Recruiting a representative sample of urban South Australian Aboriginal adults for a survey on alcohol consumption

Background
Population estimates of alcohol consumption vary widely among samples of Aboriginal and Torres Strait Islander (Indigenous) Australians. Some of this difference may relate to non-representative sampling. In some communities, household surveys are not appropriate and phone surveys not feasible. Here we describe activities undertaken to implement a representative sampling strategy in an urban Aboriginal setting. We also assess our likely success.

Methods
We used a quota-based convenience sample, stratified by age, gender and socioeconomic status to recruit Indigenous Australian adults (aged 16+) in an urban location in South Australia. Between July and October 2019, trained research staff (n = 7/10, Aboriginal) recruited community members to complete a tablet computer-based survey on drinking. Recruitment occurred from local services, community events and public spaces. The sampling frame and recruitment approach were documented, including contacts between research staff and services, and then analysed. To assess representativeness of the sample, demographic features were compared to the 2016 Australian Bureau of Statistics Census of Population and Housing.

Results
Thirty-two services assisted with data collection. Many contacts (1217) were made by the research team to recruit organisations to the study (emails: n = 610; phone calls: n = 539; texts n = 33; meetings: n = 34, and one Facebook message). Surveys were completed by 706 individuals – equating to more than one third of the local population (37.9%). Of these, half were women (52.5%), and the average age was 37.8 years. Sample characteristics were comparable with the 2016 Census in relation to gender, age, weekly individual income, Indigenous language spoken at home and educational attainment.

Conclusion
Elements key to recruitment included: 1) stratified sampling with multi-site, service-based recruitment, as well as data collection events in public spaces; 2) local services’ involvement in developing and refining the sampling strategy; and 3) expertise and local relationships of local Aboriginal research assistants, including health professionals from the local Aboriginal health and drug and alcohol services. This strategy was able to reach a range of individuals, including those usually excluded from alcohol surveys (i.e. with no fixed address). Carefully pre-planned stratified convenience sampling organised in collaboration with local Aboriginal health staff was central to the approach taken.