An investigation into how error taxonomies can generate actionable data

Abstract

Errors are phenomena that individuals and organizations are all familiar with and must accept to be inevitable. In an organizational context, errors are often feared because of the potential negative outcomes they may lead to. Therefore, much thought and effort has been invested by both academics and practitioners into understanding errors and mitigating their negative effects. One way in which organizations do this is to use error taxonomies, that are essentially systems for categorizing errors, in hopes of highlighting error trends such as error recurrence. However, a major concern is how actionable the data that these error taxonomies produce is. This study was done in collaboration with an anonymous case company at their R&D department and this company has been the sole source of empirical data. The purpose of this study is to understand the usefulness of the current error taxonomy, at the case company, in assisting to measure error recurrence patterns. The research design of the study is mixed, using both quantitative and qualitative data collection methods. The results highlight several aspects of the current error taxonomy that make it difficult to make sense of the data that is being generated. Most importantly, the data is not deemed to be actionable, and it does not provide good support for measuring error recurrence. Therefore, this study makes actionable recommendations on how the error taxonomy can be improved, based on a variety of error related literature and the empirical findings from this study. The recommendations are primarily aimed at the case company, but it is deemed that they are relatively generalizable to any organization that is working with an error taxonomy. The case company was very pleased with the results of this master’s thesis, and they assured that it will act as a valuable source of input for management in the discussions on how to proceed with their error taxonomy.