Is going beyond Rasch analysis necessary to assess the construct validity of a motor function scale?

Publication date: Available online 3 April 2018
Source:Archives of Physical Medicine and Rehabilitation
Author(s): Tiffanie Guillot, Sylvain Roche, Pascal Rippert, Dalil Hamroun, Jean Iwaz, René Ecochard, Carole Vuillerot
ObjectiveExamine whether a Rasch analysis is sufficient to establish the construct validity of the Motor Function Measure (MFM) and discuss whether weighting the MFM item scores would improve the MFM construct validity.DesignObservational cross-sectional multicenter study.Setting23 physical medicine departments, neurology departments, or reference centers for neuromuscular diseasesParticipants911 patients (aged 6 to 60 years) with Charcot-Marie-Tooth disease (CMT), facio-scapulo-humeral dystrophy (FSHD), or myotonic dystrophy type 1 (DM1)InterventionsNone.Main Outcome Measure(s)Comparison of the goodness-of-fit of the Confirmatory Factor Analysis (CFA) model vs. that of a modified multidimensional Rasch model on MFM item scores in each considered disease.ResultsCFA model showed good fit to the data and significantly better goodness-of-fit than the modified multidimensional Rasch model whatever the disease (P<0.001). Statistically significant differences in item standardized factor loadings were found between DM1, CMT, and FSHD in only 6 out of 32 items (Items 6, 27, 2, 7, 9 and 17).ConclusionIn the case of multidimensional scales designed to measure patient abilities in various diseases, a Rasch analysis might not be the most convenient whereas a CFA is able to establish the scale construct validity and provide weights to adapt the item scores to a specific disease.



https://ift.tt/2GyJg0M