As discussed in Principles of multiple regression section, multiple linear regression, multiple logistic regression and Poisson regression are all related modeling techniques. In each of these three cases, you may have one or more independent (X) variables (also called predictor variables), and you’ll have a single dependent (Y) variable (also called a response variable or an outcome variable). Despite the similarity of the three methods, choosing which regression method to use is easy; it's based on the type of dependent (Y) variable you are modeling.

•When Y is continuous use multiple linear regression

•When Y is count data (0,1,2, ...) use Poisson regression. In Prism, this is an option within  the multiple linear regression analysis.

•When Y is binary (yes/no, presence/absence, etc.) use multiple logistic regression

There's a natural hierarchy of complexity with these modeling types. If you're new to regression modeling, we recommend learning simple linear regression first. Simple linear regression has some intuitive extensions: nonlinear regression and multiple linear regression are two of them. Multiple linear regression, in turn, has two extensions available in Prism: multiple logistic regression and Poisson regression. Finally, as described above, multiple logistic regression involves fitting a model to a set of multiple independent variables. However, if you only have one independent variable, this method can be simplified into simple logistic regression (similar to the relationship of simple linear regression and multiple linear regression). In this way, simple logistic regression can also be thought of as an extension of simple linear regression.

If this all sounds a bit overwhelming, that’s ok. You can read more about each of the multiple regression methods in the Principles of Regression portion of this guide, or you can continue with any of the links below to learn more about how each method is performed within Prism.