Felix Ehrlenspiel

Research investigating competitive anxiety is typically based on the Competitive State Anxiety Inventory-2 ( Martens, Burton et al., 1990 ), its revision, or its derivatives in other languages. The psychometric quality of these scales as a whole has been determined according to classic test theory but it has not been assessed at the item level. Thus, the present research assessed the psychometric properties of the German derivative of this instrument, the German Wettkampfangst-Inventar-State (WAI-S; Ehrlenspiel et al., 2009 ) at the item level and tested a revised and extended version of the WAI-S by using an analytical approach based on item response theory. In Study 1, graded response models (GRMs) were fitted to the three subscales of the WAI-S using the original dataset collected for its development. We provided evidence indicating the sufficient psychometric properties of the somatic subscale, but we also identified some unfavorable properties of items in the cognitive anxiety and the self-confidence subscales. Thus, in Study 2, a revised and extended 16-item version of the WAI-S with 10 new items was generated and administered to a new sample of 322 German athletes. On the basis of the GRM results, we identified items with poor quality. We subsequently selected 12 appropriate items to generate a revised version of the WAI-S with stronger psychometric properties than the original instrument. We believe that the revised version of the WAI-S will increase the usability of competitive state anxiety assessments in both research and practice.