The value of non-probability surveys: collecting data on migrants through service provider NGOs
Conducting methodological defensible, logistically feasible and affordable large-scale national surveys of migrants is a serious challenge. Based on the experience of running a multi-year Migrant Rights Monitoring Project (MRMP) in South Africa, this paper outlines the pros and cons of working with and through NGOs which provide services to migrants (including asylum seekers, refugees and undocumented migrants) in order to conduct a national longitudinal survey on migrant access to basic public services. From a narrowly defined, formal methodological perspective, this access method clearly does not result in a sample which is representative of a total national population of migrants. Some of the resulting limitations are outlined in the paper, including urban bias, nationality bias, documentation bias, gender bias, and vulnerability bias, as well as effects relating to respondent duplication, strategic responses and intermediary quality and interests. The paper argues that these biases are problematic and we need to be aware of them, but that they are not significantly greater than for migration surveys which attempt to use more representative sampling methods. Furthermore, there are strong benefits of such an NGO-linked methodology. Recognising these benefits requires a wider reference point than formal methodology, namely the overall purpose of the research. Apart from making larger and more longitudinal surveys logistically and financially possible in the first place, such benefits include the formation of active and collaborative networks among organisations in the migrant rights sector; capacity building within this sector around research and the use and meaning of empirical data; and the direct integration of empirical data into local and national advocacy work.