DATA-ANALYSIS AS A HEURISTIC FOR ESTABLISHING THEORETICALLY FOUNDED ITEM STRUCTURES

The identification of structural dependencies between diagnostical ite ms is an essential prerequisite for the application of efficient adapt ive computer-based knowledge assessment procedures. A crucial question is, however, how empirically valid structures on item sets can be est ablished. On the one hand, there are methods which make use of expert knowledge or theoretical considerations which can be domain-specific, psychological, or educational. On the other hand, structures on item s ets can be derived from existing data sets by data-analytic approaches . In this work, a step towards an integration of both sides is made. W e contrast the results of an expert questioning procedure with the out comes of a data-analytic method. As a consequence of this comparison, we show how the integrated application of expert questioning resp. the ory-based methods and data-analytic procedures may facilitate establis hing well-founded structures on item sets.