Scotland's New Ethnicity Classification for Scottish Official Statistics and Recommended for Scotland's 2011 Census
Scotland's New Ethnicity Classification for Scottish Official Statistics and Recommended for Scotland's 2011 Census
5. The New Ethnicity Classification and Supporting Evidence
5.1 This section sets out the new ethnicity classification and the evidence underpinning it. The question wording and completion instruction are detailed first, followed by the question categories and terminology. For each of these, a text box at the start of the sub-section summarises the key points about the new classification. Key findings from each piece of evidence are also set out, in more detail, at Annex C. The old and new classifications are shown on pages 26 and 28 respectively, together with some different versions tested during the review. Findings on national identity are summarised and the different versions tested so far are shown at Annex F. Some of the alternative approaches that were considered and tested are also explained, together with the evidence for why they have not been adopted. These include: a geographic classification with colour terms removed and inclusion of 'Sikh' and 'Jewish' as ethnic groups.
Changes to question wording and completion instruction
Question Wording and Completion Instruction
Question Completion and Category Labelling
|
Why consider the question wording and completion rules?
5.2 Collecting ethnicity statistics is not straightforward because there is no consensus about what constitutes an 'ethnic group'. In part, this stems from the absence of any legal definition in race relations legislation. However research tends to show that 'ethnicity' means different things to different people, has a number of possible meanings which can depend on the context or situation, and evolves over time.
5.3 For these reasons, ethnicity questions on past UK censuses have not defined an 'ethnic group' in any rigid or substantive way, though in 2001 references were made, in the completion instruction, to 'cultural background' in England & Wales and Scotland (but not in Northern Ireland). Instead, they have focused on developing acceptable response options through a process of consultation and research, to ensure that respondents understand the question and complete it accurately and that the response options deliver the key information needed by data users.
5.4 The wording (and completion instruction) of an ethnic group question must strike a balance between being meaningful and acceptable to respondents, eliciting the information required by data users and ensuring high quality data is generated. The review investigated these issues using consultation and research (including question testing).
5.5 The ethnic group questions in past UK censuses were restricted to a single response (tick) per person. The possibility and implications of allowing multiple responses were considered during the review. In the 2001 Census in England & Wales and Scotland each of the categories in the question was labelled 'A' to 'E'. The review also explored the need for category labelling and people's views on this issue.
What was tested and what were the findings?
5.6 The review team developed the question wording and completion instruction, over time, using repeated cycles of evidence collection and assessment. Several versions of question wording and completion instruction were considered and tested (see pages 26 to 28). Towards the end of the review some of these were tested with members of the public from a range of ethnic groups. The findings are set out below.
How did people define their ethnicity?
5.7 Consultation and research found that for some people national identity was a dominant factor of their ethnic identity whilst for others it was, for example, country of birth, parents' birth place, country of residence, heritage, religion, race, language or community affiliation. For many it was a combination of several of these factors.
5.8 Often, there were differences of opinion about what constitutes 'ethnicity' among people from the same and different ethnic groups. What is clear is that no single concept was identified as the definitive component(s) of ethnic identity; it is multi-faceted, often subjective, and complex and does not lend itself to a standard definition which fits everybody.
5.9 In the first wave of cognitive question testing, references to 'culture' and 'background' were removed from the completion instruction (see question 3, page 27) and respondents were asked simply to indicate their 'ethnic group'. Without these references, most respondents understood what was meant by the term 'ethnic group'. However, some did struggle with its meaning and suggested, instead, the term 'ethnic background'.
5.10 A reference to 'background' was then added back into the completion instruction in the second wave of cognitive question testing (see question 4, page 27) and respondents were asked to describe their 'ethnic group or background'. Participants generally understood the term 'ethnic group'. Some felt that the word 'background' had a different meaning to ethnic group; whilst others felt it aided their understanding of this term.
5.11 Few people saw 'culture' as a defining concept of their ethnic identity. When this reference was removed from the completion instruction, and tested, it did not appear to diminish people's understanding of the term 'ethnic group'. However, social and historical factors or events were considered, by some, to have a bearing on ethnic identity and its classification.
5.12 In summary, most respondents understood what is meant by the term 'ethnic group', without the need for further definition. Although no definition has been added, a reference to 'background' has been included in the completion instruction as this was found to aid understanding of what is meant by the term 'ethnic group' for some and many referred to it when self-defining their ethnicity. This was not true for references to 'culture' and this has been removed from the completion instruction.
Should respondents be allowed to give a single response (tick) or multiple responses (ticks)?
5.13 The 2001 Census ethnicity question allowed respondents to tick only one of 14 boxes to describe their ethnic group. During the consultation some people said that, if people were allowed to tick more than one box, it would enable them to describe their ethnicity more accurately - particularly if their ethnicity was multi-faceted.
5.14 The point was reinforced by the results of the 2006 census test because, although respondents were instructed to tick only one box, about 7 per cent (3,500) ticked more than one. That may mean that they had not read the completion instruction correctly - or that they felt that a multiple response gave a better description of their ethnicity.
5.15 There are benefits in allowing multiple responses. It would enable people with multi-faceted ethnicities to reflect this in their responses, in a way which they chose, rather than obliging them to fit into a preset categorisation. It would also remove the need for a separate 'Mixed or multiple' category. So the review team looked very carefully at the pros and cons of allowing multiple responses (ticks) instead of a single response (tick) and discussed these with the other UK census offices. These discussions, and further methodological work, clearly demonstrated that the disadvantages of this approach were significant and outweighed the benefits of self-expression - particularly in a survey like the census (which counts over five million people) or a survey by a large employer, which both produce a great deal of data.
5.16 The main disadvantages of multi-ticking were identified as:-
5.16.1 It makes it more difficult to complete the form. In 2006, the University of Kent carried out a small study to find out how people with a 'Mixed' ethnic identity describe and classify their ethnicity. Participants completed three versions of an ethnic group question: a write-in box, a series of 'Mixed' tick boxes and an option to multi-tick all applicable ethnic groups. The majority found the multi-tick version the most difficult to complete. Other work carried out by SG and GROS suggested that it would not be possible to distinguish between a genuine multiple response and certain response errors. For example, respondents might tick one option at the top of the question and then select a more appropriate later option (which they had not initially seen) without crossing out the earlier response - a particular risk when there are as many tick boxes as there are in the ethnicity question.
5.16.2 It would make it more difficult to count the 'Mixed or Multiple' ethnic group. This category is intended for people whose parents or grandparents are from different ethnic groups. But a multiple response option can be misinterpreted as inviting respondents to indicate both their current ethnicity and their distant ancestry many generations ago. This would not give data users the information they need: a measurement of current ethnicity (or fairly recent ethnic heritage) rather than distant ancestry.
