ABSTRACT

Social and behavioral research often concerns problems that have a hierarchical structure, such as, for example, when individuals are nested within groups. In multilevel analysis, data structures of this nature are viewed as a multistage sample from a hierarchical population. For example, in educational research there may be a sample of schools and within each school, a sample of pupils. This structure results in a data set consisting of pupil data (e.g., socioeconomic status [SES], intelligence, school career) and school data (e.g., school size, denomination, but also aggregated pupil variables such as mean SES). In this chapter, the generic term multilevel is used to refer to analysis models for hierarchically structured data, with variables defined at all levels of the hierarchy. Typically, such research problems include hypotheses of relationships between variables defined at different levels of the hierarchy.