Creating a frame: random sampling in a non-homogeneously distributed urban migrant population in South Africa

Gayatri Singh, Brown University
Kennedy N Otwombe, University of the Witwatersrand

This paper details a research method for carrying out survey work in difficult-to-reach populations using a stratified sampling strategy as utilized in a multinational survey in Johannesburg (South Africa). The method was developed in response to a lack of baseline household distribution, unique (and often unsafe) terrain of inner city Johannesburg, need to collect methodologically sound data on social factors affecting HIV risk for urban migrants and past claims from researchers that random sampling in this context was impossible. Consequently, the success of the approach depended upon the several factors: utilization of geo-data to generate a sampling frame; mixed methods approach of the overall project; integration of the sampling strategy in questionnaire design and effective collaboration with community members. An enumeration of the method and challenges faced in the data collection is provided in this paper to demonstrate the feasibility of a random sampling approach within non-homogeneously distributed population groups.