5.16.3 It would produce an unmanageable number of outputs. If only a single response to the question is allowed, information is provided about 21 ethnic groups (plus write-in responses), which provides usable statistics. If people were allowed to tick as many boxes as they wished, there would be hundreds, and possibly thousands, of different combinations and the resulting statistics would be extremely difficult to publish or to use meaningfully.
5.16.4 It risks obscuring the size of the main ethnic groups. The ability to tick more than one box may cause people of the same ethnicity to identify using different categories, depending on their precise view of their ethnicity. While that has the advantage of allowing self-expression, it makes it difficult to obtain a robust single count of each main ethnic group in the population (a key requirement of data users). For example, people who ticked 'African' and 'Arab' would have to be included either in the 'African' or 'Arab' grouping for most outputs, and statisticians making that decision might not choose what the respondents regarded as their main ethnicity.
5.16.5 It would make it more difficult to publish statistics about small areas. One of the major strengths of the census in particular, is that it allows statistics to be published about very small geographical areas. But the census, and most other surveys, promise confidentiality. So great care is taken to avoid publishing results which allow individuals to be identified. The risk of this can increase if very small numbers are published - and if multiple ethnic group combinations were published (from multiple responses) there would be a greater risk of some combinations containing very small numbers. To avoid breaching confidentiality, such statistics could not be published - hampering users who want data about small areas, or want to cross-tabulate the answers from different questions (for example, ethnicity analysed by educational attainment and occupation).
5.16.6 It would make comparison more difficult. The 2001 Census permitted a single response only. If the 2011 Census permitted multiple responses, the results from the two censuses could not be compared, because the way people responded would have changed dramatically. If Scotland's census allowed multiple responses, but the censuses in the rest of the UK did not, it would also be impossible to produce ethnicity data for GB and the UK. The need for historical and cross- UK comparability is a key requirement of many data users.
5.17 In 1998 and prior to the 2001 Census, ONS undertook cognitive quesiton testing with 59 participants from 'non-White' ethnic groups on three different versions of an ethnic group question. Version 1 had seven categories (thirty-one tick boxes) and asked for a single response. Version 2 had five categories (twenty-three tick boxes) and asked for a single response and Version 3 had two sub-questions. The first had 6 tick boxes and asked for a single response and the second had 13 tick boxes and asked respondents to give as many responses as they liked (multi-ticks).
5.18 Some respondents found the multiple response question problematic because they failed to notice the instruction to multi-tick and then found it difficult to choose between all the options listed. Some used the multi-tick question to indicate their ancestry as well as their parentage. Some were unsure if they should tick 'British' and 'African' for example and others opted to tick 'British' but didn't tick 'African' (to specify their ethnic heritage) for example. In the last instance this would have made responses difficult to compare to the 1991 Census for the 'African' groups. Sometimes, people from 'Mixed' backgrounds answered the multi-tick question based on feelings of 'affinity' rather than ethnicity or ethnic heritage i.e. they used one of the multiple ticks to indicate the country they were raised in even though they or their parents were not born there.
5.19 Respondents were asked to say which of the three versions of the questions they preferred. Respondents identifying as 'Asian' widely criticised the multi-tick version but this seems to stem from the fact that they failed to notice the instruction to multi-tick and then found it difficult to choose between the response options listed. Respondents identifying as 'African', 'Caribbean' or 'Black' and who said they preferred the multi-tick version said it gave them more choice and made the question more specific. However, very few actually did tick more than one response and often people of similar backgrounds gave very different answers. People of 'Mixed' ethnic backgrounds did not prefer the multi-tick version. They considered it lengthy and time consuming. Some also mentioned that they would find it difficult to know where to stop. However one or two respondents preferred it because of this flexibility.
5.20 So the SG and GROS has decided that a single response (tick) will be required, as in the 2001 Census. But the number of categories has been increased from 14 to 21, giving people a wider range of choice. And people with particularly complex ethnicities can use the write-in boxes (particularly under the 'Mixed or Multiple' category) to add to the richness of the data.
What function does category labelling play and what do people think about it?
5.21 A small number of people objected to each category in the (2001 Census) classification being labelled 'A' to 'E'. The position of the 'White' category (at the top of the classification) followed by the 'Mixed' category and the remaining ethnic categories, is seen by some to imply a white-dominated racial categorisation. The labelling of the categories from 'A' to 'E' is hierarchical and is seen by some to reinforce this.
5.22 Based on these views and since it was unclear what, if any function, this labelling performed, the review team removed it from the ethnicity question tested in the 2006 census test (see question 2, page 26). However, the effect of removing category labels did not become apparent to the project team until cognitive question testing was conducted after the test. Cognitive testing allowed SG and GROS to specifically monitor and assess how individuals responded to a question with and without category labels.
5.23 At the start of wave one cognitive testing, the review team firstly tested an ethnic group question without category labels and then with labels added back in (see question 3, page 27). Without labels, it was evident that many respondents did not see (or realise) that the question has five sections. Instead, many thought that the question ended at the write-in box at the bottom of the first 'White' category. Many respondents wrote their answers here, only then realising that there were four additional response categories. Some went on to select a more appropriate category further down the list (crossing out their earlier response) but some did not. For these people the question was not functioning as intended, with people responding in error or using the least good category for them.
5.24 These findings were unexpected and the review team discussed them with the other UK census offices. It became clear that the classification presents a unique methodological challenge because it is unusually long for a survey question (the longest question on the census) and has more sections than most standard survey questions (in order to reflect the diversity of Scotland's ethnic profile). Respondents need help (visual cues) to guide them through the question and assist them to find the most appropriate response option.
5.25 The first visual cue tested cognitively was to indent each write-in box by one then two boxes (to try and visually distinguish each section of the question). This worked partially but many respondents still thought that the question ended at the first write-in box. The second visual cue tested was to label each category 'A' to 'E' and amend the completion instruction asking respondents to read sections A to E before responding. Most respondents seemed to understand the question with this layout and responded using an appropriate category. A very small number of those who took part in the question testing objected to the ordering of the sections but nobody objected to category labelling and most were content with the category ordering.
5.26 So categories have been labelled 'A' to 'E' as a visual cue to help respondents see that the question has five sections before they give a response. This will help respondents find the most appropriate response option for them, which in turn reduces the number of response errors and increases the quality of resultant data. For the same reasons, references to 'A to E' are included in the completion instruction. Most people involved in the question testing did not object to this category labelling or find it offensive.
5.27 The conclusion of this work was that, although some respondents may find the category labels 'A' to 'E' offensive because of their hierarchical nature, the use of the labels was the only effective way that could be found to ensure that people answer the question correctly - which is vital for data quality.
Why is it necessary to make the question and completion instructions stand out?
