Related Experiment Videos
Using Rasch Analysis to Validate the Motor Activity Log and the Lower Functioning Motor Activity Log in Patients With
I-Ching Chuang1, Keh-Chung Lin2,3, Ching-Yi Wu1,4
1Department of Occupational Therapy and Graduate Institute of Behavioral Sciences, College of Medicine, Chang Guang University.
Physical Therapy
|October 15, 2017
Summary
The revised Motor Activity Log (MAL) and Lower-Functioning MAL (LF-MAL) are reliable and unidimensional tools for assessing stroke patient arm function. However, a floor effect was noted, suggesting a need for easier items in future assessments.
Area of Science:
- Rehabilitation medicine
- Psychometrics
- Neuroscience
Background:
- The Motor Activity Log (MAL) and Lower-Functioning MAL (LF-MAL) are utilized to evaluate upper extremity function in stroke survivors.
- These tools assess both the amount of use and quality of movement of the more impaired arm during daily activities.
Purpose of the Study:
- To examine the psychometric properties of the MAL and LF-MAL in stroke patients using Rasch analysis.
- To validate the reliability and construct of these widely used assessment tools.
Main Methods:
- Rasch analysis was employed to evaluate unidimensionality, item difficulty, targeting, reliability, and differential item functioning (DIF).
- The study involved 403 patients for the MAL and 134 patients for the LF-MAL.
Main Results:
- Initial analysis revealed disordered thresholds and poor model fit for both MAL and LF-MAL.
- After revisions (collapsing categories, removing misfit items), the scales demonstrated unidimensionality and good reliability (coefficients .79–.87).
- No significant DIF was found, indicating no bias related to patient characteristics.
Conclusions:
- The revised MAL and LF-MAL are unidimensional and reliable measures for assessing arm activity in stroke patients.
- A floor effect was observed, suggesting the need for easier items to capture performance in lower-functioning individuals.
- The revised scales demonstrate good functioning categories and lack of response bias.