New Perspectives on Respondent-Driven Sampling with Application to Sex-Trafficked Population in Senegal

Abstract This work is to investigate whether respondent-driven sampling (RDS) yield reliable estimates when applied to extremely hidden and sensitive population under fieldwork constraints that often produce assumption violations and data deficiencies. We estimate the prevalence of sex trafficking among women aged 18–30 engaged in commercial sex in the Kédougou (urban) and Saraya (rural) departments of Senegal from RDS data collected from 561 women. We address substantial methodological challenges including estimator non-convergence and structural network fragmentation. Prevalence was estimated using Salganik–Heckathorn, Volz–Heckathorn, and Homophily Configuration Graph estimators. Further diagnostics are conducted using bootstrap-based uncertainty quantification and rigorous sensitivity analyses, including convergence diagnostics and Leave-One-Seed-Out assessments. Observed prevalence was 18.53% overall, 30.64% in Kédougou and 12.53% in Saraya. Weighted estimates are comparable across the three estimators and perverse intra- and inter-departmental contrasts. The Leave-One-Seed-Out analyses revealed substantial seed-level influence in the urban sample, where the exclusion of specific deep recruitment chains materially shifted prevalence estimates. These findings underscore the importance of explicitly incorporating seeding effects and wave-specific recruitment checks into RDS inference to improve robustness and reliability in studies of complex social phenomena.