5.28 A recognised methodological challenge for the design of self-completion survey questions is that respondents frequently do not notice or read questions and instructions before giving their response. As expected, this happened in cognitive question testing with some respondents. For example, some respondents did not realise they could only give a single response (tick) and instead gave multiple responses, whilst others did not realise there were five categories.
5.29 Within the constraints of question layout for the census, several changes have been introduced to the completion instructions, to make it more prominent. The instruction to select ' ONE' category and choose ' ONE' response in that category have been capitalised and emboldened. The instruction have been re-worded to ask respondents to choose the option which ' best describes your ethnic group'. The words ' best describes' have also been emboldened and are designed to convey to respondents that they are being asked to make a choice, because multiple responses are not permissible (see question 5, page 28).
5.30 In summary, the text of the completion instruction is capitalised and emboldened (in parts) to help increase the number of respondents who see it and read it. In this way, it is hoped that the instruction to give a single response will be followed (to minimise response errors) and that respondents understand that they are being asked to choose one response option which best describes their ethnicity.
5.31 However, it is inevitable that some respondents will not read the completion instruction and will go on to give multiple responses or respond using a category that is not the best one for them. Such responses will need to be cleaned and re-coded appropriately before they can be outputted as data (further information on this will be provided in the guidance accompanying this report and which will be published in Autumn 2008).
Changes to question categories and terminology
White' Category
'Mixed or Multiple' Category
'Asian, Asian Scottish or Asian British' Category
'African, Caribbean or Black' Category
'Other Ethnic Group' Category
'Other' Tick Boxes and Write-ins
|
Why consider the question categories and terminology?
5.32 The main focus of the review was to consider what ethnic group categories and terminology should be used in the classification to meet the needs of data users and to reflect Scotland's current and future ethnic profile. It was important to ensure that the categories and terminology used are broadly acceptable to data providers and communities and that any changes do not have an adverse impact on respondents' ability to understand and use the classification or on the quality of the data it produces. It was also necessary to work with the other UK census offices to ensure that all UK classifications are harmonised, where necessary, but that they also meet the specific circumstances of each UK country.
What was tested and what were the findings?
5.33 Between 2004 and 2008, SG and GROS developed and tested several versions of an ethnic group question using research, consultation and question testing (pages 26 to 28 show a selection of the main versions tested). The evidence was used to assess the effect of any changes to categories and terminology on their acceptability to data providers and stakeholders, the usefulness of the data and the quality of the data. The overall aims were to:
- Explore the possibility of removing inconsistency from the classification in terms of race, national identity ( i.e. Scottish-ness, 'British-ness' etc) and geography.
- Consider whether concepts of race or colour are needed in the classification and whether this is acceptable to data providers and communities.
- Explore how the concepts of national identity and ethnic identity interact.
- Consider which ethnic groups may require a new or revised tick box and assess this against users' information needs.
- Devise terminology which is meaningful to respondents and is broadly acceptable to data providers and communities.
- Assess the impact of any changes to the classification on the quality and utility of resultant ethnicity data.
- Consider the operational impact that any changes to the classification would have for the census, Scottish Official Statistics (including surveys) and other organisations that may need to use the classification for the collection of ethnicity statistics.
How do national identity and ethnic identity interact?
5.34 A person's sense of national identity is often closely related to their ethnicity. The two concepts are inter-related in this context. Therefore national identity had a significant bearing on how the proposed classification was developed and, of course, the decision to develop a separate national identity question for the census and relevant Scottish Official Statistics, where feasible. The main findings on national identity and its relationship to ethnicity are detailed below and are referenced throughout the remainder of this section.
5.35 Consultation and research found, repeatedly, that many UK-born people whose parents or grandparents were born outside the UK considered 'Scottish-ness' or 'British-ness' to be central aspects of their identity and often their ethnicity. This was particularly evident among people identifying as 'Asian' who were born in Scotland or had lived in Scotland for a long time. This was also true of some people identifying as 'African', 'Caribbean' and/or 'Black'.
5.36 With this in mind, a separate national identity question was devised to allow people to express this aspect of their identity fully - be that 'Scottish', 'British' or any other national identity - before they went on to specify their ethnic origin or heritage. The question was devised to allow people to express any national identity they wished, whether or not that was 'Scottish'. The first version was included in the 2006 census test (see question 2 in Annex F).
5.37 The majority of test respondents identified as 'Scottish' (76%) or 'British' (20%), with 2 per cent identifying a national identity outwith the UK or Republic of Ireland. Although the question allowed multiple responses, 90 per cent of respondents gave a single response.
5.38 At the same time as this, ONS were conducting cognitive testing of a similar question. Based on these findings, SG & GROS decided to test another version of the question cognitively (see question 3 in Annex F) to see if it was easier for respondents to understand.
5.39 Cognitive testing yielded some positive results but also highlighted issues for further consideration. For some the national identity question functioned as intended, for example they answered 'Scottish' here and 'Indian' on the ethnicity question. However some respondents said they wanted to identify their 'Scottish-ness' or 'British-ness' as part of their ethnicity, regardless of the national identity question. These people found it hard to separate their national identity and ethnicity.
5.40 It was clear that the national identity question worked well for some in allowing a separate expression of this aspect of their identity before they went on to express their ethnic origin or heritage. However it was also clear that the ethnicity classification should, where possible, factor in concepts of 'Scottish-ness' and 'British-ness' for those who find it more difficult to separate their national identity from their ethnicity.
Would a national identity question help measure ethnicity?
5.41 Research and consultation found that many people identify as 'Scottish' or 'British' regardless of their ethnic origin or heritage, particularly those who were born in the UK (but whose parents or grandparents were not) or were not born in the UK but have resided here for a long time. It may be possible to avoid that confusion by separating out the concept of national identity and capturing it in a separate question altogether.
5.42 Two versions of a national identity question were tested (see Annex F) to allow people to express their national identity fully - be it 'Scottish', 'British' or any other national identity - before they identified their ethnic origin or heritage. It was hoped this would improve the quality of ethnicity data and aid self-expression. These questions were included in the 2006 census test or were tested cognitively. A similar approach was adopted across the UK.
5.43 The evidence suggests that the majority of people understood and did not object to the national identity questions. However the questions tested to date were shown to have some limitations. Some respondents interpreted national identity very widely and in different ways, taking it to mean country of birth, nationality/citizenship or countries they would like to live in, amongst other things. In part, this is due to the subjective nature of this concept but it may also reflect the question wording and response options used.
