I am puzzled with logistic regression! When to do binary, when to do multiple and when to do stepwise logistic regression? What are the criteria of data? How to understand the output and how to write those in main results. Please can anybody help!
When the outcome is binary then you would do binary (simple or multiple) logistic regression. Simple means you have only one predictor. Multiple means you have more predictors. Stepwise is an (unrecommended) method for multiple logistic regression to select some predictors among all. Understanding the output and interpreting the results is a complicated task. I advise you to search some web sites or RG.
Rajat, Mehmet has given an excellent answer. In addition I would look at this set of notes if you are familiar with multiple regression by ordinary least squares. Best Wishes, David PS A logit model is the model in logistic regression.
Multiple logistic regression is sensitive to the presence of multicollinearity and this makes using stepwise regression less recommended unless you carefully study multicollinearity between the regressor variables.