Psychometric variables of the EMUN-AR scale using Rasch analysis

Abstract

Introduction: Self-report scales have been considered to be useful for evaluating and monitoring symptoms of manic-depressive illnesses. The EMUN-AR scale is one of such scales but it has not been validated.
Objective: To validate the EMUN-AR scale using Rasch models.
Materials and methods: The EMUN-AR scale measures 26 items to assess comprehensively symptoms of manic-depressive illnesses in three domains: Frequency, severity, and level of disturbance caused by the symptoms. To test the psychometric properties of this scale, we used the exploratory factorial and Rasch analyses. The Rasch analysis included unidimensionality, reliability, item fit, threshold ordering, and person-item maps.
Results: The study included 267 hospitalized patients, most of them with a manic episode diagnosis (n=204, 76.4%), and receiving pharmacological treatment. The factor structure was summarized in four domains: Activation, inhibition, depressive ideation, and maladaptive behavior. The Rasch analysis supported unidimensionality of the EMUN-AR factors, satisfactory levels of reliability, and appropriate item fit, except for one of the items measuring death or suicidal thoughts. However, the scale did not measure adequately the mild forms of the illness given its redundant and unordered thresholds.
Conclusion: The EMUN-AR was modified in two aspects: An item whose marginal indicators were poorly adjusted and the reduction in the number of categories. In its current form, the EMUN-AR is appropriate for measuring severe forms of the illnesses, but it does not adequately measure the mild forms of manic-depressive illnesses.