Mokken scaling and principal components analyses of the CORE-OM in a large clinical sample

In a sample of 543 adult National Health Service (NHS) patients referred to a Psychological Therapies Service, the responses to the Clinical Outcomes in Routine Evaluation-Outcome Measure (CORE-OM) self-report questionnaire were examined using conventional principal components analysis (PCA) and a unique application of Mokken Scaling Procedure (MSP). Following the theoretical views of G. A. Foulds, it was suggested that some items more properly belong to the universe of attitudes and traits rather than that of symptoms and states. Accordingly, the analyses were carried out both with and without the CORE-OM Risk domain items. Both PCAs produced a very large first component of Psychological distress, while the small second component differs. With all items included, the second component was of Risk. With the risk items excluded, the second component was now Functioning. The MSP results, respectively, were of a five-item scale of Functioning (impaired by depression) and on the second analysis, a five-item Functioning scale (impaired by anxiety). There was discussion on the criteria for item selection, the time scale specified in questionnaire instructions and the optimum number of items required for a symptom scale. It was concluded that the CORE-OM item pool did not conform to its purported face validity domains and subdomains, but predominantly constitutes a large Psychological distress scale with considerable item redundancy.