Leigh A Lehman1, Michelle Woodbury, Craig A Velozo. 1. Department of Psychology, University of South Carolina Upstate, 800 University Way, Spartanburg, SC 29303, USA. LALehman@aol.com
Abstract
OBJECTIVE: The Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire is a global scale evaluating the impact of upper-extremity disorders. We sought to validate or challenge the unidimensional factor structure of the DASH. METHOD: Secondary analysis was performed on data collected from outpatient clinics. Factor analysis was performed in two steps, exploratory factor analysis (EFA) followed by confirmatory factor analysis (CFA). To provide further support for unidimensionality, fit statistics were calculated using the Andrich Rasch rating-scale model. RESULTS: EFA revealed three potential factors (eigenvalues = 18.40, 1.56, and 1.54). CFA was performed fitting a three-factor model. Tucker-Lewis Index (.99) and standardized root mean square residual (.05) values indicated good fit. Comparative fit index (.89) and root mean square error of approximation (.13) did not. When divided into three constructs, only one item misfit. CONCLUSION: More research is needed to determine situations in which division of the DASH may enhance interpretability.
OBJECTIVE: The Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire is a global scale evaluating the impact of upper-extremity disorders. We sought to validate or challenge the unidimensional factor structure of the DASH. METHOD: Secondary analysis was performed on data collected from outpatient clinics. Factor analysis was performed in two steps, exploratory factor analysis (EFA) followed by confirmatory factor analysis (CFA). To provide further support for unidimensionality, fit statistics were calculated using the Andrich Rasch rating-scale model. RESULTS:EFA revealed three potential factors (eigenvalues = 18.40, 1.56, and 1.54). CFA was performed fitting a three-factor model. Tucker-Lewis Index (.99) and standardized root mean square residual (.05) values indicated good fit. Comparative fit index (.89) and root mean square error of approximation (.13) did not. When divided into three constructs, only one item misfit. CONCLUSION: More research is needed to determine situations in which division of the DASH may enhance interpretability.