Making social science accessible

In the context of a substantial increase in the number of people in England identifying as ‘British’ in the 2021 census, John Denham and Ailsa Henderson explain how changing the survey question on national identity is likely to have influenced the results and why this matters for how we understand the dynamics driving our politics.

According to the Guardian, over half of those resident in England identified ‘as British only’ in the 2021 census. Behind this headline result the ONS itself warned, was the outcome of changes to the structure of the national identity question, with ‘British’ listed first among the list of options. Just ten years earlier when the ONS offered ‘English’ as the first choice and ‘British’ the fifth, 60% of England’s residents appeared to be ‘English only’. Neither result has ever been replicated in any other survey.

Those running social surveys know that there are multiple ways to ask about national identity. You can ask respondents to select any that apply (the question used in the 2021 census), or the single national identity that fits a person best.

Others prefer the Moreno question, which asks individuals to indicate the relative weight of English and British identity: English not British, More English than British, Equally English and British, and so on. The questions deliver slightly different results. In most surveys of opinion in England that use this question, ‘equally English and British’ is the largest group, yet only 15% of census respondents ticked both boxes.

In ‘forced choice’ questions, English and British have been in a dead heat for several years. One widely used social survey, the British Social Attitudes survey has shown only a gradual shift, primarily from ‘More English than British’ to ‘Equally English and British’ over the past 20 years.

Changing survey questions, even something as seemingly minor as changing the order of the response categories can influence the results. Changing response options – both the number and order – inhibits comparisons with older data. In a UK context, having a different response order across the state likewise inhibits comparison. Scottish and Welsh identity are listed first in the Scottish and Welsh census, so it is difficult for us to know if England feels more British than other parts of the state, or whether this is an artefact of the question.

Response order also matters. We know on election ballots this works to the advantage of candidates with surnames towards the top of the alphabet. We know it works on survey questions, but particularly we know it matters on national identity response options. Research from both Quebec and the UK shows that changing the order of the Moreno categories, so that they run British not English, More British not English, rather than the usual English not British etc advantages the categories listed first. Mention English national identity first and the figures jump. Mention British national identity first and the figures jump.

In England, the two dominant national identities are more tangled and perceived to be less in tension. When we ask people to list what makes them proudest to be English or British respondents tend to mention much the same thing. A large portion of the electorate feels strongly British and strongly English.

This tangling effect, particular to England, suggests that any order effect of the question would be stronger in England than in would be in, say, Scotland, where respondents would be more likely to search further down the list to find the Scottish label. This doesn’t mean that the English are more British, but instead that the relationship between the two identities is different, something which the survey question does nothing to illuminate.

To complicate things further, in England we cannot think of Englishness and Britishness as two primary colours to be mixed in different proportions. There is a Britishness that is closely associated with Englishness that often shares strongly held values of patriotism and is socially conservative, while the cosmopolitan Britishness of the ‘British not English’ often takes nation lightly and is socially liberal. Both differ on the extent to which they value the UK as a union.

It’s easy to see where the ONS went wrong. They first ignored their own rules that questions should change as little as possible from one census to another. More important they did not learn from 2011 that, unless explicitly offered mixed identities, most people will tick one box and quickly move on.

National identity matters for our understanding the dynamics driving our politics. National identities represent much more than the flag we might wave or the team we support. They carry ideas of the nation: its history, who belongs, how it should be governed, its understanding of sovereignty.

This is well recognised in analysis of Scotland and Wales but, though less widely discussed, is a powerful force in England too. These ideas of national identity can be politically mobilised, not necessarily by overt or explicit nationalism but equally well by appeals to the underlying ‘world views’ they hold. In the EU referendum, Leave was effective in connecting with the English feeling that sovereignty should be exercise at national level (and an English tendency to equate ‘Britain’ with a very English view of the UK).

Similar appeals gave Johnson’s Conservatives a dominating lead amongst English identifying voters in the ‘Get Brexit Done’ election of 2019 – an election where Corbyn’s Labour won amongst the ‘more British’.

Our measures of national identity have a real-world importance. If, as the ONS headline figures suggest, there has been both a big shift from ‘English only’ to ‘British only’ we might assume that the political divides around Brexit and attitudes toward the union have been decisively resolved. If being ‘English and British’ is now a small minority identity it might also be assumed that ‘British’ and ‘English’ have now evolved as two, largely distinct, identities. Neither is true. The interplay between different ideas of Britishness, Englishness, and Englishness with Britishness are likely to be more rather than less important in the future.

It’s possible to argue that the census is a poor vehicle to measure identity at all. It is, after all, a subjective question connected to identity, attitudes and values in a way that most other census questions are not. If the reality is that a national identity question must remain, then consistency, both over time and across the different census operating in the UK is critical.

The problem now, as the Guardian report shows, is that the erroneous figure will pass quickly into public misconception.

By Professor John Denham, Director, Centre for English Identity and Politics, University of Southampton; and Ailsa Henderson, Professor of Political Science and Co-Director, Future of England Survey, University of Edinburgh.


Sanchismo reloaded: what’s going on in Spanish politics?

Of course the pandemic was political: the Covid Inquiry and the constitutional question

Do Labour supporters back a softer Brexit?

Is immigration costing the Conservatives votes?

Has Reform UK really come of age?

Recent Articles

Subscribe to our newsletter

* indicates required