Item order effects in the evaluation of posttraumatic stress disorder symptom structure

Factor analytic research has demonstrated consistently that the 3-factor DSM-IV model of posttraumatic stress disorder (PTSD) symptom structure provides a poorer fit than alternative 4- and 5-factor models. In the current study we examined whether order of item presentation accounts for these findings. In a large sample (N = 1,311) of trauma-exposed undergraduates we conducted a series of confirmatory factor analyses using the PTSD Checklist and Posttraumatic Stress Diagnostic Scale, which present symptom items in the same order as DSM-IV, and the Detailed Assessment of Posttraumatic Stress, which presents items in a different order. Across all 3 measures the 3-factor DSM-IV model provided a relatively worse fit and the 5-factor dysphoric arousal model provided a relatively better fit compared with other tested models. We also examined the distinctiveness of 2 pairs of symptom clusters that appear in the dysphoric arousal model--avoidance versus numbing and dysphoric arousal versus anxious arousal--by comparing their patterns of associations with external correlates. Avoidance and numbing demonstrated differential associations with external correlates, as did dysphoric arousal and anxious arousal. Taken together, results indicate that order effects are unlikely to account for differences in relative fit between leading models of PTSD symptom structure. We discuss the need for future research in this area, especially studies designed to evaluate order effects more directly.