Socio-economic and Demographic Covariates of Infant Mortality in Sub-Saharan Africa: Application of Analysis of Variance (ANOVA) and Multiple Regression Models

Abstract

The main thrust of the study is investigate the main socio-economic and demographic antecedents and proximate factors influencing Infant Mortality Levels and Patterns in Sub-Saharan African
 The source of data of the study is largely from the 2001-2005 UNDP Human Development Reports, National Census Results, DHS data and Demographic survey results of Sub-Saharan African Countries.
 The analytical techniques include analysis of univariate and bivaraite frequency distributions, ANOVA, correlation matrix and Multivariate regression models. Each country is taken as the unit of observation and a total of 42 countries are covered. The dependent variable is the level of Infant Mortality whereas 14 independent variables as the predictors influencing the Infant Mortality conditions are considered. 
 The model outputs suggested that all the fourteen independent covariates appeared to influence the levels of infant mortality positively or negatively, but in varying values of statistical significance, due to multicollinearity among the independent variables as manifested in the correlation matrix which appear to hinder clear understanding of the impact of each independent variable on infant mortality. However, the stepwise regression output identified the critical covariates of the infant mortality levels.
 In conclusion, the study indicated that persistent high infant mortality level in the African -Continent has yet to continue for some decades before coming down to acceptable levels. The main reasons which largely appeared to maintain high Infant Mortality Levels among the Sub-Saharan Countries were noted to be the widespread poverty conditions and low level of health services and education in rural Sub-Saharan Africa. Consequently, the high infant mortality in rural Africa has been observed to be influenced by high fertility level, family planning programmes, education and employment, among other background and proximate variables.