy to answer any of the census test questions. Respondents identifying as 'British', 'English' and 'Other' were most dissatisfied with the national identity question.

Links: 2006 census test questionnaire (Scotland)

Evaluation of the 2006 census test (April 2007)

2006 census test follow up survey (2006)

69. Following the 2006 census test, GROS conducted a small follow-up survey to ask respondents about why they responded as they did and their views on the test questions; including those on national identity and ethnic group. Interviewees described their ethnicity in their own words and then re-answered the national identity and ethnic group questions included on the 2006 test. The ethnic group question used in the 2001 Census was then completed and question preferences were requested. Since these findings are based on small numbers and are not drawn from a representative sample they have been treated with caution.

70. The national identity and ethnic group questions did not appear to work as intended for some people identifying as 'Scottish Asian'. Nine (of 44) respondents, who identified as 'Asian: Pakistani' on the 2001 ethnicity question, selected 'European: Scottish' or 'European: British' on the 2006 question, even though they had already identified as 'Scottish' or 'British' on the national identity question. In effect, their 'Asian' ethnic heritage disappeared from the statistics.

71. It was intended for the above respondents to identify as 'Scottish' or 'British' on the national identity question and 'Asian' on the ethnic group question. The unintended responses probably occurred as 'Scottish' was under a 'European' heading in 2006 rather than a 'White' heading as in 2001 and since all references to 'Scottish' and 'British' were removed from the 2006 'Asian' category but were included on the 2001 'Asian' category. This demonstrated one effect on responses of removing references to colour and national identity from the 2006 ethnic group question.

72. Seven respondents identified as 'White: Scottish' or 'White: Other British' on the 2001 ethnicity question but identified as 'Asian' on the 2006 ethnic group question and 'Scottish', 'British' or 'Other' on the 2006 national identity question. These individuals may have benefited from the inclusion of a national identity question i.e. it allowed them to assert their 'Scottish-ness' or 'British-ness' before going on to identify as 'Asian' on the ethnic group question (as intended).

Link: 2006 census test follow-up survey (November 2006)

Analysis of write-in responses on the 2006 census test (2007)

73. Following the census test, GROS provided SG with the responses of those people who had written their ethnicity in the open boxes provided, rather than using the tick boxes provided. This information gave some indication of whether the category headings and tick boxes were interpreted as intended and also whether the tick boxes listed under each category were adequate. However since this analysis was based on small numbers and since the test was not drawn from a representative sample, these findings have been treated with caution.

74. The table below shows how respondents answered the write-in options under each ethnic group category on the 2006 census test. Three types of response were possible: a tick only, text only or tick and text.

75. Across all six ethnic group categories, most respondents chose to provide a text answer for the write-in options and very few provided a tick only. This may suggest that most respondents will write descriptions of the ethnicity when they don't use a tick box. This is helpful because it makes it possible to examine (and even publish) responses to each write-in box.

Ethnic Group Write-In Boxes	Type Of Response Given
	% Tick only (N=148)	% Text only (N=750)	% Tick and Text (N=1,103)	Total (N=2,001)
European: Other	6	39	54	100
Multiple ethnic groups	10	27	63	100
Asian: Other	6	32	63	100
Arab: Other	11	61	27	100
African or Caribbean: Other	18	56	26	100
Other ethnic group: Other	14	25	61	100

'European: Other' written responses

76. The 'European' write-ins were the most varied (297 different descriptions were given in total). Of the most frequent descriptions, 9 were 'non-European' ethnicities (suggesting that people were identifying here because they were born or had lived for some time in Scotland or another European country or that they were identifying distant European ancestry) . Of these, the most common were 'Asian' ethnicities, followed by 'African', 'Australian' and 'American' ethnicities.

77 . Three of the most frequent descriptions were ethnicities relating to people from Central or Eastern Europe, namely 'Polish', 'Latvian' and 'Albanian'. 'Polish' was the second most frequent write-in, after 'Asian'. Ethnicities described as 'White' (not intended) were also given, but infrequently.

