Estimating the Prevalence of Domestic Servitude Among Tanzanian Women Who Have Returned from Working Overseas

Author: Dank, Meredith; Zhang, Sheldon; Vincent, Kyle; Stoeltje, Mackenzie & Wanz, Mengfei

Abstract: The authors conducted a regional prevalence study on domestic servitude among Tanzanian women who worked abroad as part of the Prevalence Reduction Innovation Forum. Two prevalence estimation methods were employed within the study region: (1) Multi-Stage Probability Proportional to Size, revised to Stratified Simple Random Sampling (STSRS), with a sample of N = 1,052, and (2) Vincent Link-Tracing Sampling (VLTS) with a sample of N = 788. Both methodologies found that over half of Tanzanian women were victims of domestic servitude while working abroad; the STSRS sample had a rate of 69.1% of domestic servitude and the VLTS sample had a rate of 59.9%. Both sampling methods were relatively effective in producing population estimates. However, these two estimation strategies provided somewhat different estimates for population size, and even within VLTS, the population size estimates were fairly discrepant. From the perspective of applied social sciences, both estimation strategies appeared to work relatively well. Operationally, STSRS was easier to implement because of the significantly reduced complexity in tracking and tracing links required for VLTS. Additionally, the VLTS design is the most ideal method for estimating the size of a hidden domain of a highly mobile and hidden population, such as the one in this study.

Keywords: domestic servitude, human trafficking, prevalence estimation, stratified simple random sampling, link-tracing sampling