Author(s):
Pablo Mateos, Dr* - University College London
Paul Longley, Prof - University College London
Abstract:
Understanding of the nature and detailed composition of ethnic groups in today's increasingly multicultural societies remains key to a vast swathe of social science. In practice, however, much research remains hamstrung by the quality and availability of ethnicity classifications, and consequent shortcomings in our ability meaningfully to subdivide populations.
This paper proposes an alternative approach to researching the ontology of ethnicity based on the origin of surnames and forenames. This methodology assigns an individual's forename and surname to one of 185 Cultural Ethnic and Linguistic groups (CEL), which are weighted according to name scores, in order to assign the most probable CEL allocated to each individual at very fine geographical levels. This method has been applied to the UK Electoral Roll and European telephone directories as well as to several health registers in London. The accuracy of the method has been evaluated using separate datasets where the self-reported ethnicity of individuals has been recorded. It is argued that this methodology better reflects the complex dimensions of ethnicity and can be more flexibly adapted to the particularity of each study.
The outcome of the research is an improved methodology for classifying population registers, as well as small areas (typically postcodes), into cultural, ethnic and linguistic groups, that makes possible the creation of much more detailed, frequently updated, representations of the multicultural geographies of contemporary cities.
