The easy interpretability of the visual representation of classification trees, that we have illustrated in this example, has added much to the popularity of this method, e.g., in medical applications. However, the downside of this apparently straightforward interpretability is that the visual representation may be misguiding, because the actual statistical interpretation of a tree model is not trivial. Especially the notions of main effects and interactions are often used rather incautiously in the literature, as seems to be the case in Berk (2006, p. 272), where it is stated that a branch that is not split any further indicated a main effect. However, when in the other branch created by the same variable splitting continues, as is the case in the example of Berk (2006), this statement is not correct.
