Hierarchical models, or multilevel models, are used to represent data that might vary on, you guessed, different levels. Michael Freeman, from the University of Washington Information School, provides an introduction to the method using a scrolling format. The transitions give a good sense of how the model can change, depending on your approach.