Nehehe,
The technique illustrated in my article is used to create data warehouse dimensions that represent code sets. This scenario isn't a one-size-fits-all solution, but more often than not, it's quite helpful.
In a transactional model, users can sometimes select multiple values for a given data point. My example uses an "item" code set. Since customers can purchase multiple "items," a record would be saved for each item in the transactional model. In the analytical model, the data must often be pivoted into a single row that represents the "items" customers purchased. This is a common practice in data warehouses that employ star schemas where you would only have a single row in each dimension that represents a given fact (measure).
I hope this sheds a little more light. 🙂
-Craig