Multilevel analysis for repeated measures - hierarchical linear model as an alternative to the analysis of variance

The repeated measures design is a popular research method, in some cases more suitable then the independent groups design. The most common method of data analysis for the repeated measures is the mixed effects univariate analysis of variance (ANOVA). The strict assumptions of this statistical method, however, cannot always be met, therefore an alternative is needed. This article presents the analysis of the repeated measures design in the hierarchical linear model, which is characterized by more liberal assumptions and better modelling ability. In two examples the author describes how in a simple research design the results of ANOVA and the hierarchical linear model are almost identical, while in more complex designs the hierarchical linear model allows for more accurate analyses than ANOVA. peerReviewed publishedVersion