Geoadditive Latent Variable Modeling of Count Data on Multiple Sexual Partnering in Nigeria




Summary The 2005 National HIV/AIDS and Reproductive Health Survey (NARHS) in Nigeria provides evidence that multiple sexual partnering increases the risk of contracting HIV and other sexually transmitted diseases. Therefore, partner reduction is one of the prevention strategies to accomplish the Millenium Development Goal of halting and reversing the spread of HIV/AIDS. We consider the numbers of girlfriends, casual, and commercial partners of heterosexual men, reported in the NARHS study, as observed indicators of their latent attitude toward multiple partnering. To explore the influence of risk factors on this latent variable, we extend semiparametric methodology for latent variable models with continuous and categorical indicators to include count indicators. This allows us to simultaneously analyze linear and nonlinear effects of covariates, such as sociodemographic factors and knowledge about HIV/AIDS, on attitude toward multiple sexual partnering, which in turn influences the observable count indicators. The results provide insights for policy makers who are aiming to reduce the spread of HIV and AIDS among the Nigerian populace through partner reduction.