Lulu Zhang

Graduation Semester and Year

2016

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Information Systems

Department

Information Systems and Operations Management

First Advisor

Radha Mahapatra

Second Advisor

Sridhar P. Nerur

Abstract

The fundamental question of the present study is to investigate whether fit between representation and task type matters while developing software. The UML (Unified Modeling Language) includes diagrams that allow developers to represent software artifacts at different levels of abstraction. These diagrams facilitate design, specification, and testing during the software development life circle. However, they are expensive and difficult artifacts to maintain because of the l time pressure that developers in the software industry typically experience. The present study hypothesizes that the selective usage of diagrammatic representations would justify the cost of maintaining these representations. Student subjects participated in controlled laboratory experiments in which two factors were manipulated: task types and diagram types. The subjects’ comprehension of the system being developed as well as their coding performance were compared across experimental treatments. The results indicate that the effectiveness of diagram usage depends on the task type. The fit between diagram and task type has a positive relationship with both comprehension and coding quality performance. However, comprehension does not always fully mediate the relationship between fit and coding performance. The findings suggest that developers do not need all types of UML diagrams, but would benefit by choosing the ones that are compatible with the task at hand. Implications of the study for practitioners and academics are discussed as well.

Keywords

Diagrammatic representation, Experiments, Programming performance, Comprehension, UML

Disciplines

Business | Management Information Systems

Comments

Degree granted by The University of Texas at Arlington

Download

Share

COinS