Human-Centered Child Trafficking Prevalence: A Latent Class Analysis of Exploitation Sectors

Abstract Child trafficking and the worst forms of child labor are pervasive yet difficult to measure. In three hotspot districts in Sierra Leone’s Eastern Province (Kailahun, Kenema, and Kono), a baseline mixed-methods study estimated that approximately one-third of children aged 5–17 had experienced child trafficking, underscoring the need to understand how exploitation varies across children’s experiences. This study uses latent class analysis to identify profiles of trafficking-related experiences among children who met the study’s operational criteria for child trafficking, supplementing a binary “trafficked/not trafficked” classification with a profile-based characterization of heterogeneity in exploitation experiences. We analyzed caregiver-reported data from a baseline household survey of 2,025 children aged 5–17 in Kailahun, Kenema, and Kono who met an operational definition of child trafficking experience. Models were estimated using all available manifest-indicator data across 19 indicators spanning six trafficking-related labor sectors, five hazardous tasks performed away from home, and eight force, fraud, and coercion experiences. A five-class model was retained based on fit, stability, parsimony, and interpretability. The classes were: Class 1, moderate coercion with minimal hazards (25.8%); Class 2, high multi-domain exploitation (7.8%); Class 3, high coercion with moderate hazardous work (10.6%); Class 4, environmentally hazardous work with relatively low coercion (30.8%); and Class 5, heavy-load labor with low coercion (25.1%). Class 2 showed the highest burden across labor sectors, hazardous tasks, and force, fraud, and coercion indicators, whereas Class 3 was marked primarily by high coercion. External validation showed that Class 2 had the highest orphanhood and lowest school enrolment. Class distributions differed across districts, with Class 4 most prevalent in Kailahun and Kono and Class 5 most prevalent in Kenema. Child trafficking experiences in Eastern Sierra Leone are heterogeneous and not well captured by a single binary classification. Latent class analysis-derived profiles can improve understanding of exploitation patterns and help inform more targeted, context-specific prevention and protection responses.