Current and Predicted Fertility using Poisson Regression Model: Evidence from 2008 Nigerian Demographic Health Survey

Abstract

Nigeria with persistent high growth rate is among top ten most populous countries. Monitoring key mechanisms of population dynamics particularly fertility in Nigeria is long overdue. Periodical availability of data on fertility and other demographic indices is scarce, hence this study. Our objective was to build a non-linear model to identify fertility determinants and predict fertility using women’s background characteristics. We used 2008 Nigeria Demography and Health Survey dataset consisting of 33,385 women with 31.4% from urban area. Fertility was measured using children ever born (CEB) and fitted into multi -factors additive Poisson regression models. Respondents mean age was 28.64±9.59years, average CEB of 3.13±3.07 but higher among rural women than urban women (3.42±3.16 vs 2.53±2.79). Women aged 20-24years were about twice as likely to have higher CEB as those aged 15-19years (IRR=2.06, 95% CI: 1.95-2.18). Model with minimum deviance was selected and was used to predict CEB by the woman. (Afr J Reprod Health 2014; 18[1]: 71-83).

Keywords: Fertility, Incidence rate ratio, Poisson prediction, children ever born, Nigeria,