On Choosing a Model for Measuring

This paper describes an approach to the issue of selecting a measurement model that is based on a comprehensive framework for measurement consisting of four conceptual building blocks: The construct map, the items design, the outcome space, and the measurement model. Starting from this framework of building blocks, the measurement model selected must conform to the constraints imposed by the other three components. Specifically, to preserve the interpretability of the construct map, the models must preserve the order of items throughout the range of person locations, and must do so in a way that is consistent with the interpretational requirements of the map. In the case of item response modeling, this translates into selecting models that have a constant slope—i.e., they must be from the Rasch family. In the conclusion, some next steps in investigating these issues are discussed.