78. Many respondents wrote in two descriptions for their ethnicity. Many of these were two UK ethnicities, for example 'Scottish: Irish', (expected in some cases), others wrote-in a UK and non- UK ethnicity, for example 'Scottish: Pakistani' (intended for the 'Multiple ethnic group category) and a fair number of respondents wrote two non- UK ethnicities, for example 'Polish: German' (expected in some cases).

'Multiple ethnic groups' written responses

79. The 'Multiple ethnic groups' category, by its very nature encourages people to write-in their responses. Most combinations of ethnic groups listed, were therefore expected.

80. Of the most frequent descriptions 'Scottish: Pakistani' was top. Other frequent ethnic group combinations given included: a UK and non- UK ethnicity e.g. 'British: Chinese', two UK ethnicities e.g. 'English: Scottish' (intended for the 'European' category) and two non- UK ethnicities e.g. 'Portuguese: Burmese'.

81. The colour term 'Black' was referenced several times e.g. 'Black: African', 'Black Scottish' but 'White' was referenced less often in this way e.g. 'White: Kosovan'. Some respondents wrote in a single ethnic group (not intended) but this was quite seldom.

'Asian: Other' written responses

82. Of the most frequent descriptions 'Sri Lankan' was top, followed by 'Malaysian', 'Afganistani' and 'Filipino'. Also responses such as 'Iranian' and 'Persian' were fairly frequent.

'Arab: Other' written responses

83. Of the most frequent descriptions, 'Kurdish' was top.

'African or Caribbean: Other' written responses

84. Of the most frequent descriptions, 'Somalian' was the top, followed by 'South African'. Few people chose to write-in 'Black'.

'Other ethnic group: Other' written responses

85. Of the most frequent descriptions, 'Somalian' was top, followed by 'Kurdish' 'Australian' and equally tied 'North American' and 'South American'. However, far higher numbers wrote in 'North American' in the 'European: Other' write-in box.

Wave one cognitive question testing - Scotland (2007)

86. Following a competitive tendering exercise, the review team commissioned Ipsos MORI to undertake independent cognitive question testing of a provisional set of ethnic group and national identity questions (see question 3, at Annex F), in June 2007. The religion question being developed for the next census was also included to test for any interaction effects, as people tend to relate it to ethnicity.

87. Cognitive question testing is being used across the UK to develop ethnicity classifications. It is a well recognised research technique used by survey designers to trial questions with members of the target audience to see if they are meaningful to respondents and function as intended. A benefit of this technique is that it mimics, to some degree, the situation whereby a census form is delivered to a person's home and they then complete it with minimal, if any, assistance. In this instance, the overall testing aims were to find out how different versions of Scotland's questions were interpreted by respondents and whether the questions elicited the required information. The acceptability of, and preference for, different category terms (particularly the 'Black', 'African', and 'Caribbean' category terms) were also explored.

88. Eighty-eight interviews were conducted with male and female members of the public from different ethnic backgrounds, covering a range of ages and locations in Scotland. Given the focus on acceptability of colour labels, in particular 'Black', a 'boosted' sample of 30 people self-identifying as 'Black', 'African' or 'Caribbean' was included. The sample also included people identifying as 'Asian, Asian Scottish, Asian British' (20), 'White Scottish, British' (10), 'White African' (7), 'Sikh'(7) and 'Eastern European' (14).

89. An iterative approach was adopted i.e. questions were modified and re-tested if significant problems arose in terms of respondents ability to answer the questions or where changes were desirable. Question 3 on page 26 shows the version of the question that was tested in the final iteration. Six slightly different versions of the form were used over the testing period.

90. One version of the national identity question was tested; 'Which nation or nations do you identity with most? Tick all boxes that apply.' A copy of the question is shown at Annex F. No changes were made during testing.

91. Two versions of the ethnic group question were tested:

1. `What is your ethnic group? Read the options below and tick ONE box to indicate your ethnic group'

2. 'What is your ethnic group? Choose ONE section from A to E, then tick the appropriate box to indicate your ethnic group'.

92. Changes were made to categories and tick boxes during the testing period for each iteration. A 'Polish' tick box was included under the 'White' category. The write-in boxes were indented to see if this visual cue made people recognise that this was a single question with five separate categories (many respondents thought the question ended after the first 'White' category). Then a further indentation was tested. Finally, each category was labelled 'A' to 'E' (for example, 'A White', 'B Mixed or multiple ethnic groups' etc) to see if this visual cue made people recognise that this is a single question with five separate categories.

