How do individuals decide which modeling scripts to use during systems analysis and design?

Abstract

Because most information systems in real-world domains are complex, practitioners often use different types of conceptual modeling scripts to understand them. Modeling methods such as UML, for example, provide more than a dozen scripts for practitioners to use. We study how script readers decide which of several different scripts to use during systems analysis and design tasks. We carried out a free-simulation experiment to test how users select scripts based on two factors: combined ontological completeness and ontological overlap. We find that participants indeed decided to select more than one script to achieve a more complete domain representation. But, when they selected more than three scripts, they decided to remove scripts and reduce combined completeness to increase the clarity of the combined representation. Our results indicate that script readers prefer relatively less complete scripts with high levels of clarity over more complete, but more overlapping, script combinations. We detail the implications these findings have for the theory and the practice of conceptual modeling.