Provides alternatives to the definitions of additive and multiplicative method effects in multitrait-multimethod data given by D. Campbell and E. O'Connell (1967). The alternative definitions can be formulated by means of constraints in the parameters of the correlated uniqueness model (H. Marsh, 1989). (SLD)

Demonstrates the advantages of Rao's Distance (C. Rao, 1949) over the root mean residual square for model comparisons using an analysis of a sample of multitrait-multimethod data. A simulation study shows that the true orderings of intermodel proximities are received with a fair degree of accuracy. (Author/SLD)