An empirical approach to on-line learning in SIETTE

Abstract

SIETTE is a web-based evaluation tool that implements CAT theory. With the help of a simulation program, different empirical experiments have been performed with SIETTE with two different goals: a) to study the influence of the parameters of characteristic item curves and selection criteria in test length and accuracy; and b) to study different learning strategies for these parameters. The results of the experiments are shown and interpreted.