Background : Physical activity (PA) is essential to adolescent health. International research has shown ethnic disparities in PA, and the method used to classify ethnicity may influence how disparities are detected and interpreted particularly in increasingly multiethnic societies. This study examined how different ethnicity classification methods affect adolescent PA outcomes. Methods : Secondary analysis was conducted using data from the 2023 Voice of Rangatahi survey in Aotearoa New Zealand (n = 19,776; ages 12–19 y). PA was measured using a discrete scale (0–7 d/wk) and a binary variable (>4 vs ≤4 d/wk). Ethnicity was self-identified and classified using 4 methods: Prioritized, Total Response, and Single/Combination (8- and 15-group). Analyses applied regression and nonparametric tests. Results : Overall, 24.3% identified as multiethnicity. Asian adolescents consistently reported the lowest PA outcomes (mean = 3.51–3.67 d/wk; 34.4%–36.3% achieving >4 d/wk), while multiethnic Pacific Peoples/European reported the highest (mean = 4.34 d/wk; 52% achieving >4 d/wk). The Single/Combination 15-group method provided the greatest intergroup differentiation (χ 2 (13) = 218.65, P < .001), whereas the Total Response method attenuated contrasts. The Single/Combination 8-group method demonstrated the best model fit (lowest Akaike Information Criterion/Bayesian Information Criterion). Conclusions : Ethnicity classification strongly influences observed disparities in adolescent PA. The Single/Combination 8-group method may be most suitable for research applications due to better model fit and parsimony, while the 15-group method offers finer differentiation to guide policy and the design of targeted interventions. Selecting context-appropriate classification methods is critical for accurately identifying disparities and informing equitable, culturally responsive PA interventions.
Li et al. (Thu,) studied this question.