Abstract:
Ethnicity is an important variable in health and social science research, but classifying
ethnicity is complex because of the construct’s fluidity and multiplicity. There are several
possible ethnic classification methods, each with strengths and limitations. Existing literature
suggests that researchers’ decisions on ethnic classification method can influence research
findings. However, there is little empirical research on how these classification methods affect
quantitative research in samples with higher multi-ethnic prevalence (e.g., >10%). Situated
within a critical quantitative paradigm, the current thesis addresses this gap by utilising largescale
datasets from increasingly multi-ethnic cohorts (adults, adolescents, and children) in
Aotearoa New Zealand to examine the effects of ethnic classification method on reported ethnic
group size, demographic composition, and substantive outcomes.
The corresponding three studies found that, first, ethnic classification method can
substantially impact outputted ethnic group sizes, especially when there are higher rates of
multi-ethnic identification (e.g., among children and adolescents). Second, there was a high
rate of discrepancy between the two popular methods of administrative-prioritisation and selfprioritisation
in each age cohort (≥60%). These discrepancies were systematically associated
with contextual characteristics such as neighbourhood ethnic composition and socioeconomic
deprivation. Third, using adolescent mental health outcomes as a case study, it was found that
ethnic classification method can affect substantive outcomes, both within nominal ethnic
groups (by an effect size of up to d = 0.12), and between nominal ethnic groups (by an effect
size of up to d = 0.25).
Together, the results indicate that researchers’ choice of ethnic classification method
affects who is included or excluded from ethnic groups, the demographic composition of these
ethnic groups, and conclusions about the extent of ethnic disparities. This, in turn, influences
knowledge construction, practice, and policy. Given that each ethnic classification method has
strengths and limitations, the studies in this thesis highlight the importance for researchers to
critically and transparently select the most appropriate ethnic classification method for their
research question and context. The power of the researcher, combined with the subjectivity of
ethnicity measurement, also emphasises the need for researchers to be critical and reflexive
throughout the wider research process.