Statistical Modeling On Number Of Children Ever Born From Women Of 15-49 Years In Ethiopia

Abstract

High level of fertility and rapid population growth has an impact on the overall socio-economic development of the country in general and maternal and child health in particular; which leads to increased obstetric and medical risks of mothers. According to UN, 2009 reports more developed regions have fertility levels below replacement; whereas, least developing regions have five or above Five children per women. The core objective of this study was to identify key factors that influence high fertility of Women of Child bearing age and to assess between and within regional heterogeneity of determinants of fertility in Ethiopia. Method: Data from the 2011 Ethiopia Demographic and Health Survey which is a nationally representative survey of mothers in the 15-49 years age groups were used to identify determinant factor of fertility for woman of child bearing age (n=4976) in Ethiopia. In this paper, the descriptive statistics of the total children ever born data exhibit the presence of over-dispersion in the data set; we have used Negative Binomial Regression Model and Generalized Poisson Regression Model. These two models have statistical advantages over standard Poisson regression model and are suitable for analysis of count data that exhibit either over-dispersion or under-dispersion and also generalized linear mixed models (GLMM) were used to assess between and within regional heterogeneity determinant of fertility in Ethiopia using 2011 EDHS data set. Results: The results obtained from Generalized Poisson model, Negative Binomial model and GLMM showed that Age of mother, Age at first birth, Age at first marriage, status of education of parents, place of residence, Religion, contraceptive use and status of breast feeding were significantly affect number children ever born in house hold and only Age of mother between 20-39 years, Religion was positive effect for children ever born. It found that Generalized Poisson regression model has statistical advantages over standard Poisson regression model and Negative Binomial Regression model because it was suitable for analysis of count data that exhibit either over-dispersion or under-dispersion. For GLMM it was also found that model with two random intercepts was the best description of the data to address the between and within-regional heterogeneity of fertility. Conclusion: We can conclude that delaying early marriage, extend Age at first birth, Using contraceptive method and Breast feeding were the most determinant factors in reducing number of children ever born in house hold.