Modeling the Effect of Mediation on HIV Prevalence in Kenya using a Logistic Regression Model
Ruth Naomi Wanga *
Department of Mathematics Masinde Muliro University of Science and Technology, P.O Box 190-50100, Kakamega, Kenya.
David Alila
Department of Mathematics Masinde Muliro University of Science and Technology, P.O Box 190-50100, Kakamega, Kenya.
Everlyne Akoth Odero
Department of Mathematics Masinde Muliro University of Science and Technology, P.O Box 190-50100, Kakamega, Kenya.
Robert Keli
Department of Mathematics Masinde Muliro University of Science and Technology, P.O Box 190-50100, Kakamega, Kenya.
*Author to whom correspondence should be addressed.
Abstract
The control of HIV/AIDS demands different interventions based on various HIV risk factors directly or indirectly affecting HIV prevalence through a mediator variable. There is however limited literature on how these risk factors interact with each other and in turn affect HIV/AIDS prevalence in presence of mediator factors [1]. A logistic regression model formulated in presence of mediation was found to fit both simulted and real data from 2018 Kenya Population-based HIV Impact Assessment (KENPHIA) survey well and had a higher predictive power as compared to the model formulated in absence of mediation. This was accomplished by using Binary logistic regression to fit the models and estimating the model parameters using Maximum Likelihood Estimation in R. Akaike’s Information Criterion was used to determine amount of data lost by respective models and McFadden’s \(R_2\) to evaluate the adequacy of the model fit.
Keywords: Mediation, logistic regression