5.44 When the second re-worded question was tested (see question 3 on page 85), it was misinterpreted by some to mean affiliation to the Scottish National Party ( SNP) government or as a test of loyalty to Scotland for which there was a 'right' answer (with a potential here to introduce bias into the data). Some respondents continued to want to express their national identity, particularly their 'Scottish-ness' or 'British-ness' on the ethnic group question itself (regardless of a preceding national identity question). Data users tended to recognise that a national identity question aids self-expression of increasingly complex identities but some were unable to identify a need for this information beyond that. An emerging policy interest around national identity was identified as it is a possible indicator of social cohesion and inclusion (among people from different ethic backgrounds) and this was felt to merit further work to develop a robust question.
5.45 GROS, in partnership with SG, will continue to develop and test a national identity question to address these limitations, with the aim of making it available in Autumn 2008. The national identity questions shown at Annex F, are test versions only and do not represent the final version that will be used in the census or otherwise. The final version of the question, could be subject to substantial change to address some of the known limitations. This question will be recommended for next census (subject to Parliamentary approval) and will be included in relevant Scottish Official Statistics, where it is feasible to do so. ONS (and WAG) are likely to adopt a national identity question for the England & Wales census. GROS will continue to liaise with ONS (and WAG) when developing the national identity question. As stated previously, it is not possible to show ONS's findings on national identity in this report as ONS will not be reporting these until Winter 2008, when its testing programme will be complete.
5.46 Alongside the work to develop a satisfactory national identity question, references to 'Scottish' and 'British' have been included in the new ethnicity classification, in response to those consulted who wanted to identify their 'Scottish-ness' etc as part of ethnicity (whether or not a separate national identity question is asked). This approach has the additional benefit of allowing the ethnicity classification to function very effectively on its own (to provide users with the information they require) in those instances where it is not possible for users to ask a separate national identity question when collecting ethnicity statistics.
Is it acceptable to use colour to describe ethnicity?
5.47 The most complex and contentious issue (and one which led to the review) is the acceptability of using colour terms (particularly the term 'Black') to describe ethnicity. The review team gave a particular focus to the exploration of this issue using consultation (both formal and with stakeholders), research (in-depth interviews and focus groups) and question testing. This has been an important issue for all UK census classifications and other UK census offices adopted a similar approach.
5.48 The review team aimed to find out if it is acceptable to use colour terms to describe ethnicity and whether the inclusion of colour terms in an ethnicity classification serves a legitimate purpose for data users. They also tested classifications with and without colour terms (see question 1 to 5 on pages 26 to 28) and examined the effect of this on the quality of responses and resultant data. Findings are detailed below.
5.49 Consultation, research and question testing demonstrated, repeatedly, that there are opposing views on the acceptability of colour terms as ethnicity descriptors. Some people are in favour of them, whilst others are opposed to them. This was particularly the case regarding use of the term 'Black' to describe people with 'African' or 'Caribbean' ethnicities. People from both these ethnic groups expressed opposite views, with opposition to the term 'Black' tending to be stronger among people of 'African' ethnicity. Relatively few people expressed a view just on the use of the term 'White': most were opposed to the use of any colour terms.
5.50 Those opposed to the term 'Black' gave a range of reasons for this. Some said that skin colour refers to race and that this is separate from ethnicity. Others argued that racial terms such as 'Black' and 'White' are socially constructed concepts with no objective reality or validity.
5.51 Some people expressed concern that colour terms were used inconsistently in the classification in the 2001 Census i.e. the 'White' and 'Black' categories referred to colour terms but the remaining categories ( e.g. 'Asian') did not. This was seen by some as an unequal and divisive approach to ethnicity classification. The positioning of the 'White' category at the top of the classification, followed by 'non-White' categories, was seen by some to imply a white-dominated racial categorisation.
5.52 Those in favour of retaining the term 'Black' gave a range of reasons for this. Some believed that the term 'Black' was an accurate description of their skin colour or the culture or music they affiliate with. In this way, they felt that they would wish to use the term 'Black' to describe their ethnicity and they did not see it as relating solely to their race.
5.53 A range of people said they were proud to be 'Black' and so wanted the option to identify in that way. Some people born in the UK said they identified as 'Black', 'Black Scottish' or 'Black British' because they had never lived in or visited Africa or the Caribbean. Others felt that people should be given a choice to identify as 'Black' or 'White' if they wished but that this should not be imposed on anyone. Some people expressed a wish to identify as both 'Black' and 'African' or 'Black' and 'Caribbean'.
What purpose do colour terms play in an ethnicity classification?
5.54 In consultation, data users were asked if they need colour terms to be included in the classification. Many asked for colour terms to be retained, in order for the classification to provide data on visible ethnic groups. This would allow them to monitor and tackle discrimination and inequality on the grounds of visible difference (particularly skin colour, which is known to be a key trigger of discrimination). Some added that this would be essential for them in order to meet the statutory requirements of race relations legislation. However, some data users did not believe that colour terms are required in the classification to meet their statutory requirements, nor did they say they needed to measure visible ethnic difference.
5.55 A key requirement of data users is that the new classification should, at some level, be comparable with the old classification. This would enable them to produce ethnicity data over time in order to monitor the impact of policies to tackle discrimination and inequality. However, data users also recognised that the classification would benefit from modernisation. Some, but not all, data users consulted felt that a new classification without colour terms would be too different from the existing classification to allow comparability to be maintained at the level they required.
What effect would removing colour terms from the classification have on quality of responses and data?
5.56 One of the main criticisms of the 2001 census classification was its inconsistent use of colour and geography. For example the 'Black, Black Scottish or Black British' category used colour terms and geography but the 'Asian, Asian Scottish or Asian British' category used only geography. The review team responded by devising a revised version of the classification with colour labels removed and based on consistent geography (with the exception of an 'Arab' category). The 'White' category was replaced with a 'European' category and the 'Black, Black Scottish or Black British' category with an 'African or Caribbean' category (see question 2, page 26). A separate national identity question was asked before the ethnic group question where respondents could express their 'Scottish-ness', 'British-ness' etc (see question 2, Annex F). These questions were included in Scotland's 2006 census test (with approximately 50,000 Scottish households). This was developed at an early stage in the review, before the full range of evidence on the acceptability and user need for colour terms had been collected and assessed.
5.57 Analysis of the test results revealed that removing colour and national identity from the classification has a number of adverse effects on response quality and data quality. These are outlined in brief below and are discussed in the remainder of this section of the report.
5.58 Some 'Asian Scottish' or 'Asian British' respondents ticked 'Scottish' or 'British' on the national identity question and then ticked 'European: Scottish' or 'European: British' on the ethnic group question. In effect their 'Asian' ethnic heritage disappeared and became invisible in the data. However data users need this information to tackle discrimination and inequality and to deliver services.
