Improving Research Methodology as Part of Doing Software Process Improvement

In the early writing on software process improvement (SPI), it was suggested that the group of people responsible for designing and implementing the SPI system should apply the principles of SPI on themselves, making sure their own processes were no less mature than the processes they were trying to improve. Looking at maturity level theory from various literature sources, it seems reasonable to say that the highest maturity level for people running SPI programs corresponds to the process of doing scientific research. We look at action research as a possible model for implementing SPI through a research driven climbing of the maturity level ladder, asking what sort of constraints the SPI theory puts on the design of the action research, and what sort of contributions can be made to SPI theory by applying action research. These two opposite questions are investigated by looking at a longitudinal case study of a SPI project that has gradually evolved into an action research project. The case illustrates a situation where it is unclear to which extent the practical problems need new theoretical insights to be solved, and also to which degree some theoretically chosen research problems are really relevant for solving the practical issues at hand. Based on this lack of convergence between relevant practical problems and interesting theoretical problems, we suggest one should take advantage of this expected dichotomy. One way of doing this could be to focus action research on how to improve the SPI process, only slightly guided by what practical SPI problems the practitioners are dealing with at the time of research. We make two contributions to theory by this suggestion on how to focus; (1) the action research approach, being an establish scientific approach in IS research, provides a systematic framework for producing new SPI knowledge to be published in SPI outlets and thus provides SPI with an external quality control loop for improving the SPI implementation, and (2) the SPI framework provides ideas on how to control action research efforts, suggestion one might control and improve the AR process just like one would control and improve any process in a SPI project. More specifically, using SPI to control and improve the AR process may typically mean that we are providing guidelines on how to select initial research questions, how to select theory, how to discuss the results and how to select research questions and theory for the next iteration. In conclusion, the marriage of SPI and action research provides a framework for systematically researching how to reach the highest possible SPI maturity level in the quickest and best possible manner.