ABSTRACT
The general estimating equations (GEE) approach, initially developed by Zeger and Liang (1986), is used in lieu of generalized linear models (GLMs) when there is statistical dependence among observations. So, for example, use GEE if using ordinal, logistic, or Poisson regression for pre–post or repeated measure analysis. GEE can also be used for clustered data structures as an alternative to multilevel models, especially when there are many small groups (e.g., many small discussion groups), because approaches such as multilevel models are better used with larger groups of at least 20–30 students each so that individual regression equations can be estimated for each group (Hox, 2010). However, of the two situations (repeated measures and clustered observations), the repeated measures application is stressed in this chapter (and the IBM SPSS procedure is designed only for repeated measures).
