Estimating the Prevalence of Forced Labor Among Domestic Workers in Tunis, Tunisia: A Direct Comparison of Time-Location Sampling and Mark-Recapture Strategies

Author: Zhang, Sheldon; Vincent, Kyle; Cole, Katrina; Rydberg, Jason; Thompson, Shanna Rose; Dank, Meredith & Sanchez, Gabriella

Abstract: Funded under the Prevalence Reduction Innovation Forum (PRIF) initiative to advance knowledge on human trafficking prevalence estimation strategies, we applied two strategies commonly used to study hard-to-reach populations— (1) time-location sampling (TLS) and (2) mark-recapture (MR) in a well-defined geographical area on domestic workers in Tunis, Tunisia. Our TLS design reached a final sample of 1,029 unique individuals, while MR resulted in a final sample size of 1,016, of which 923 were unique individuals. Using the same definitional criteria as all other PRIF teams, we found that both strategies yielded very similar prevalence estimates – about one of every four domestic workers were potential victims of domestic servitude at their present job in Tunis. The TLS strategy found that 22.55% of the sample (24.65% [95% CI 0.21, 0.29] for the target population in general) were potential victims of domestic servitude, while the MR strategy yielded a rate of 25.89% (26.38% [95% CI 0.23, 0.29] for the target population in general). It is our conclusion that both strategies appeared to have worked well, producing similar prevalence estimates. From an operational point of view, the MR approach was much easier than TLS to implement because of the significantly reduced complexity in the field setup.

Keywords: domestic servitude, human trafficking, mark-recapture, prevalence estimation, time-location sampling