ABSTRACT
Although the correlated trait-correlated method (CTCM) model is limited in different ways and its application usually shows serious problems, it is still widely used in analyzing multitrait-multimethod (MTMM) data. Taking a design-oriented view it is discussed why the CTCM model is not a reasonable model for analyzing MTMM data. Focusing on three typical types of research designs (designs with only interchangeable methods, designs with only structurally different methods, designs with a combination of interchangeable and structurally different methods) it is shown that no general CTCM model can be defined in a reasonable way. However, several variants of this model seem to be reasonable if certain conditions are fulfilled (e.g., uncorrelated trait and method factors). The trait and method factors in these restricted models have a clear meaning, but the meaning differs between the restricted models depending on the research design. We present a new restricted CTCM model for the combination of interchangeable and structurally different methods that contains as many method factors as methods in the design (as in classical CTCM models). However, the model allows only some, but not all, method factors to be correlated.
