The Application of Generalizability Theory to Reliability Assessment: An Illustration Using Isometric Force Measurements

• Published in: Physical therapy Vol. 73; no. 6; pp. 386 - 395
• Main Authors: Roebroeck, Marij E, Harlaar, Jaap, Lankhorst, Gustaaf J
• Format: Journal Article
• Language: English
• Published: United States American Physical Therapy Association 01.06.1993; Oxford University Press
• Subjects: Accelerated life testing; Adult; Analysis of Variance; Dynamometer; Dynamometers; Female; Humans; Isometric Contraction - physiology; Joints; Knee; Knee Joint - physiology; Medical research; Methods; Middle Aged; Muscular system; Physical Therapy Modalities - methods; Physiological aspects; Range of motion; Reference Values; Reproducibility of Results; Weights & measures
• ISSN: 0031-9023; 1538-6724
• DOI: 10.1093/ptj/73.6.386

The objective of this study was to examine the potential use of the generalizability theory for assessing reliability of muscle force measurements in clinical applications. Reliability is expressed in terms of standard error of measurement (SEM) and the indexes derived from the SEM. Using generalizability theory, potential sources of measurement error can be recognized and estimated. Ten healthy women, aged 23 to 47 years (means = 29.5, SD = 7.1), participated in the study. The method is illustrated by presenting a pilot study. Repeated measurements of maximal isometric knee extension force were carried out by two therapists. The time interval between measurement occasions was 1 week. A functional prototype of a hand-held dynamometer was used, and measurements were carried out according to standardized test protocols. Relatively important sources of measurement error were associated with interaction effects between subject and therapist as well as with interactions including the factor occasion. The SEM of the net knee moment was estimated for various hypothetical applications of the dynamometer. It is concluded that the SEM, the corresponding confidence interval, and the smallest detectable difference are practical measures for expressing reliability of measurements on an individual. Generalizability theory is a powerful tool for estimating the magnitude of multiple sources of measurement error and for assessing the reliability of measurements tailored to specific clinical applications.