In a Monte Carlo simulation, a methodology was developed to investigate the existence of radex properties among objective test items. In an experiment with items covering four categories of Bloom's cognitive domain taxonomy, the items did not have the factorial properties of a radex with four levels of complexity. (Author/BW)

Using three different forms of a 61-item administrator/supervisor questionnaire, the effects of both order and grouping were examined in separate analyses. For low-inference type items, factor structures obtained from survey data appear to be affected by item order, and, to a lesser extent, by grouping items. (Author/PN)