Weibull Accelerated Failure Time (AFT) Model for Time – To – Divorce Data with Cure Fraction

Abstract

We present the use of survival analysis to investigate divorce rate and marriage stability among ever married women in Kano metropolis. Specifically, Weibull Accelerated Failure Time Mixture Cure Model (Weibull AFT MCM) was used to analyze time-to-divorce data. We used the cure model because the Kaplan-Meier divorce survival curve has indicated the possibility of marriages that will forever remain without divorce (cure fraction). This is against the usual regression models used, because they do not incorporate the possibility of cure fraction. Also, comparing the Weibull homogeneous model with the Weibull AFT MCM indicates the superiority of the use of the cure model in terms of log-likelihood (-620.8 versus -323.6) and AIC (1245.6 versus 641.6). Therefore, the aim was to estimate the proportion of marriages that will forever survive divorce, divorce survival curve, as well as to ascertain the influence of some factors on the time-to-divorce as well as on the probability of remaining in marriage without divorce. The estimates of divorce fractions from Kaplan-Meier and Weibull AFT MCM are close to each other (73.4 versus 72.2). Factors that significantly cause divorce include Husband employment status, Domestic violence and Discussing family matters with spouse. While, level of education and Discussing family matters with the spouse significantly influences time-to-divorce. It is seen that, discussing family matters with the spouse is a factor that significantly influence both the incidence and latency sub models. Findings may benefit the government and other stakeholders in marriage stability planning policies.