Connected Threads: Africa, HIV and Intimate Partner Violence Using the INLA-SPDE Approach

This work aims at evaluating the relationship between intimate partner violence (IPV) and HIV using survey data from Zambia and Zimbabwe— specifically, data from the Demographic and Health Survey. Sub-Saharan Africa’s population is strongly affected both by HIV and IPV. Indeed, the literature has paid great attention to the association between these two phenomena and how their contextual characteristics affect their association and their spatial distribution. The sample used in this work includes women living in Zambia and Zimbabwe who were tested for HIV during the survey and who were asked about their experiences with abuse. Two different perspectives of analysis are here considered: first, women as individual respondents, then as people being part of a neighbourhood. The estimation results of a logistic regression model and the post estimation results of a Bayesian hierarchical Beta regression model with the integrated nested Laplace approximation (INLA) are presented. HIV is the response variable. Results show that women who are both physically and emotionally abused by their partners are more likely to be affected by HIV, both when considered individually and within their neighbourhood; maps allow to visualise HIV prevalence within regions.