Test Reliability and Errors of Measurement at Several Levels of Absolute Isometric Strength

Abstract

Abstract Five maximal isometric wrist flexion strength trials of 5-sec. duration and spaced 1-min. apart were secured on two separate days one week apart. Intraclass correlation techniques were used to partition error variance and true score variance estimates on 75 college females classified into high, middle, and low levels of absolute strength groups; 30 male subjects were also included. Data were analyzed for left and right wrists for the female subjects and for the right wrist of males. The low strength females demonstrated the smallest error variance estimates for trials and days on both wrists. Middle and high strength females exhibited error variance estimates two to three times larger than the low strength females. Right wrist data on male subjects showed the largest error variance, with trials and day error variances some two to four times larger than the female subjects. Since error variance seems to be related to the magnitude of the strength score, further caution is indicated in the interpretation of reliability estimates for strength measures expressed in the form of a single reliability coefficient. The possibility that the importance of motivation in reliable assessment of maximal isometric strength may have been exaggerated was also discussed.