Development and evaluation of a lightweight root cause analysis method in software project retrospectives

Timo O.A. Lehtinen

    Research output: ThesisDoctoral ThesisCollection of Articles

    Abstract

    Software projects are famous for their problems. The most common problems include low software quality and schedule and cost overruns. According to the theory of causality, problems escalate from the causal relationships of mutually exclusive events. The consequences of the problems motivate software companies to improve their work practices. Improvement of work practices is based on the detection of occurred problems and the ana-lyses of their underlying causes. "Retrospective" is a term that refers to post-project activities where the occurred problems are considered in order to make improvements. Software project retrospectives include the detection of the occurred problems, reasoning their causes, and developing corrective actions. There are many retrospective methods that follow these three phases.Root cause analysis (RCA) is a structured investigation of problems to detect their under-lying causes. This dissertation considers the applicability of RCA in the retrospectives of small- and medium-sized software product organizations. The research focuses on three research problems. The first problem is to explain how to conduct RCA in collocated and distributed software project retrospectives. The second problem is to study whether RCA is perceived as efficient and easy to use. The third problem is to consider whether the outcome of RCA indicates how the causes of project failures are interconnected. In this dissertation, an RCA method (ARCA) is developed. Thereafter, it is evaluated in a to-tal of six industrial cases and in one controlled student experiment. Additionally, a software tool, called ARCA-tool, for improving the ARCA method is developed. The tool is evaluated in two industrial cases. The ARCA method includes four steps: target problem detection, root cause detection, corrective action innovation, and documentation of results. It requires ap-proximately five hours of team work. A total of 97 participants evaluated the ARCA method and it was used in a total of 19 software project retrospectives.One of the key contributions of this dissertation is the ARCA method, which is applicable to collocated and distributed retrospectives with the assistance of ARCA-tool. Another key cont-ribution is the empirical evaluation of the ARCA method. The method is perceived as useful, easy to use, and cost-efficient for detecting the causes of software project problems and deve-loping corrective actions. Furthermore, the method helps to understand what happened, whe-re it happened, and why it happened. It reveals interconnections between software process areas, which is important for process-improvement activities.
    Translated title of the contributionKevyen juurisyyanalyysimenetelmän kehitys ja evaluointi ohjelmistoprojektien retrospektiiveissä
    Original languageEnglish
    QualificationDoctor's degree
    Awarding Institution
    • Aalto University
    Supervisors/Advisors
    • Lassenius, Casper, Supervisor
    • Mäntylä, Mika V., Advisor
    • Itkonen, Juha, Advisor
    Publisher
    Print ISBNs978-952-60-5907-5
    Electronic ISBNs978-952-60-5908-2
    Publication statusPublished - 2014
    MoE publication typeG5 Doctoral dissertation (article)

    Keywords

    • root cause analysis
    • software engineering
    • retrospective
    • software project failure
    • usefulness
    • cost-efficiency
    • ease of use
    • field study
    • case study
    • controlled experiment
    • design science

    Fingerprint Dive into the research topics of 'Development and evaluation of a lightweight root cause analysis method in software project retrospectives'. Together they form a unique fingerprint.

    Cite this