Objective: Little research exists using the Trauma Symptom Inventory (TSI) with military samples especially those having both PTSD and blast injuries. The purpose of this study was to determine whether Item Response Theory (IRT) analysis could be used to create a shorter TSI for this population. Method: Participants included 1,017 Marines and sailors. Heat maps, Item Information Functions (IIF) and the multidimensional IRT package in the R program were used to determine which items could be eliminated. Results: Preliminary results display the following alpha coefficients for the TSI's 10 clinical scales: 0.86, 0.92, 0.92, 0.93, 0.90, 0.85, 0.82, 0.86, 0.85 and 0.75 (M = 0.866, SD = 0.055), which are comparable to Briere's TSI manual, except for the Inconsistent Response validity scale, with results of 0.93 compared to 0.55. Exploratory Factory Analysis (EFA) was also conducted, giving comparable results with Briere's TSI manual, except the loading of the Sexual Concerns scale, giving results of 0.53 compared to 0.78. EFA also concluded that there are two independent factors that compose the TSI: Generalized Trauma and Distress and Self-Dysfunction. The following scores are the loadings from each scale in Factor 1 for AA, IE, DA, DIS and D scales respectively: 0.83, 0.81, 0.77, 0.80, and 0.72. Conclusion: Preliminary results from heat maps and IIF's suggest that a significant number of items can be deleted from the TSI while still maintaining the same precision and accuracy as the longer version.
Item Response Theory Analysis of the Trauma Symptom Inventory in Military Related Blast Injuries
A Sekely, W Smith, A Puente; A-69
Item Response Theory Analysis of the Trauma Symptom Inventory in Military Related Blast Injuries. Arch Clin Neuropsychol 2015; 30 (6): 510-511. doi: 10.1093/arclin/acv047.69
Download citation file:
© 2017 Oxford University Press×