THE TRUE COST OF PAIR PROGRAMMING: DEVELOPMENT OF A COMPREHENSIVE MODEL AND TEST

Abstract

To advance the nomological net and theory, this dissertation proposed a comprehensive pair programming research model where the relationships among system complexity, programming methods, pair composition, effort, duration, defect rate, knowledge transfer, and various cost constructs were investigated. A multi-method, multi-study empirical approach was adopted. The survey method was employed for Study 1, and the bootstrap simulation method for Study 2. The reponses from 191 industry software developers and the simulation results suggest the previous conclusions regarding pair programming are limited in nature and the pair programming approach may not be as desirable in all situations as was previously assumed. The pair programming approach clearly adds value in situations where it is appropriate but certain conditions must be met for this goal to be achieved. Pair composition must be taken into account, and it is important to examine the interactions of multiple cost factors such as defect, effort, duration, and knowledge transfer and consider their combined effect on the ultimate goal of the project.