Categorical Data Analysis Using The SAS® System, 2nd Edition

12.3. Simple Poisson Counts Example

The following data come from a cross-sectional study of 400 patients who had malignant melanoma (Roberts et al. 1981). The site of the tumor and the histological type were recorded. These data are also analyzed in Chapter 16, “Loglinear Models,” from the perspective of the loglinear model framework, in which you consider all dimensions for classification to be response variables (not response and explanatory variables) and you determine if there is association within that set of variables.

It is reasonable to consider the tumor counts to be distributed as Poisson and to determine whether those counts are influenced by site and type of tumor. By taking the log of the counts, you can fit a loglinear model.