ABSTRACT
Not all multilevel data are purely hierarchical. For example, pupils can be nested within the schools they attend, and within the neighborhoods in which they live. However, the nesting structure may be less clear when we consider the schools and neighborhoods. It is likely that there is a tendency for pupils to attend schools in their own neighborhood, but there will be exceptions. Some pupils will attend schools in other neighborhoods than the one they live in, and, in particular, schools that are located close to the border of two neighborhoods may be expected to draw pupils from several neighborhoods. As a result, it is not possible to set up an unambiguous hierarchy of pupils within schools within neighborhoods. We could, of course, arbitrarily set up such a model structure, but to do so we would have to include several schools more than once, because they appear in several different neighborhoods.
