Abstract
Objective: To examine internal construct validity of the Neurobehavioral Functioning Inventory (NFI) by applying
Rasch analysis. Setting: An outpatient rehabilitation program trial in New Zealand employing a goal-setting
intervention in people with traumatic brain injury (TBI). Participants: One hundred eight people (mean age = 46
years; 73% male) between 6 months and 5 years post-TBI. Design: Rasch analysis of the NFI (Partial Credit Model).
Results: Three NFI subscales were not unidimensional and at least 4 items in each subscale had disordered response
categories. Two items showed differential item functioning by age, 1 item by educational attainment, and 2 items
were found to misfit the overall construct. These items were excluded from the total score calculation. The revised
scale fit the Rasch model and supported the internal construct validity of the NFI. Conclusions: Current scoring
of the NFI subscales for people with TBI in New Zealand does not meet the requirements of the Rasch model. The
revised version of NFI can improve the interpretation of scores but should be further tested with people with TBI
in other settings.
