Can we, should we, measure ethnicity?

This paper discusses some of the methodological issues in the measurement of ethnicity in the social survey. Measurement encounters two sets of interrelated problems: those of the definition and nature of ethnicity as a social phenomenon and its individual subjective experience and those of the derivation of valid and reliable categories. Commonly used single-item measures fail to capture many ethnic groups and may be of limited value. Extending categories in these will capture more groups, but can be unwieldy and produce small cell counts, whereas multiple measures can be more successful, but are practically less viable in general surveys. We conclude that whilst these difficulties are probably insurmountable, a strategy that is sensitive to the presence of ethnic groups in particular localities is a way forward, but this should also be part of a more sociologically nuanced approach to the collection of ethnic data.