The data were generated such as to resemble the key results of Kitsantas et al. (2007). However, the variables age and alcohol_per_month, that are used only in a discretized form by Kitsantas et al. (2007), were generated as numeric variables to illustrate the selection of optimal cutpoints in recursive partitioning. The generated data set, as well as the R-code used for all examples, are available as supplements.