93. As this was in-depth qualitative research, the sample was not intended to be statistically representative of Scotland. Rather, it was designed to reflect the range and diversity of respondents who potentially would have trouble working out the meaning or intention of the question or have difficulty locating a response option that they feel accurately describes them. Respondents' experiences of completing the test questions were reported along with some recommendations from the contractors

Findings on the national identity question

94. Most people understood the meaning of the question. However, there were some misinterpretations. A few thought it referred to the ethnicity or nationality of people they socialised with, a few thought it referred to countries they liked or would like to visit and a few thought it referred to their legal status i.e. citizenship or nationality. The word 'identity' caused some ambiguity and there was also some confusion about the meaning of the word 'nation'

95. Some respondents felt the question was important because it meant they could express their 'Scottish-ness' or 'British-ness'. Several respondents, and British respondents in particular, thought that the purpose of the question was to explore their views on Scottish devolution and independence. There was a view that immigrants should try to integrate and therefore the "correct" answer was 'Scotland' and/or 'Britain'.

96. Many respondents did not notice the instruction to 'Tick all boxes that apply' and then felt uncomfortable identifying with one nation only. The write-in box was frequently described as too short.

Suggestions for national identity

97. The contracted researchers suggested deleting the national identity question. If the question was deleted, their suggestion to include 'Scottish' and 'British' in the Asian response options on the ethnic group question ( e.g. 'Pakistani', 'Pakistani Scottish' and 'Pakistani British'') was considered even more important. If the question was retained, the suggestion was to increase the length of the write-in boxes and emphasise the instruction to 'Tick all boxes that apply'.

Findings on the ethnic group question

98. The main problem with this question was the layout. Many respondents thought the 'White' section represented the whole question because they interpreted the "Other white ethnic group" write-in box space as signalling the end of the question. They went on to write their ethnicity ( e.g. 'Black: African') in that space, rather than in a later section of the question as was intended.

99. Indenting the write-in boxes did not adequately correct for this perceptual difficulty and subsequent response errors. Each category was then labelled 'A' to 'E' and the completion instruction amended to reflect this. This visual cue was then tested to see if it helped respondents to perceive the question correctly (as having five categories) and then to respond using the most appropriate category.

100. At the point when the category labels were introduced, there were not enough interviews remaining with 'non-White' respondents to draw any conclusions about whether putting a letter in front of each section heading (A to E), solved the problem. However there appeared to be very little objection to the lettering. This was tested further in the second wave of cognitive testing (detailed below).

101. Most respondents were happy with the order of the categories and assumed it reflected the numbers of people from different ethnic groups in the Scottish population. Most respondents seemed to understand the term 'ethnic group'. However, some felt that the explanation of ethnic group in the 2001 Census would make it easier to understand the question.

102. Respondents who understood why ethnicity data are collected were more likely to accept the question. Virtually all respondents were comfortable with the 'White' category heading. Respondents who could identify as both 'Scottish' and 'British' sometimes found it difficult to choose between a 'Scottish' or 'British' tick box. All 'Polish' respondents were happy with the 'Polish' tick box. Some people from other 'Eastern European' countries supported a 'Polish' tick box whereas some others saw this as unfair.

103. Several respondents found the term 'Mixed' offensive and preferred 'Multiple ethnic groups'. Others were unfamiliar with this term and wanted to retain the 'Mixed' term. There was some confusion about who should be included in the 'Mixed or multiple ethnic group' category. Some thought it was for combinations of 'White' ethnic groups. Others thought it was for combinations of one 'White' ethnic group and a 'non-White' ethnic group. Some thought the 'Asian, Asian Scottish, Asian British' category was a mixed or multiple ethnicity and this also caused some confusion.

104. Respondents were generally happy with the phrasing of the 'Asian, Asian Scottish, Asian British' category heading as recognising the identities of 'Asian' people born and raised in the UK. Most were happy with the 'Asian' response categories but several UK-born people identifying as 'Asian' found it difficult to accept the response categories as they did not include any reference to 'British' or 'Scottish'.

