Rasch calibration of the 25-item Connor-Davidson Resilience Scale.

Rasch modeling was used to examine the 25-item Connor-Davidson Resilience Scale within adults (n = 410) in a weight management program. Rasch analysis assessed model-data fit, item difficulty and person's resilience level, an item-person map to evaluate relative distribution items and persons, and rating scale function. Four misfit items were identified and removed. Item difficulty ranged from 1.25 to 1.19 logits (higher logit values indicate more difficult items). Persons' resilience level had wide distribution (resilience = 2.27 ± 1.56 logits). Item difficulty levels did not adequately assess higher resilience levels. An improved inventory that measures a wider range of resilient behaviors would improve measurement quality.