Dimensionality of the Chinese Version of the Cognitive Failures Questionnaire

Abstract

This study aimed to examine the measurement properties of the Chinese version of Cognitive Failures Questionnaire (CFQ) using the Rasch model analysis. This model allows us to construct linear measures from ordinal scale. In this technique, one first transforms different performance raw response scores to a “logit” metric. Then, each raw performance score is transformed to a linear logit scale of “difficulty” along which different levels of difficulty can be directly compared with one another. A total sample of 262 subjects (102 patients with traumatic brain injury (TBI) and 160 normal controls) were recruited. Equal interval measures of CFQ were developed that exhibited good reliability and validity. The item and subject separation were found to be 3.29 (similar to Cronbach alpha of 0.92) and 5.57 (Cronbach alpha of 0.97). Principal component analysis of correlations among standardized residuals showed that all 25 items cohered to define a single construct of general cognitive failure. A common metric on the individual item scores was also established to discriminate patients from the normal controls. In conclusion, these findings indicate that Rasch Model analysis can offer us an alternate method to determine the measurement properties of the CFQ as well as to set up an efficient profile measure of everyday life cognitive failures to discriminate patients from normal controls.