F14 Record linkage with partially missing values in multiple systems estimation

Abstract

""Population size estimates are important for understanding social phenomena and for informing political decisions. However, privacy restrictions that prevent the disclosure of identities make accurate population estimates difficult. This study addresses these challenges related to victims of trafficking in the Netherlands. I focus on transforming covariate marginal frequencies into joint frequencies in order to understand the relationships within the dataset. Current approaches to transform marginal frequencies into joint are limited in capturing the underlying relationships between covariates. To solve this problem, this work proposes a methodology that uses the Kronecker product to efficiently generate potential combinations of covariate levels. The dataset used consists of marginal frequencies from six population registries associated with Dutch anti-trafficking organizations. This provides insight into the distribution of covariates and highlights differences in gender, age, nationality and type of exploitation. Analysis of joint frequencies reveals patterns and insights, highlighting the vulnerability of specific populations and the prevalence of various forms of exploitation. This result shows the importance of analysis of joint distributions to obtain meaningful results. The proposed methodology bridges the gap between marginal and joint distributions, contributes to the estimation of multiple systems, and improves our understanding of human trafficking in the Netherlands. ""