Reliable estimation of child trafficking prevalence at fine geographic scales is critical for targeted program and policy interventions, yet such estimation is often hindered by sparse or incomplete data. This study investigated the use of small area estimation (SAE) methods—one frequentist and the other hierarchical Bayesian—to improve prevalence estimates using data from a household survey of 3070 households across three hotspot districts in Sierra Leone. SAE models were implemented at the chiefdom level (m = 40), the lowest administrative unit, and compared against traditional district-level estimates. Both SAE approaches yielded chiefdom-specific prevalence estimates and predictive probabilities that exhibit substantial heterogeneity and greater precision than the district-level estimates. By borrowing strength across small areas, SAE methods achieved notable reductions in variance and enhanced inference for rare and sensitive outcomes. These findings underscore the utility of SAE for producing reliable local-level estimates in contexts where direct estimation for small geographic areas is infeasible or insufficient. The study highlights the methodological and practical implications of SAE to inform localized interventions and encourages its broader application in research on hidden or hard-to-reach populations, including those affected by crime victimization and human rights violations.
Yi et al. (Wed,) studied this question.