Multistate model of life course events: analysis of transition to family formation and first birth with application in Southern Africa

Lawrence Kazembe, University of Namibia
Ndeyapo M Nickanor, University of Namibia

Fertility patterns in Southern African countries including Namibia, South Africa, Lesotho and Swaziland, have been falling to an average of TFR of 3.2 children per woman from 4.5 in the early 2000s. Changing marriage patterns, increased education level among women, and improved socio-economic status have been some of the factors associated with the declining fertility patterns. However, several pathways exist that may explain lower fertility rates, for instance sexual debut, delayed entry into marriage and or delayed first birth. Examining such pathways and associated factors may help understand the falling fertility patterns. In this study, we use multistate models (MSM) to explore time to the event - first birth, through multiple stages. Multistate models (MSM) are useful to analyze life course events, in a situation where transitions to intermediate states are equally important, apart from the ultimate event (first birth). We extend MSM to examine factors associated with transition from birth to the woman’s first birth, through intermediate stages such as sexual debut and marriage. MSM are fitted, through multiple survival models, to analyze four stages: birth[1]-first sex[2]-marriage[3]-birth[4]; or birth[1]-first sex[2]-birth[4]-marriage[3]; or birth[1]-marriage[3]-birth[4]. We apply the MSM models to study fertility patterns in Namibia, Malawi, and Swaziland using the recent Demographic and Health Surveys (2005-2013). In our model we adjusted for education level, place of residence, modern contraceptive use and other socio-demographic variables. Models were implemented using the Bayesian Inference. The sensitivity of the model to prior assumptions is explored.