Evaluating the Impact of User Story Quality on the Modelers Ability to Understand and Structure Requirements

Scrum is essentially driven by user stories (US). The development team indeed uses, to fill the project and sprints’ content, sentences describing the user expectations w.r.t. the software. US are often written “on the fly” in structured natural language so their quality and the set’s consistency are not ensured. The Quality User Story (QUS) framework intends to evaluate and improve the quality of a given US set. Other research has built a unified model for tagging the elements of the WHO, WHAT and WHY dimensions of a US; each tag representing a concept with an inherent nature and granularity. Once tagged, the US elements of the set can be graphically represented through an icon and the modeler can link them when inter-dependencies are identified. One or more Rationale Trees (RT) are then built. This paper presents the result of an experiment conducted with novice modelers aimed to evaluate how well they are able to build a RT out of (i) a raw real-life US set (group 1) and (ii) a new version of the US set improved in quality using QUS (group 2). The experiment requires test subjects to identify the nature of US elements and to graphically represent and link them. The QUS compliant US set improved the ability of the test subjects to make this identification and linking. We cannot conclude that the use of the QUS framework improved the understanding the problem/solution domain but it increased the ability of modelers to identify Epic US. All in all, building an RT has a positive impact on identifying the structure of a US set’s functional elements.