Supporting Software Inspections through Fault Content Estimation and Effectiveness Analysis

Sammanfattning: This thesis is focused on methods to improve software inspections. Software inspections is a method used to increase and assure software quality by carefully scrutinizing software artefacts throughout the development. Inspections improve product quality through the detection and removal of faults, i.e. imperfections that could lead to failures. This thesis reports on research to support decisions during the execution of the inspection process through fault content estimation and efficiency analysis. Fault content estimation is concerned with estimation of the remaining number of faults in the inspected documents. The estimate provides support to decisions regarding whether the artefacts should be reinspected or released to the subsequent development phase. The thesis shows through analysis of empirical data that improvements of estimation methods can be made, for example, by utilising experience data. Moreover, the thesis includes two replications, one that disagrees with the previous results of one method’s capability of avoiding extreme estimates, and one that shows conformance with the recommendation to prefer a sub-estimator instead of a composite version. Effectiveness analysis is proposed to support the management of inspection resources. A technique that uses sampling to guide the decision of where to focus the inspection effort is proposed. The technique is evaluated in a simulation and shows potential to increase the efficiency of software inspections. In addition, this thesis investigates the use of inspection effectiveness analysis on both individual and group level. The analysis can be used to increase knowledge and awareness of inspection impact and thereby support the planning of inspections. The thesis also presents a feasibility study of how similar measures could be used to benchmark inspection processes in search for best practices.