Performing the analysis: a series of designed steps ensuring you are entering the correct information into the model. A useful convention to code dichotomous variables: assign “1” to presence and “0” to absence of condition; the variable’s mean will be equal to the condition’s prevalence. The reference group choice will not affect results but will affect how results are reported. Choose your reference category based on the main hypothesis. Interaction terms are entered by creating a product term: a variable whose value is the product of two independent variables. For proportional hazards or other survival time, enter: starting time, outcome of interest’s date, censor date.

Variable selection techniques: automatic procedures determining which independent variables will be included in a model. It is usually better for the investigator to decide what variables should be in the model rather than using a statistical algorithm.

Working with a biostatistician should be an iterative process. Especially with complicated studies, consult them at each analysis phase. For conducting the analysis, use the stat package your research group uses so you will always be able to get help when needed.