105. Most 'Sikh' respondents ticked the 'Indian' response option rather than the 'Sikh' option and believed that the 'Sikh' option should be removed as they felt that Sikhism was a religion not an ethnicity. Some felt uncomfortable at being asked to choose between 'Sikh' and 'Indian'.

106. Under the 'African, Caribbean or Black' category, inclusion of a separate 'Black, Black Scottish or Black British' tick box caused considerable confusion, because it could be seen to imply that 'African' and 'Caribbean' people were not 'Black'. Those identifying as 'Black African' and 'Black Caribbean' were confused as to whether to tick 'African' or 'Caribbean' or 'Black' but most eventually choose 'African' or 'Caribbean' and not 'Black'. In most cases this did not stem from any objection to the word 'Black', though a few people did object to the term.

107. Within the 'African, Caribbean or Black' category, the inclusion of the terms 'Scottish' and 'British' against each tick box was generally seen as positive, but it did cause some confusion about whether, for example, 'African, African Scottish, African British' was a 'Mixed or multiple ethnic group'.

Suggestions for the ethnic group question

108. The contractors made a number of suggestions for the development of the ethnic group question based on their findings.

109. Additional testing of a question with category labels ('A' to 'E') with more respondents was suggested, to see if it functioned as an effective visual cue. In the completion instruction, consideration should be given to amending the phrase 'ethnic group' to 'ethnic background'.

110. They suggested consideration should be given to including an explanation about why it is necessary to collect information on ethnicity from the census. The 'Polish' tick box should be retained.

111. The phrase 'Mixed or multiple ethnic groups' should be retained as people are familiar with the term 'Mixed' (even though some find it offensive) but are as yet unfamiliar with the term 'Multiple'. However, consideration should be given to including an explanation of who should be in this group and consider moving this category so it does not appear after the 'White' category. It was suggested that the write-in box be extended so that people can adequately describe their ethnicity and that this category be tested with more people from this group.

112. Retention of the heading 'Asian, Asian Scottish, Asian British' was suggested but that consideration should be given to adding 'Scottish' and 'British' to the individual response options within this category, for example 'Pakistani, 'Pakistani Scottish' or Pakistani British'. It was suggested that the 'Sikh' tick box was removed from this category.

113. The contractors thought that consideration should be given to changing the 'African, Caribbean or Black' category heading back to the 2001 Census wording i.e. 'Black, Black Scottish and Black British' and to consider deleting the 'Black, Black Scottish or Black British' response option.

Links: Cognitive Question Testing Scotland's Census Ethnicity Classification - Main Report (2008)

Cognitive Question Test Scotland's Census Ethnicity Classification - Summary Findings (2008)

Wave two of cognitive question testing - Scotland (2008)

114. In early 2008, GROS commissioned Ipsos MORI to undertake cognitive question testing of some questions proposed for the 2011 Census. One aim, was to test some of the related questions with respondents, in context, i.e. questions which are planned to sit near each other on the census form. One of these was the ethnic group question and so this provided scope to test some further changes to the question following the findings of wave one testing earlier on.

115. As the end of the ethnicity classifications review was approaching, there was only a limited amount of time available to conduct this testing. Therefore the number of participants was kept small, focusing on outstanding testing issues. The overall aims were to find out if introducing the term 'background' to the completion instruction helped respondents understand the meaning of an 'ethnic group', to see if the category labels 'A' to 'E' acted as an effective visual cue for respondents, to test the new 'Asian' tick boxes and to test a 'Jewish' tick box in the 'Other Ethnic Group' category.

116. Overall, interviews were conducted with 88 members of the public and the ethnicity classification was tested with a sub-set (26) of these people.

117. Of the 26 people recruited, 7 self-identified as 'Jewish', 14 as 'Asian, Asian Scottish or Asian British', 2 as 'Polish' and 3 as 'African, Caribbean or Black'. A final version of the ethnic group question tested is shown at question 4 on page 27.

118. In this round of testing, most of the respondents interviewed seemed to understand the layout of the question. A few seemed to pause after they had read the 'White' section but then realised there were subsequent sections to choose from. The few respondents who did tick two response options were mainly the 'Jewish' respondents, an issue discussed below.

