ABSTRACT
In this study the problem of relating a qualitative dependent variable to one or more independent variable which may or may not be qualitative. This problem has its multivariate analogous as well, when the dependent and independent variables are related by a logistic function, the model is often referred to as a logistic regression. A step-up [forward] selection procedure from the set of the independent variables, involving all the 63 variables in the data set were carried out. The forward selection procedure begins by fitting the intercept. It then solves each of the models with only one independent variable. A linear logistic regression [LLR] models were fitted for the response variable Y, is 0 for a child having diarrhea disease and 1 for a child having no diarrhea using the selected subset of variables obtained from the set-up procedure for Daqahlia and Sohag Governorates together based on 1020 observations, for Daqahlia Governorate based on 660 observations, for Sohag Governorate based on 330 observations