Abstract
[Purpose] The purpose of this study was to investigate the item-level psychometrics of a Korean translation of the Gross Motor Function Measure (GMFM) using Rasch analysis. [Methods] GMFM data for 206 children (mean age: 4.9 years) with cerebral palsy were collected in Korea. Focusing on the psychometric properties of the item measures, person measures, and rating scale categories, Rasch analysis was performed. Principal components analysis in combination with fit statistics was used to verify the unidimensional construct of GMFM. [Results] The original 4-point rating scale revealed that it was not being used in the expected fashion and could be improved by collapsing into a 3-point version. Eighty-eight items showed a logical item difficulty hierarchy. Furthermore this revision instrument appears to have good internal construct validity showing a person separation reliability of 0.99. The GMFM has a person separation index of 10.67 which indicates that the instrument separates the sample into approximately 15 statistically significant strata. [Conclusion] The rescaled version of the GMFM shows outstanding item-level psychometric qualities. This study suggests that the Korean version of the GMFM produces valid measures of gross motor ability in children with CP and furthermore appears to be sensitive in differentiating children with different gross motor abilities.