119. For the last four interviews, the write-in boxes at the end of each section were indented slightly. It was not possible to tell whether the indentation specifically had an effect on how the final respondents answered the question as it is not possible to differentiate between this change and the current layout using only four interviews.

120. It was suggested that the current layout of the question should be retained. This appears to be clearer to respondents than the previous versions tested. It was suggested that if someone responds in the 'White' section and again using another section, the first response should be disregarded. This is because evidence suggests that multi-ticking occurred due to the layout of the question.

121. Respondents felt that the ordering of the response categories ('White' first, etc) was appropriate and mentioned that this ordering was most likely due to the number of people from different ethnic backgrounds present in the population.

122. In this round of testing, there was no objection from respondents to the lettering before section headings ( e.g. 'A White', 'B Mixed or multiple ethnic groups', etc). The contracted researchers suggested that the lettering and ordering used in the question be retained.

123. Participants generally understood what the term ethnic group meant. Most described this as their parents' backgrounds, cultural background or skin colour. Some could not describe the meaning of the phrase 'ethnic group' but felt they understood what the question was asking as they have filled similar questions out so many times before.

124. There were mixed feelings among respondents about the use of the word 'background' in the instructions. Some felt that 'background' had a different meaning to 'ethnic group'. Others felt it aided understanding of the question. However, the term 'background' did not seem to confuse respondent's understanding. And it aided the understanding of some respondents. Therefore, it was suggested that the word "background" should be retained.

125. Most 'Jewish' respondents were confused by the inclusion of a 'Jewish' response option. They could not understand why they were being asked to state their religion again, as they had already been asked to do so in the preceding religion question (in the census, a religion question appears before the ethnic group question).

126. Nearly all of the 'Jewish' respondents felt that the 'Jewish' response option should be removed. The reasons given for these views were that Judaism is a religion not an ethnicity and that they felt uncomfortable having to chose between saying they were 'Scottish' or 'Jewish' when they are both. Some also felt that if 'Jewish' was included then other groups should be included such as 'Sikh', 'Muslim' etc.

127. Most of the 'Jewish' respondents ticked both the 'Scottish' response option and the 'Jewish' response option. It is difficult to untangle the reasons behind these choices. Respondents were familiar with choosing the 'Scottish' response option as an answer to this type of question and therefore did not look at the later sections in the question. However, some did not initially notice the 'Jewish' response option due to the layout issues discussed previously, and then noticed the response option after they had initially chosen the 'Scottish' response option. The layout issue is more difficult to clarify with 'Jewish' respondents as one of the response options they may choose to tick is in the 'White' section of the question.

128. Most respondents felt that if the 'Jewish' response option was to be retained, then it should be labelled 'Jewish, Jewish Scottish and Jewish British'. This was because they felt this would at least include some reference to Scotland in the response option. However, most felt they would still choose the 'Scottish' response option and preferred the removal of the 'Jewish' response option. The contractors suggested that the 'Jewish' response option be removed.

129. All 'Asian' respondents felt that the section heading was acceptable. The contractors suggested that the section heading should be retained in its current format.

130. Several participants mentioned the importance of including a reference to Scotland and Britain in the 'Asian' response options. These included most of the 'Asians' respondents who were born in Scotland or Britain, as well as those who had lived in Scotland/Britain for a long time but were born elsewhere.

131. 'Asian' respondents understood the meaning of 'Indian Scottish', 'Indian British', 'Pakistani Scottish', etc as someone who was born in Scotland or Britain but whose parents were from India, Bangladesh or Pakistan. A few respondents felt this included older 'Asian' participants, who had lived in Scotland/Britain most of their lives, as well as younger 'Asians' who were born and brought up in Scotland or elsewhere in the UK.

132. Some 'African: Caribbean' respondents felt that the 'African, African Scottish or African British' and 'Caribbean, Caribbean Scottish or Caribbean British' response options were for people from 'Mixed or Multiple Ethnic Groups'. Nevertheless, on balance, the contracted researchers suggested that the current wording of the response options in the 'Asian' and 'African, Caribbean or Black' section should be retained.