.

This study investigated the geographical variations in the sexual behaviours of Nigerian women aged 15-49 years, using geoadditive multinomial logit model. The dependent variable was sexual abstinence which was viewed within the context of “Recent sexual activity”. This was categorized for analysis as “Active in the last 4 weeks, “Had postpartum abstinence” and “Not active in the last 4 weeks” and “Had postpartum abstinence” was made the reference category. Effects of all categorical covariates were assumed to be linear and hence estimated in the usual parametric form while the effect of age was assumed to be nonlinear. The geographical information at state level was also explored in the model to assess the impact of spatial distribution on sexual behaviours of the women. Inferences were based on Markov Chain Monte Carlo (MCMC) technique. Findings revealed some spatial variations in the sexual patterns among Nigerian women at less aggregated state level. The results of this study can provide insights to states that possibly need interventions on matters relating to the sexual risks among the Nigerian women, and this can reduce the spread of HIV and AIDS among the Nigerian women through controlled sexual activities.
Keywords: Multinomial Logit Model, Abstinence, Bayesian, Geoadditive, MCMC