We can also compare means between more than two groups. For instance, in the main
textbook example we may wish to compare Sales averages between small, medium and big
customers. Here, there are three categories to compare.
In this case, you cannot use t-tests (there are more than two groups). Some people
try to analyze all possible pairs with t-tests but this is wrong because you can’t
assess the overall picture if you’re always excluding one or more of the groups. This
would lead to greater error in your testing for reasons I won’t go into here.
The correct statistical approach will be a variant on the Analysis of Variance (ANOVA)
family of procedures. There are various versions of ANOVA procedures, which suit different
situations.
This introductory book does not cover ANOVA; the reader should refer to intermediate
statistics texts.