5.59 The 'White' category was replaced with 'European'. This seemed to encourage some respondents - born in Scotland (or long term residents) or in Europe - to use this category over a more appropriate one, for example those with an 'Asian' ethnic origin. Others seem to have identified distant 'European' ancestry rather than their more recent ethnic heritage or origin. These response patterns became apparent from an analysis of responses given in the 'European: Other' write-in box. Of the top twenty most frequent write-in responses given here, nine were 'non-European' ethnicities , the most common of which were 'Asian', 'African', 'Australian' and 'American' ethnicities.
5.60 This meant that the 'European' category did not always allow data to be collected that users require. If this happened on the census (or a survey), these respondents would need to be identified and re-grouped (where possible) before their responses could be outputted as ethnicity data. This is time and resource intensive and increases the risk of data errors. Additionally it would not be possible, for some respondents, to identify where this had occurred, for example those who ticked 'Scottish' rather than those who wrote in 'North American' etc. It is incumbent on SG and GROS to devise a classification which maximises data quality (and usefulness) before it is processed for outputting.
5.61 Testing a question without colour terms prompted some people who felt strongly that colour terms should be retained to contact the review team. These views were explored further and it was found that views on the acceptability and need for a colour-based classification are polarised, among both data providers and data users. Some are in favour of retaining colour terms whilst some are opposed and others are ambivalent to their inclusion.
5.62 Given the issues outlined above, a decision was taken to re-introduce the terms 'White' and 'Black' into the classification. However, it was decided to re-introduce (and test) the 'Black' term in a different way from the last census, to try and achieve a compromise between the opposite views of some data users and data providers (see question 3, page 27). This is discussed in more detail below.
5.63 A robust classification should try to minimise people from the same ethnic group identifying using different categories, in order to derive as robust a count of that group as possible. Many data users requested that the new classification be comparable with (though not necessarily identical to) the previous classification, so that they can monitor ethnicity trends over time. Removal of colour labels would prevent this.
5.64 These findings show some of the difficulties which arise when attempts are made to remove inconsistency, by classifying all ethnic groups in a standardised way. Different ethnic groups identify using different terminologies and based on different histories. In some cases, they experience different inequalities and types of discrimination which the classification must make visible in the statistics to allow these to be tackled. Consistency has been added when this did not comprise data quality (for example the addition of the term 'Scottish' and 'British' throughout the classification). However, SG and GROS have not imposed the terminology of one ethnic group on another in an effort to remove inconsistency, particularly where this would diminish data quality.
Why are the categories and tick boxes in this order?
5.65 Categories and tick boxes are not listed alphabetically. In general, their ordering is based on population size in Scotland (with larger groups appearing higher up the list) or in order to maximise the quality of responses by reducing response 'errors' ( i.e. a methodological reason). In some instances another factor, specific to a category or tick box, was taken into account. Examples of these ordering principles are provided below.
5.66 Based on population size the 'White' category is listed first; followed by 'Asian, Asian Scottish, Asian British', then the 'African, Caribbean or Black' category and 'Other Ethnic Group' listed last. Similarly the 'Scottish' tick box is placed at the top of the 'White' category and the 'African' tick box at the top of the 'African, Caribbean or Black' category etc. according to population size.
5.67 Testing across the UK has shown that respondents have a tendency to only read down the classification as far as the first tick box they find suitable. So the ordering of categories and tick boxes makes a difference to the responses given. The 'Mixed or Multiple' category is positioned near the top of the classification, after the 'White' category. This is because question testing showed that these respondents were likely to miss this category if it was placed at the bottom of the classification (since they had responded using one or more of the tick boxes higher up the classification).
5.68 Additionally, the positioning of the 'Polish' tick box before the 'Any other white ethnic group' write-in box is to signpost that respondents from other countries which joined the EU since 2003 should give their response here. The 'British' tick box is positioned after the UK tick boxes to encourage respondents to tick these, since data users are more interested in counts of specific UK groups than a count of 'British'.
5.69 There are several tick boxes which are positioned for another reason. The 'Chinese' tick box is positioned at the bottom of the 'Asian, Asian Scottish or Asian British' category for continuity with the 2001 Census (these respondents are used to seeing it here) and since it is less connected culturally and historically to Pakistan, India or Bangladesh. The 'Black, Black Scottish or Black British' tick box sits at the bottom of the 'African, Caribbean or Black' category because, as a new tick box, the size of this population is unknown.
5.70 SG and GROS also discussed category and tick box ordering with the other UK census offices to ensure that, where possible, they are harmonised across UK classifications but also meet the specific circumstances of each UK country ( i.e. the different size of ethnic group populations across the UK).
Changes to the 'White' category
5.71 A variety of changes have been developed and tested for the 'White' category. It has been expanded to include new tick boxes for 'English', 'Welsh', 'Northern Irish' and 'British' (see question 5, page 28). Previously, only 'Scottish', 'Other British' and 'Irish' tick boxes were listed.
5.72 Research suggests that the 'British' identity is declining in England and Scotland and that people are increasingly identifying as 'Scottish' or 'English' rather than 'British'. For example, the New British Social Attitudes Survey (2007) 13 found that in 1974 three in ten people (31%) in Scotland said they were 'British' and now half this proportion (14%) do so. The proportion choosing 'Scottish' has gone up from 65 per cent to 78 per cent. The findings were similar for England.
5.73 However, question testing by SG and GROS shows that a fair proportion of Scottish residents continue to identify as 'British' when given the option. For example on the 2006 census test, 20 per cent of respondents identified as 'British' on the national identity question and 9 per cent identified as 'European: British' on the ethnic group question (it should be noted that the sampling used for the census test means these proportions are not representative of the population as a whole - see Annex C paragraph 59).
5.74 In this way, the new UK tick boxes allow those with 'White' ethnicities to identify as 'Scottish', 'English', 'Welsh' etc, but do not force people to identify in this way by providing a separate 'British' tick box. Given that research suggests a continuing, long-term trend away from the 'British' identity, these changes should mean that the classification will become increasingly relevant in future years and when the next census takes place in 2011.
5.75 SG and GROS research found that some data users wanted all 'British' nationalities to be included in the classification, with some users reporting that sub- UK tick boxes would better help them meet their data needs. Some stakeholders and data users reported that the UK tick boxes could potentially enable them to use the classification to monitor statistically, for the first time, the level of anti-English discrimination occurring in Scotland, for instance if the classification was used in a crime survey or to collect statistics on crime
5.76 Cognitive testing found that almost all respondents were comfortable with a 'White' category heading. The few who objected to it, objected to all colour terms in the classification. Some respondents found it difficult to choose between the 'Scottish' or 'British' tick box but this did not prevent them making a choice.
5.77 A new 'Polish' tick box has been included in the 'White' category. The review team undertook consultation and research to find out what information is required by data users on people who have recently migrated from central or Eastern Europe and how these people classify their ethnicity. The available sources of data were Accession Monitoring Reports 14 produced by the Home Office, analysis of the 2001 Census and 2006 census test, consultations with data users and cognitive testing.
