Using an inclusive weighting approach to enhance the utility of Indigenous-governed health and wellbeing data from non-probability surveys — The Association Specialists
  1. Schedule
  2. Oral Presentations - Indigenous and community-led and governed data collection, measurement and reporting
  3. Using an inclusive weighting approach to enhance the utility of Indigenous-governed health and wellbeing data from non-probability surveys

Using an inclusive weighting approach to enhance the utility of Indigenous-governed health and wellbeing data from non-probability surveys (24213)

Christopher D McKay 1 , Benjamin Harrap 1 , Katherine Thurber 1 , Bernard Baffour 2 , Joanne Thandrayen 1 , Bronwyn Wilkes 1 , Bobby Maher 1 , Siena Montgomery 1 , Ray Lovett 1
  1. Yardhura Walani, Australian National University, Canberra, ACT, Australia
  2. School of Demography, Australian National University, Canberra, ACT, Australia

Background

Mayi Kuwayu: The National Study of Aboriginal and Torres Strait Islander Wellbeing is a large-scale cohort study that is Indigenous-led and -governed, embeds Indigenous Data Sovereignty principles, and includes community developed cultural wellbeing indicators. Wave 1 participants were recruited using multiple sampling strategies during 2018-2021 with the intention of capturing diversity within the population, while enabling individual and community-level self-determination in participation. We aimed to develop survey weights to calculate population prevalence estimates and enhance the utility of Mayi Kuwayu data for Aboriginal and Torres Strait Islander people.

Methods

We used a post-stratification method to generate survey weights with population benchmarks sourced from the Australian Bureau of Statistics (ABS) 2019 estimated resident population (2016 Census). Benchmark variables included age, sex, remoteness area, state/territory, and area-level socioeconomic position. The range, variance, and proportion of extreme weights were assessed to guide final benchmark selection. We compared a standard weighting approach that excludes participants with missing benchmark data (e.g., non-binary or missing gender) to a more inclusive approach. Weighting performance was assessed by comparing weighted prevalence of Mayi Kuwayu health variables against comparably measured health variables in the National Aboriginal and Torres Strait Islander Health Survey (NATSIHS) 2018-19.

Results

Weighted prevalence of benchmark variables had closest alignment with ABS population distributions when all benchmarks were used. Some health variables had a weighted prevalence consistent with NATSIHS estimates (e.g., depression, anxiety and high blood pressure), but others varied significantly (e.g., asthma and self-assessed health). The prevalence estimates calculated using the inclusive approach had negligible differences to those estimated using the standard approach and had slightly more precision.

Conclusion

The weighting approach used by Mayi Kuwayu enhances the utility of Indigenous-governed data for health evidence generation, and demonstrates an ethical and inclusive approach to survey weight generation that could be adopted by other studies.