| Popis: |
Rasch’s unidimensional models for measurement are conjoint models that make it possible to put both texts and readers on the same scale. Causal Rasch Models (Stenner et al. Frontiers in Psychology 4:1–14, 2013) for language testing fuse a theory of text complexity to a Rasch Model making it possible for a computer to response illustrate texts read by language learners during daily practice. Causal Rasch Models are doubly prescriptive. First, they are prescriptive as to data structure (e.s., non-intersecting item characteristic curves.) Second, they are prescriptive as to the requirements of a substantive theory. One consequence of this fusion of a Rasch Model with a substantive theory is that individual-centered growth trajectories can be estimated for each reader even though no two readers ever read the same article or respond to a single common item. Rather than common items or common persons being the connective tissue that makes comparisons of readers possible, common theory is the connective tissue just as is true in, say, human temperature measurement where each person is paired with a unique thermometer. Thus, although the instrument is unique for each person on each occasion, a text complexity theory makes it possible to convert counts correct to a common reading ability metric in each and every application. |
