Validating code reviews conducted as part of a multi-level peer review process in a large-scale software development
Bo I. Sandén, Raymond T. Tillman · 2006
Software quality is a goal in industry and government software developments. Software defects continue to plague development activities and the cost involved in identification of defects is high. The earlier in the process that a defect is found, the less it costs to repair it. Peer reviews have been used to identify defects at early stages in the development process. Peer review groups of varying sizes and formality have been used for years. But research results have been mixed as researchers attempt to identify effective peer review methodologies. This research compares code peer reviews performed by nominal groups using electronic defect reporting (electronic reviews) with informal groups (informal desktop reviews) who performed informal desktop reviews involving informal defect reporting. Nominal groups were collections of individual reviewers who examined code independently at their desks and reported their findings to a review coordinator who consolidated results and took corrective actions. Informal groups examined the code independently and then walked through their findings informally at the desk of the developer while looking at the code in an editor. The results of this research show that the electronic reviews by nominal groups were significantly more effective than the informal desktop reviews by informal groups since the reviewers reported significantly more defects for the time each reviewer spent in the electronic reviews. The number of defects identified by nominal groups approximated that which was observed in previous research for more formal review processes. The informal desktop reviews resulted in significantly fewer defects being identified. Although the reviews examined twice the amount of code for time expended, they only identified one tenth the number of defects.