Examination of a bifactor model of the Obsessive-Compulsive Inventory-Revised

The current study evaluated a bifactor model for the Obsessive-Compulsive Inventory Revised (OCI-R) in two samples, with (n=188) and without (n=408) obsessive-compulsive disorder (OCD). Since the OCI-R has been traditionally conceptualized as a correlated six-factor instrument, our goal was to evaluate the fit of a bifactor model against three competing models (one general factor, six correlated factors and a hierarchical model). Results revealed that the bifactor model presented the best fit and that all OCI-R items were influenced by a general factor and by one of the six dimensions. Model-based reliability estimated via Omega hierarchical for the total score suggested that OCI-R items are multidimensional, with the general obsessive-compulsive (OC) factor accounting for little more than half of the variability in the items of the total score, with the remaining variability accounted for by the specific OC domains and other sources of variance. Measurement invariance of the bifactor model was also supported among those with and without OCD. Implications for better understanding the structure and further use of the OCI-R are discussed.