Key findings

• • Ethnic group is not a fixed characteristic during a person's life.

• • Four per cent of all people chose a different ethnic group in 2011 than in 2001. This is twice the level of instability found for the decade 1991 to 2001.

• • Among the main ethnic group categories, 1 per cent of White British in 2001 changed their ethnic group when asked again in 2011. Four per cent of Bangladeshi and 26 per cent of White Irish groups changed their ethnic group.

• • A larger proportion of those in Mixed and ‘Other’ ethnic groups changed their recorded category (for example, 43 per cent of ‘Mixed White & Black African’ in 2001 identified with another census category a decade later).

• • This should concern official statistical agencies responsible for setting ethnic group response categories in censuses and surveys, because ‘Mixed’ and ‘Other’ categories are the fastest growing groups as a result of natural change and international migration.

• • A change of ethnic group category in the census is often due to a person's background fitting more than one category; it need not involve a conscious change of ethnic identity.

• • The main categories of ethnic group can be compared from one census to another, but the residual ‘Other’ categories cannot. A recommended comparison of most stable groups identifies seven groups for comparison from 1991 to 2011, and 12 groups for comparison from 2001 to 2011.

• • When a category names a region or a country, someone born there is more likely than others to stay with that category.

Introduction

Statistics of ethnicity are often treated rather like birthplace, sex or date of birth, as if there is little doubt about either the categories that have been used or their acceptance by respondents to the question. But while the measurement of ethnicity has become a norm in Britain and many other countries, the categories used are not at all fixed. And its reliability is limited: four in every 100 people asked their ethnic group in 2001 changed their answer when asked again in 2011. This chapter explores that reliability, asking why recorded ethnicity may change over time, and who is most likely to change their ethnic group.

Those who collect and review statistics of ethnicity are in no doubt about their fallibility.