Abstract
ABSTRACTBackground: The introduction of the WHO ICD-11 created a need for reliable and valid measures of Post-Traumatic Stress Disorder (PTSD) and complex PTSD (CPTSD). The International Trauma Questionnaire (ITQ) has been applied to different samples and cultures. Previous research using the original ordinal polytomous items supported the construct validity of the ITQ across different language versions in a sample of refugees in treatment and found evidence of locally dependent items and differential item functioning (DIF) relative to gender and time since trauma in the PTSD subscale. Another strand of research on the measurement properties of the ITQ has used dichotomized items and focused on discovering the model that best described the data found no evidence of DIF.Aim: To investigate the consequences of using dichotomized ITQ PTSD items for detecting DIF, other departures from the Rasch model as well as implications for measurement precision.Methods: We used Rasch and graphical log-linear Rasch models for the analysis, as these models have previously been employed in the only psychometric study using the original polytomous ITQ PTSD items.Results: The use of dichotomized PTSD items lead to detection of less DIF than previous research with the polytomous items, and it accentuated local dependence between items and DIF relative to gender. Measurement by scores over dichotomized items increased the standard error of measurement and reduced the reliability to a level, where psychometric theory would conclude that the measure of PTSD was inapplicable. In contrast, previous research has shown measurement by the polytomous ITQ PTSD items to have precision and reliability sufficient for screening for PTSD.Conclusions: The original polytomous ITQ PTSD items are recommended for purposes of studying measurement properties of the ITQ and treatment effects.