5.78 Home Office statistics show that since May 2004, around 70 per cent of migrants from the first Accession countries (A8) to Scotland were 'Polish' nationals. Responses given on the 2006 census test show that, in general, people from these countries used national identity, nationality or country of birth to describe their ethnicity i.e. 'Poland'/'Polish', rather than the term 'Eastern European'.
5.79 Data users consistently reported a strong demand for information on recent migrants from A8 countries. Many believed that this information could be best captured on the census using a number of questions including ethnic group. Repeatedly, users said they needed information on recent A8 migrants for language and housing provision and to monitor discrimination. The need for this information is particularly strong because there is a lack of reliable, alternative sources of statistics at Scotland and sub-Scotland level.
5.80 The next census will include a set of questions aimed to capture information specifically on migration. Final decisions have not yet been made, but questions currently proposed for 2011 include: country of birth, month and year of entry to the UK, citizenship ( i.e. passport entitlement) and usual address one year ago. The ethnic group question and a new language question can be added to this question set to provide data users with the kind of migration information they identified above.
5.81 The new 'Polish' tick box was tested cognitively by the review team (see question 4, page 27). 'Polish' respondents were content with this tick box, though some respondents from other areas of Central and Eastern Europe thought it was unfair to single out 'Polish' in this way. Whilst A8 nationals have migrated to Scotland from a total of eight countries, there is not sufficient space on the census form to include a separate tick box for each of the groups.
5.82 It was decided to test a 'Polish' tick box only because this is by far the largest A8 group resident in Scotland and since a broader 'Eastern European' tick box would not meet users' specific information needs. Allied to this, an analysis of verbatim responses given on the census test shows that people from these groups do not identify their ethnicity using the term 'Eastern European'. As such they may find this term offensive or unmeaningful.
5.83 The 'Polish' tick box is positioned above the 'Any other white ethnic group' write-in box to signpost to respondents from other A8 countries that they should provide their response here. Following the 2011 Census, GROS has agreed to output counts of these groups (subject to data quality and disclosure rules) in order to meet the information requirements of data users.
5.84 A new 'Gypsy/Traveller' tick box has been included in the 'White' category. This was developed and tested because there is a very strong requirement for information on this group among data users. This is because there are few robust counts of this group (at national and local level) even though they are known to experience high levels of discrimination, deprivation and inequality.
5.85 The term 'Gypsy/Traveller' is the official term used by the SG following consultation with these communities. As with some of the other terms used in the classification, 'Gypsy/Traveller' is an umbrella term used to describe a number of different ethnic groups.
5.86 Because this is a small group in the general population, it is necessary to use an umbrella term (as opposed to more specific tick boxes for the individual groups) to avoid small numbers and the difficulties this creates for useful analysis of results. Overall, the tick box has been welcomed by data users and stakeholders but some felt that 'Gypsies' and 'Travellers' are two distinct groups who should not therefore be joined together using a single term.
5.87 A decision was taken to position the tick box under the 'White' category, rather than the 'Other ethnic group' category. Some stakeholders advised this, however, the main reason was to ensure that these respondents did not tick a box under the 'White' category (such as 'Scottish') at the top of the classification causing them not to respond using a 'Gypsy/Traveller' tick box at the bottom of the classification. Therefore its positioning under the 'White' category is designed to maximise response rates and the quality of data for this group.
5.88 Some data users said that the term 'Gypsy/Traveller' may not be adequate for people wanting to identify as 'Roma' or 'Romany', many of whom are known to have migrated to Scotland recently. This group is known to face high levels of discrimination and so data users are keen for specific information on this group. As this is a very small group in the general population and the available space on the census form is limited, it is not possible to provide a separate tick box for this group.
5.89 Respondents identifying as 'Roma' or 'Romany' may tick the 'Gypsy/Traveller' tick box or they may write-in their ethnicity in one of the 'Other' write in boxes. It may be possible to distinguish between 'Roma' and 'Gypsy/Traveller' on the census by analysing the results of the ethnicity question together with the country of birth question. Given this, GROS may be able to produce a count of this group (subject to data quality and non-disclosive data) in order to meet the information requirements of data users. If this is not possible, it may be necessary for data users to conduct specific research or a survey, at a local level, to gather information on this group.
Changes to the 'Mixed or multiple ethnic groups' category
5.90 Early on in the review, research and consultation by the review team found that people's views on the acceptability of the term 'Mixed ethnic group' were opposing. Some are in favour of it, whilst others are not. The term 'Multiple' was suggested as an alternative by those not in favour of the term 'Mixed'. For the 2006 census test, the review team re-termed the existing 'Mixed ethnic group' category to 'Multiple ethnic groups' (see question 2, page 26). Subsequent consultation revealed that some people were confused by the term 'Multiple'. In part, this was because the term 'Mixed' has become the accepted term which some people from these ethnic groups identify with, expect to see and understand.
5.91 A further term was therefore devised: 'Mixed or multiple ethnic groups' (see question 3, page 27) and this was tested cognitively. Use of the word 'Mixed' continued to be viewed unfavourably by several respondents, who said it could be offensive. Others favoured the word 'Multiple' instead. Some respondents said they were unfamiliar with the term 'Multiple' and that the word 'Mixed' should appear next to it, to explain its meaning. Following testing, the contracted researchers suggested that the term 'Mixed or multiple ethnic groups' could be used for the time being and as people become familiar with the word 'Multiple' it may be possible to phase out the word 'Mixed'.
5.92 Cognitive testing also revealed that there was some confusion about who should be included in the 'Mixed or multiple ethnic groups' category. In part, this was due to its position after the 'White' category and before the remaining categories. However it was also due to use of the terms (used in the 'Asian' and 'African, Caribbean or Black' categories) 'Indian Scottish' and 'Caribbean Scottish' etc. To some people these looked like 'Mixed or multiple' ethnicities.
5.93 The researchers suggested adding a definition of this group and repositioning the category at the bottom of the classification. However, as discussed previously, the position of this category is to maximise response rates and minimise response errors. Equally, a constraint of the census is that the meaning of the questions should be self-evident because definitions require additional space on the census form and this space is restricted for each question on the census.
5.94 Cognitive testing also revealed that the single write-in box in this category was too short to allow respondents to write-in more than one ethnicity. Two write-in boxes are provided on the new classification (see question 5, page 28). This provides more space and the dual write-in box also signifies, visually, that more than one ethnicity should be entered here.
Changes to the 'Asian, Asian Scottish or Asian British' category
5.95 All the 'Asian' tick boxes have been re-worded and re-ordered. References to 'Scottish' and 'British' are included against each tick box, for example, 'Bangladeshi, Bangladeshi Scottish or Bangladeshi British'. As previously explained, tick boxes have been re-ordered, in part, to reflect the population size of these ethnic groups in Scotland.
