Multi-Perspective Analysis of Software Development: a method and an Industrial Case Study

Abstract: In November 2009 a project was started between Chalmers University of Technology and aSwedish company developing safety critical software, in this report referred to as CompanyA. Thegoal of the project is to investigate and improve CompanyA's Verification and Validation practices.A large contributing factor to why the research project was started is because of a European projectthat will affect CompanyA’s market domain.
The study, from which the data for this report was taken, was conducted as a master thesis atCompanyA with the goal of producing a state-of-practice description of the company. The researchdesign was based around an analytical method using an analytical matrix that was developedspecifically for the study. The matrix consisting of two dimensions where the first dimensionconsists of the four BAPO aspects of Business, Architecture/Technology, Process and Organization.The second dimension of the matrix consists of three points in time, Past, Current and Future (PCF).The matrix was constructed in this way to allow collection and analysis of data from multipleperspectives as well as from a chronological perspective. The matrix provides a broad yet definedscope giving a good basis for practice oriented exploratory research. The analytical matrix can beused to structure the research in all aspects of a study, from the definition of research questions toresult demonstration. Data elicitation was conducted using several different research methods, suchas interviews, structured observation, document analysis, watercooler discussions and surveys. Themethods were used in different stages of the study to triangulate the findings and increase thevalidity of the collected data.
The focus of this report is to describe the research design and the general methodology that wasused during the study, primarily the BAPO/PCF analytical matrix, and to show its strengths andweaknesses through an industrial case study. By using the matrix it was possible to show how thecompany evolved over time in the BAPO aspects and to discuss how these changes are tied to eachother. An example would be how the company’s expansion in the international market has affectedthe company’s processes and the process evolution. The methodology hence proved to be effectivein finding information on a broad scale, which is beneficial for exploratory inductive research. Theresults from the industrial case study show that the methodology, with the analytical matrix at itscore, is a powerful research tool for any exploratory empirical research within the domain ofsoftware engineering.