The effect of layout on the comprehension of UML class diagrams: A controlled experiment

Abstract

The results of a controlled experiment assessing the effects of different layout strategies on the comprehension of UML class diagrams of two software systems is presented. Six different categories of software comprehension tasks, with varying degrees of difficulty, are used to assess the layouts. Each task consists of several questions aimed at measuring the comprehensibility of a layout. The study involved 45 participants of varied experience in software design and programming ability. A report on the quantitative analysis of accuracy, speed, confidence level and preference of solving the tasks is given. Results indicate that clustered layouts demonstrate significant improvement in subject accuracy and speed in solving the problems in a majority of tasks.