An empirical study of factors that affect user performance when using UML interaction diagrams

During the requirements process it is of key importance that all representations used are clearly understood by those who must use them. Therefore it is essential to ensure that those representations are presented as effectively as possible. The research reported in this paper relates to an empirical study carried out to investigate factors which might affect user performance when using UML interaction diagrams. Several variables were investigated in the study; these were identified from the related literature and earlier research by us as being important in understanding interaction diagrams. The independent variables investigated in the study were diagram type, user pre-test and post-test preference, individual's cognitive style, text direction, scenario type and question type. Time taken to formulate the correct answer was the dependent variable used as the measure of performance. Statistical analysis of data showed significant differences for several variables, including diagram type, preference, and scenario type (p<0.05).