5.96 Until the later stages of the review, the wording of the 'Asian' tick boxes remained largely unchanged because research and consultation found that, in general, people accepted this category and they did not seek to changes to it. In part, these changes to the new 'Asian' category were made because some 'Asian' respondents said that their 'Scottish-ness' was an important facet of the ethnic identity as well as their national identity. If data users are not able to ask a national identity question, for example due to space on their surveys or operational constraints, the ethnicity classification will continue to reflect many people's sense of 'Scottish' and 'British' national identity when it is used as a standalone question.
5.97 As detailed earlier, it is clear that many UK-born people of 'Asian' ethnic heritage, feel that their 'Scottish-ness' or 'British-ness' is an important facet of their ethnicity. However, data users require information on people's ethnic heritage in order to tackle discrimination and inequality and deliver services effectively to different ethnic groups. As with the other categories, these changes required a balance to be struck between self-expression on the one hand and users' data needs on the other, when developing the 'Asian' category.
5.98 As detailed in paragraphs 5.34 to 5.46, a separate national identity question is currently being developed for the census and the SG's key surveys. When tested, previous version of the question did not always function as intended. In part, this was because some respondents saw their national identity as part of their ethnicity. This was particularly evident for some 'Asian' people. Given, this, the review team had to consider some way of capturing 'Scottish-ness' and 'British-ness' within the ethnic group question, regardless of whether or not a separate national identity question was asked.
5.99 The first wave of cognitive testing also found that, whilst the majority of 'Asian' respondents felt the existing category and tick boxes to be adequate (see question 3, page 27), they said that the tickboxes could be improved by adding references to 'Scottish' and 'British', for example, 'Pakistani' becoming 'Pakistani, Pakistani Scottish or Pakistani British'. This was also seen to mirror the tick boxes under the 'African, Caribbean or Black' category. The researchers also suggested that this change was even more advisable in the event that a national identity question was not asked.
5.100 These changes were introduced and tested in the second wave of cognitive testing (see question 4, page 27). On the whole they were positively received by 'Asian' respondents, who felt that including references to Scotland and Britain was important for those 'Asians' who were born in the UK or who had lived here a long time but were born elsewhere. They understood the meaning of 'Indian Scottish', 'Bangladeshi British', 'Pakistani Scottish' etc as someone born in the UK whose parents were from Indian, Bangladesh or Pakistan or who has one 'Asian' and one 'Scottish' parent.
5.101 The researchers suggested retaining the new wording. By including a national identity question in the census and relevant Scottish Official Statistics (where feasible), all 'Asian' respondents will be able to specifically identify their national identity fully - be it 'Scottish', 'British' or any other national identity - on this question before they are asked to specify their ethnic origin or heritage.
Changes to the 'African, Caribbean or Black' category
5.102 The previous 'Black, Black Scottish or Black British' category has been re-worded to 'African, Caribbean or Black' and a new tick box has been added for 'Black, Black Scottish or Black British'. References to 'Scottish' and 'British' have been included in the remaining tick boxes i.e. 'African, African Scottish or African British' and 'Caribbean, Caribbean Scottish or Caribbean British'.
5.103 These changes have been introduced in order to accommodate (as far as possible) the opposing views which were expressed in consultation and research on the acceptability and user need for colour terms to describe ethnicity (as detailed above). References to 'Scottish' and 'British' have been introduced for the same reasons as their introduction to the 'Asian' category.
5.104 The compromise position adopted by the review team is to provide respondents with the option to identify as 'Black' should they wish but not to impose this on other people who want to identify as 'African' or 'Caribbean'. This was not possible on the previous classification. As a compromise, these changes do have a number of limitations on data quality and not everybody is likely to agree with them.
5.105 Cognitive testing found that some 'African' and 'Caribbean' respondents were confused by having a separate 'Black' tick box, in that they were unsure who this was meant for. Some wanted to identify using two tick boxes such as 'African' and 'Black' for example, but realising that only a single response is permitted, the majority of those involved in the testing chose to tick 'African' or 'Caribbean' rather than 'Black'. A few respondents objected to the term 'Black'.
5.106 References to 'Scottish' and 'British were introduced into each tick box under the 'African, Caribbean or Black' category (see question 3, page 27) for the same reasons as the 'Asian' category. In cognitive testing, most respondents from the 'African, Caribbean or Black' group welcomed this change, but some believed that there should be three separate tick boxes for 'African', 'African Scottish' and 'African British' etc. However, in order to accommodate this, the category would have to extend to ten tick boxes rather than four and this is not possible due to space constraints on the census (and other surveys). By including a national identity question on the census and relevant Scottish Official Statistics (where feasible), all 'African', 'Caribbean' and/or 'Black' respondents will be able to express their national identity fully - be it 'Scottish', 'British' or any other national identity - on this question before they are asked to specify their ethnic origin or heritage.
Changes to the 'Other Ethnic Group' category
5.107 A new tick box for 'Arab' has been included in the 'Other Ethnic Group' category. In the 2001 Census, people identifying as 'Arab' did not have a tick box but they were the largest group (1,959) to write-in their ethnicity. Since this is a sizeable ethnic group and one which is likely to have grown since 2001, a decision was taken to include a new tick box for this group.
5.108 In addition, consultation showed that data users expressed a need for information on this group. Yet it is difficult for them to obtain a robust count using the 2001 classification because 'Arab' respondents wrote in their responses across all five ethnic group categories (probably due to the wide geographic spread of 'Arabic' countries).
5.109 The need for information on this group is also heightened because they are known to have experienced greater discrimination since the terrorist attacks in the USA on 11 September 2001, in London on 7 July 2005 and at Glasgow International Airport on 30 June 2007. The number of 'Arab' people living in Scotland has increased in recent years both prior to, and since, the war in Iraq and this group may have distinct service needs.
5.110 The terminology for this tick box was decided based on consultation with stakeholders and by analysing the write-in responses of people identifying as 'Arab' on the 2001 Census. Stakeholders said the term 'Arab' was more acceptable, accurate and useful than the term 'Middle Eastern' (which refers to a political geography). Analysis of write-in responses on the last census also bears this out to some extent, as people were far less likely to describe themselves as 'Middle Eastern' than as 'Arab'.
5.111 A decision was taken to term the tick box 'Arab' rather than 'Arab, Arab Scottish or Arab British' (as with some of the other tick boxes). Analysis of the write-in responses on the 2001 Census results showed that the majority of 'Arab' people did not identify as 'Arab Scottish' or 'Arab British' and those who did tended to give their response in the 'Mixed ethnic group' category. In the main, however, people tended to self-identify as 'Arab' or 'Arabic' or as 'Arab Syrian', 'Saudi Arab', 'Arab Iraq' for example. By including a national identity question on the census and relevant Scottish Official Statistics (where feasible), all 'Arab' respondents will be able to express their national identity fully - be it 'Scottish', 'British' or any other national identity - on this question before they are asked to specify their ethnic origin or heritage.
