Abstract
Declining survey response rates threaten the inferential basis of quantitative social research, and evidence on effective recruitment outside Western contexts is thin. We randomised 2,400 households across three recruitment strategies: unannounced visit, advance letter and community intermediary. Intermediary recruitment achieved a 61 per cent response rate against 29 per cent for unannounced visits, but introduced measurable clustering on political attitudes. Researchers face a genuine trade-off between response rate and sample independence that is rarely acknowledged in fieldwork reports.