Achieving Generality over Conditions: Combining the Multitrait Multimethod Matrix and the Representative Design of Experiments

摘要

Doubts about the generality of results produced by psychological research have been expressed with increasing frequency since Koch observed, in 1959, that there is 'a stubborn refusal of psychological findings to yield to empirical generalization'. Campbell and Fiske's (1959) multitrait multimethod matrix and Brunswik's (1956) representative design of experiments are combined and extended in order to increase our ability to generalize over conditions in both experimental psychology and the study of individual differences. A study of expert judgment illustrates the application of Campbell and Fiske's methodology to an experimental design involving multiple concepts and methods, and in addition, criterion measures for each concept. The criterion measures make it possible to complement Campbell and Fiske's internal validity matrix with an external validity matrix. Inclusion of the correlations among criteria in the external validity matrix is consistent with Brunswik's argument that generalization over conditions depends upon the representation of ecological relations among experimental conditions. Procedures are described for computing measures of convergent and discriminant validity for each matrix and for combining the data from both matrices.