5.112 Earlier on in the review, stakeholders recommended that a 'Middle East' and 'North African' tick box be included under an 'Arab' category and this was included in the 2006 census test (see question 2, page 26). This formulation was also in line with the geographic re-configuration of the ethnic group question at this stage. However further consultation with stakeholders revealed that a single 'Arab' tick box should be sufficient to capture the information required rather than a separate category. Given the constraints of space on the census form, the review team came to a decision to include one tick box only.
Did you test tick boxes for any other ethnic groups?
5.113 Although 'Sikh' and 'Jewish' will be included as specific groups in the census question on religion, some stakeholders from both communities asked for 'Sikh' and 'Jewish' to also be classified as ethnic groups, as they face discrimination regardless of whether they practice their religion. They also pointed out that they are recognised as ethnic groups following judicial interpretation of race relations legislation.
5.114 The review team tested 'Sikh' and 'Jewish' tick boxes in the 2006 census test and did additional testing with members of the 'Sikh' and 'Jewish' public (see questions 2, 3 and 4 on pages 26 to 27).
5.115 On balance, the evidence does not support the inclusion of 'Sikh' and 'Jewish' in the new classification, for several reasons. When the question was tested with a small number of 'Sikh' and 'Jewish' members of the public they were confused by inclusion of these tick boxes, which they saw as religions not ethnicities. They found it hard to choose between the 'Sikh' or 'Indian' tick box and the 'Jewish' or 'Scottish' tick box, often ticking both in error. Census questions must be answered without detailed instructions, by the entire Scottish population, so questions must aim to be easy to answer and avoid causing confusion where possible.
5.116 If people choose 'Indian' over 'Sikh', 'Scottish' over 'Jewish' and vice versa, all four populations are undercounted. Since a large number of 'Sikhs' are 'Indian' and a large number of 'Jewish' people are 'Scottish' this may produce substantial undercounts for the smaller 'Indian', 'Sikh' and 'Jewish' groups (but would have a lesser effect on the larger 'Scottish' group). Undercounts cannot accurately inform resource allocation, discrimination monitoring or service provision. Some 'Sikh' stakeholders are opposed to 'Sikhism' being classified as an ethnicity as they argue that the faith was founded as a universal religion, inclusive and open to all ethnic groups.
5.117 None of the data users consulted identified specific needs for information on 'Sikh' and 'Jewish' ethnicity, and some were confused by its inclusion in the census ethnic group question rather than the census religion question. However, this may be because data users would be able `to gather information on 'Sikh' and 'Jewish' groups from the census religion question.
5.118 While these groups have not been included in the classification, they are known to face discrimination, which must be monitored. In the guidance which will follow this report in Autumn 2008, SG will recommend that organisations consider asking both a religion and ethnicity question when collecting ethnicity statistics, even though collecting statistics on religion is not currently a statutory requirement for public authorities (as is the case for ethnicity). The main SG official surveys will ask both questions as will the 2011 Census.
5.119 This would facilitate self-expression but would also allow data users to cross-tabulate the religion, ethnic group and other survey questions to produce rich analysis of any different outcomes for these groups. If they wish to, 'Jewish' people can identify as 'Jewish' ethnicity in the write-in boxes provided on the classification. The same is true for those who wish to identify as ethnically 'Sikh'. The write-in boxes provide this flexibility. It has been recommended to GROS, that they output statistics on the number of people who write-in 'Sikh' and 'Jewish' (subject to sufficient numbers to avoid disclosure).
What are the limitations of the new classification?
5.120 Whilst the new classification has been adopted to ensure that the question wording and completion instruction is meaningful and acceptable to respondents and produces high quality data to meet the main information needs of data users, there are clearly a number of limitations.
5.121 In the 'White' category, some respondents may find it difficult to choose between tick boxes where they have a choice, for example the 'Scottish' tick box and the 'British' tick box or the 'Scottish' tick box and the 'Gypsy/Traveller' tick box. This is a matter of personal preference on the part of the respondent and they may have less difficulty in making this choice when a preceding national identity question is asked. However, because people of the same ethnic group might tick different boxes, this might produce a slight undercount of each group. This is most likely to affect the 'British' population count since it is known that the 'British' identity is declining. Because respondents have a choice they may be more likely to multi-tick (in error), if they don't notice the instruction to provide a single response only.
5.122 Although the new classification gives respondents the choice to identify as 'Black' or not, some of those who object to colour terms may continue to object to this formulation and may elect not to answer an ethnic group question using this classification. There may also be objection to the fact that colour is not used consistently across the ethnic group categories. Again some people may choose not to answer the question on this basis. If this were to happen, it would produce an undercount of the 'African', 'Caribbean' and 'Black' communities. This would have an adverse impact because the census is used to deliver services and allocate resources to these communities, as well as to tackle discrimination and inequality.
5.123 Some people may use the 'Black' tick box to identify as 'Black' in a political way rather than ethnically i.e. interpret the 'Black' tick box to mean all minority ethnic groups. Based on the review's findings, the likelihood of this occurring is fairly small, however it would incorrectly increase the 'Black, Black Scottish or Black British' population count.
5.124 People of 'African' and 'Caribbean' ethnicity have the option to identify as 'Black' if they wish. Where this happens, this will produce a slight undercount of the 'African' and 'Caribbean' populations at tick box level, however the population count for the overall 'African, Caribbean or Black' category will not be undercounted. Yet this had to be balanced, against the need to ensure that the classification makes provision for people to identify as 'Black' if they wish (which was expressed clearly in the evidence base).
5.125 There is a small chance that some 'White Africans' will answer using the 'African' tick box rather than the 'White: Other' tick box (as intended). However, in specific question testing with a small number of 'White Africans', the majority answered using the 'White: Other' write-in box.
5.126 An additional limitation of this category is that respondents will not be able to identify, specifically, as 'African Scottish' or 'Caribbean British' for example because it is only possible to provide tick boxes which read 'African, African Scottish or African British' or 'Caribbean, Caribbean Scottish or Caribbean British'. It is not possible to allow this degree of self-identification because this would require the category to include 10 tick boxes and there is not sufficient space to provide for this on the census form. The same holds true for the 'Asian' category, which would have required 13 tick boxes to enable this. However, the preceding national identity question will allow respondents from both groups to express their 'Scottish-ness', 'British-ness' (or any other national identity) fully, before expressing their ethnic origin or heritage.
There is a problem
Thanks for your feedback