The Stata Journal (2012) Vol. 12 No. 1, pp. 162-164.
When including a set of indicator variables (often called dummy variables) the standard approach is to leave one category out, the so-called reference category. It is possible to include an indicator variable for the reference category if one excludes the constant, and this can provide a useful alternative view on the same model. This idea can also be extended to interaction effects. In this tip I discuss a couple of examples of how to use this trick.