Evaluating the Impact of Pair Testing on Team Productivity and Test Case Quality – A Controlled Experiment

Abstract

One of the main objectives of software testing is to uncover the maximum number of faults while consuming the least amount of resources. This research is an attempt to investigate the utility of an unconventional testing technique called pair testing in achieving this goal. In pair testing, two individuals sit together at one keyboard to test the software. An empirical study was designed and conducted to evaluate the performance of pair testing vis-à-vis conventional testing. Six pairs of testers divided into two different groups - one using pair testing and the other using conventional testing - participated in a controlled experiment involving three separate projects. The productivity of the groups and the quality of their work were quantitatively evaluated and compared. The results of comparison revealed that the group using pair testing spent more effort but the quality of its work was better.