PURPOSE: The Motor Assessment Scale (MAS) has shown to be a reliable and valid instrument for measuring motor functioning for stroke patients. The purpose of this paper was to study dimensionality and scalability of the instrument. METHOD: A total number of 137 patients were evaluated at admission and on discharge. To study dimensionality and scalability, Partial Credit Models (PCM) were applied for the eight items in the instrument and also for the upper limb and lower limb activities separately. Studies were performed to compare subgroups such as young/old, men/women and diagnosis to examine if the items had the same meaning for the subgroups. RESULTS: The items showed the same ordering for admission and discharge. Infit and outfit values varied between 0.5 and 1.9 at admission and 0.5 and 1.7 on discharge. CONCLUSIONS: We conclude that the scalability of the different items is not optimal. For some items, the top and/or bottom levels are overrepresented and for other items the middle levels tend to cluster. There seems to be a single construct except for one of the items. The instrument also showed the same meaning for the different subgroups.
PURPOSE: The Motor Assessment Scale (MAS) has shown to be a reliable and valid instrument for measuring motor functioning for strokepatients. The purpose of this paper was to study dimensionality and scalability of the instrument. METHOD: A total number of 137 patients were evaluated at admission and on discharge. To study dimensionality and scalability, Partial Credit Models (PCM) were applied for the eight items in the instrument and also for the upper limb and lower limb activities separately. Studies were performed to compare subgroups such as young/old, men/women and diagnosis to examine if the items had the same meaning for the subgroups. RESULTS: The items showed the same ordering for admission and discharge. Infit and outfit values varied between 0.5 and 1.9 at admission and 0.5 and 1.7 on discharge. CONCLUSIONS: We conclude that the scalability of the different items is not optimal. For some items, the top and/or bottom levels are overrepresented and for other items the middle levels tend to cluster. There seems to be a single construct except for one of the items. The instrument also showed the same meaning for the different subgroups.
Authors: Sophie Wang; C Janice Hsu; Lauren Trent; Tiffany Ryan; Nathan T Kearns; Eugene F Civillico; Kimberly L Kontson Journal: PM R Date: 2018-02-21 Impact factor: 2.298