Is a Picture worth a Thousand Words?

Use cases specifications are artifacts employed in all stages of software development, from the requirements elicitation to implementation. During this process, issues related to ambiguity, redundancy, inconsistency, and incompleteness can affect these specifications. These issues can harm software engineers' understanding and, consequently, affect the software quality. Given this context, this paper describes an empirical study to evaluate two different use cases specifications approaches (textual and graphical-based forms). We compared the approaches by assessing the degree of correctness and the time spent to generate the specifications. In addition, we performed an analysis focusing on evaluating the ease of use and usefulness of each approach. The quantitative results showed that textual form and graphical-based specifications presented similar levels of correctness and the time spent to generate them were also similar. The qualitative results indicated that the subjects had difficulties using both approaches; however, subjects stated that graphic-based specifications were easier and more useful to specify use cases.