each graded by a different judge. When all students complete all assignments, and each judge evaluates one of the assignments, we have a two-facet crossed design. If we want to assign a single grade to the students, this constitutes a design with one trait and multiple methods (i.e. the cross-classification of assignments and judges). In this design, the person scores are decomposed into seven parts: the grand mean, the effects from students, assignments, and judges, and all the two-way interaction terms. The three-way interaction and the error components cannot be distinguished. When all students complete different assignments evaluated by different judges, we have a two-facet nested design. In practice, many designs are partly nested. For instance, all students complete the same assignments evaluated by different judges, or each student completes his/her assignments, which are then evaluated by all judges, but the subset of assignments completed is different for each combination of students and judges. For an elaborated description, we refer to Shavelson and Webb (1991).
Was this article helpful?