Visual Posture Observation Error and Training

Abstract

The purpose of this study was to determine people's ability to visually estimate postural angles of the shoulder, trunk, and wrist. One application of these findings is to determine the effect of estimation error on common risk analysis tools that incorporate posture. Considerations are given to the effect of training, video mode, gender, body region, and subject characteristics on estimation error. Absolute error, algebraic error, and subject confidence are used to characterize visual estimation abilities. Results indicate that visual estimation error ranges between 7 and 10 degrees. Error further increased with wrist postures and female observers. Due to estimation errors, analysis tools that include posture are less accurate in predicting risk of injury. Eight, 12, and 14 percent of shoulder, trunk, and wrist postures, respectively, were misclassified causing Rapid Upper Limb Assessment (RULA) scores to shift by at least one point. For the Strain Index, forty percent of wrist postures were misclassified by participants causing as much as a two-thirds change in the final score.