USING COUNT REGRESSION MODELS TO DETERMINATE FACTORS INFLUENCING FERTILITY OF SUDANESE WOMEN

This paper models the factors influencing fertility in Sudan by using generalized linear models; Poisson regression models and negative binomial regression models. The results show the statistical advantages and aptness of the standard Poisson and negative binomial models for analyzing count data. Both models are used to predict the average number of children ever born (CEB) to women in the Sudan. The findings show a significant relationship between fertility and age at first marriage, gap between births, infant mortality, high level of education and wealth index. The negative binomial regression predicts fertility of Sudanese women better in many respects compared to the Poisson regression models; however, it failed to detect the correct relationship between currently using family planning methods and the number of children ever born.