Several approaches have been proposed to classify populations into ethnic groups using people's names, as an alternative to ethnicity self-identification information when this is not available. These methodologies have been developed, primarily in the public health and population genetics literature in different countries, in isolation from and with little participation from demographers or social scientists. The objective of this paper is to bring together these isolated efforts and provide a coherent comparison, a common methodology and terminology in order to foster new research and applications in this promising and multidisciplinary field. A systematic review has been conducted of the most representative studies that develop new name-based ethnicity classifications, extracting methodological commonalities, achievements and shortcomings; 13 studies met the inclusion criteria and all followed a very similar methodology to create a name reference list with which to classify populations into a few most common ethnic groups. The different classifications' sensitivity varies between 0.67 and 0.95, their specificity between 0.80 and 1, their positive predicted value between 0.70 and 0.96, and their negative predicted value between 0.96 and 1. Name-based ethnicity classification systems have a great potential to overcome data scarcity issues in a wide variety of key topics in population studies, as is proved by the 13 papers analysed. Their current limitations are mainly due to a restricted number of names and a partial spatio-temporal coverage of the reference population data-sets used to produce name reference lists. Improved classifications with extensive population coverage and higher classification accuracy levels will be achieved by using population registers with wider spatio-temporal coverage. Furthermore, there is a requirement for such new classifications to include all of the potential ethnic groups present in a society, and not just one or a few of them. Copyright (c) 2007 John Wiley & Sons, Ltd.
