Empirical validation of pair programming

Abstract

This paper discusses an empirical assessment of pair programming. Several agile processes arose and have been adopted in industry with the promise of producing working software in the early phases of production process and to make the organization flexible and adaptable with respect to the changing environment. The outcomes of the experiments can be summarized as follows: (i) pair programming improves the productivity of developers with respect to solo programming; (ii) it also fosters knowledge transfer between the developers of the pair: the experiment suggests that this effect is emphasized when both the components own the same educational background; and, finally, (iii) distributing the components of a pair could seriously deteriorate expected benefits, if an appropriate support for communication and collaboration is not set up.