Similarities between retrospective and actual anxiety states

The purpose of the study was to examine the fit and calibration of the items in Spielberger, Gorsuch, and Lushene's (1970) State-Trait Anxiety Inventory during measurement of actual and retrospective anxiety. Subjects in the actual anxiety situation (n = 113) and in the retrospective anxiety situation (n = 55) were administered the inventory, and 20 scale items were analyzed individually by the probabilistic Rasch Model (Wright and Masters, 1982). Comparisons between the items' values in both situations revealed that 17 of the 20 items were rated similarly. In the retrospective anxiety state, 9 items (misfits) failed to discriminate between high- and low-anxious subjects, but only 6 failed to do so in the actual anxiety state. Despite the similarities, we recommend that the scales be modified to yield a more reliable measure of anxiety and to discriminate more accurately among subjects with varying levels of anxiety.