Business Statistics Made Easy in SAS

Data Assumptions and Alternatives when Comparing Categories

Introduction to Data Assumptions

When considering comparison of categories, some (but not all) traditional parametric statistics rely on certain assumptions, including but not limited to:

1. Normality: Many statistical tests assume normality of various things, including:

   1. The statistical estimates under review.
   2. In some cases only the actual data itself should be normal. For instance, to use traditional tests to compare the productivity of two factories, the productivity data of each factory needs to be normally distributed. This is a troublesome assumption, as many datasets are not normal in distribution.
   3. Residuals: In some of the techniques, the data doesn’t need ...