Abstract
Cognitive perspectives and IRT models in the development of achievement tests: A preliminary explo ration of information functions. Achievement tests are good tools to implement the new psychometric technologies which are playing an important role in the modern definition of achievement. The view of achievement has changed from the behaviourism, which used to view it as the measurement of overt behaviours without reference to internal cognitive processes, until nowadays in which from both, Cognitive Psychology and Performance Assessment movements, the need to explore the internal processes brought into play by the examinee is considered as necessary. From this cur rent view achievement is assimilated to the measurement of the mental structure of a knowledge structure. Item response theory is the basic instrument for scaling individual knowledge. This is a preliminary study in which we explore the information functions of several IRT models as scaling models of complex knowledge. We standardize the concept of mental model, coming from performance assessment, asking a relatedness (5 point scale) judgement from the examinee for all pairs of a series of concepts. In this work we explore and compare the information functions of several politomous models, and one dichotomous model (two parameter model). Results show that generalised partial credit model is appropriate to measure this new conception of achievement.
