INTERPRETING RESULTS FROM RASCH ANALYSIS 2. ADVANCED MODEL APPLICATIONS AND THE DATA-MODEL FIT ASSESSMENT

The Rasch Analysis transforms raw scores of questionnaires into measures, satisfying fundamental measurement axioms (e.g., unidimensionality, linearity, generalizability). These measures may foster the advancement of the scientific assessment of behaviours, perceptions, skills, attitudes, and knowledge so frequently faced in Physical and Rehabilitation Medicine, not less than in social and educational sciences. The flaws of raw scores are well known: they are poor approximations of objective measures. A previous companion article illustrated the most frequent graphic and numeric representations of results obtained through Rasch Analysis. The present paper presents developments and advanced practical applications to construct questionnaires and person measurements. Suggestions are given on interpreting and managing the inevitable discrepancies between observed scores and ideal measures (data-model "misfit"). Twelve practical take-home messages for the appraisal of published results are provided