The impact of item order on the factor structure of the PTSD Checklist for DSM-5.

Abstract

The PTSD Checklist for DSM-5 (PCL-5) is the most widely used self-report measure of posttraumatic stress disorder (PTSD) and is frequently modeled as having four correlated factors consistent with the DSM-5 symptom structure. Some researchers have argued that item order may influence factor structure. Although two studies have examined this, they were both based on DSM-IV criteria, and neither utilized a randomized design. Thus, this study aimed to determine whether item order impacts the factor structure of the PCL-5, using two independent samples of community participants (N = 347, 67.7% female, 85.3% White) and veterans (N = 409, 83.6% male, 61.9% Black/African American). Approximately half of each sample was randomized to receive the PCL-5 in the original fixed order, whereas the other half received a uniquely randomized version. We compared the DSM-5 four-factor model to several theoretically relevant models and found improved model fit in the seven-factor hybrid model, community sample: ∆χ² = 153.87, p < .001; veterans: ∆χ² = 152.61, p < . 001. Consequently, the DSM-5 four-factor and seven-factor hybrid models were retained for invariance testing. Across both samples, measurement invariance was examined between the randomized and fixed-order groups. Configural invariance, partial metric invariance, and partial scalar invariance were achieved in both samples, ps = .054-.822, suggesting that the fit of the DSM-5 four-factor structure and the seven-factor hybrid model, as measured using the PCL-5, are not due to order effects. These findings support the continued use of the PCL-5 in a fixed fashion.