CITATIONS DO NOT SOLVE PROBLEMS

Campbell and Fiske's 1959 article made the now-obvious statement that measures of the same variable made by different methods should agree (converge) and certainly should agree better than measures of different variables made by those several methods. Almost everyone accepted that statement, yet the multitrait-multimethod matrices published today show little or no improvement over the published examples cited in 1959. The article raised the conceptual problem of how a variable and its measurement should be linked and the methodological problem of clarifying the nature of methods and their effects. Given these unresolved problems, it is perhaps not surprising that the question of the appropriate statistical analysis of these matrices has also no consensual answer. Perhaps methods and traits or contents are so thoroughly intertwined that their interaction cannot be adequately analyzed.