The Aphasiology Archive
University Library System, University of Pittsburgh
    

Rasch Modeling of Revised Token Test Performance: Construct Validity, Known-Groups Validity, and Sensitivity to Change

Hula, William D. and Doyle, Patrick J. and Mikolic, Joseph M. and McNeil, Malcolm R.
Rasch Modeling of Revised Token Test Performance: Construct Validity, Known-Groups Validity, and Sensitivity to Change. In Clinical Aphasiology Conference: Clinical Aphasiology Conference (2004 : 34th : Park City, UT : May 2004) / : (2004).

Abstract

The Revised Token Test (RTT) is a test of auditory language processing based in
classical test theory. Rasch analysis is an alternative psychometric approach that
employs probabilistic models of item difficulty and patient ability and can be used to
address issues of construct validity and test sensitivity, among others. In this
presentation, the Rasch model will be summarized and its utility for improving test
psychometrics will be demonstrated. Rasch analysis generally supported the construct
validity of the RTT, and Rasch-derived RTT scores demonstrated greater ability to detect
longitudinal differences in stroke survivors than did classically derived RTT scores.

EPrint Type:Clinical Aphasiology Abstract
Subjects:UNSPECIFIED
ID Code:1609
Conference:Clinical Aphasiology Conference: Clinical Aphasiology Conference (2004 : 34th : Park City, UT : May 2004)
Additional Information:ABSTRACT ONLY. No full text available.