In an experiment error-related emotional reactions of users (n = 16) working on a data base system were investigated with the aid of a post-action video self-confrontation. First, error data were classified according to the taxonomy of Frese and Zapf (1991). Based on current theories of emotion; assumptions concerning a) temporal position of emotional reaction within error process, b) correlation between error type and degree of emotional reaction, and c) correlation between degree of emotional reaction and performance given in error handling were investigated. The results allow a differentiation of the assumptions concerning error-related emotional reactions of users into different qualities of emotion. There are indications that a reaction of anger can already be expected at the stage of error detection and that the degree of helplessness experienced during error detection rises as the complexity of error increases. Furthermore, there is a positive correlation of the degree of helplessness experienced during error detection and the error handling. The results imply suggestions for the conception of error training for users as well as system design. Special attention should be paid to the stage of error detection, and to the sensitive stage of reactions of helplessness which reduce performance.
