DRAFT

Ideal measurement requires unidimensionality, additiv ity and objectivity. If this ideal could be achieved in the human sciences, it is expected that observations of the ability of persons with respect to items would be consistent with a stochastic approximation to Guttman’s scaling (Guttman 1944). Rasch presented a model that is built upon these expectations and described it with respect to probabilities: A person having a greater ability than another should have the greater probability of solving any item of the type in question, and similarly, one item being more difficult than another one means that for any person the probability of solving the second item correctly is the greater one. (Rasch 1960, 1980, 1993 1993 #52) as cited in (Wright 1999). Wright (1999) has nicely stated the assumptions, process and utility of Rasch measurement in the human sciences. As part of his exposition, Wright describes some of the ‘mistakes ’ (p. 65) that scientists have been making in their attempts at measurement and outlines the requirements for success in measurement. Among these principles Wright includes (a) avoid the use of models that fail to converge such as the two-parameter logistic (2PL) and three-parameter logistic (3PL) item response theory (IRT) models; (b) use models that have sufficient statistics to allow for independent estimation of parameters; and (c) do not use models that ”cause the hierarchy of relative item difficulty to change at every ability level