Classifying multiple ethnic identifications: Methodological effects on child, adolescent, and adult ethnic distributions

Background: The burgeoning global multi-ethnic population, in conjunction with the importance of accurate ethnic group counts for research and policy purposes, make classification of multiple ethnic responses a complex but important issue. There are numerous possible classification approaches, differing in ethical implications and ease of statistical application. Objective: This study empirically examines the validity and consistency of three comparatively accessible ethnic classification methods (total response, administrative-prioritisation, and self-prioritisation) in increasingly ethnically diverse age cohorts (adults, adolescents, and children). Methods: We utilised secondary data from two large-scale studies in Aotearoa/New Zealand which asked children (N = 6,149; responded via mother proxy), adolescents (N = 8,464), and adults (N = 11,210) to select (1) all the ethnicities they identified with, and (2) their main ethnicity. The data were coded, then analysed using descriptive statistics and z-tests for proportional differences. Results: The majority of multi-ethnic participants were able to select a main ethnic group when required, but around 20% could not or refused to do so, and there was over 60% discrepancy between self-prioritised ethnicity and administrative-prioritised ethnicity. Differences by age group and ethnic combination were apparent. Comparison of overall ethnic group proportions outputted by the three classification methods revealed within-group variation, particularly where there were higher rates of multi-ethnic identification. Contribution: This study empirically demonstrates that researchers’ choice of ethnic classification method can have a strong influence on ethnic group proportions. Researchers should therefore select the classification method most appropriate for their research question and clearly report the method employed.