Exploratory Comparison of Expert and Novice Pair Programmers

Abstract

We conducted quasi-experiment comparing novice pair programmers to expert pair programmers. The expert pairs wrote tests with a higher instruction, line, and method coverage than the novice pairs and changed the given program skeleton to a larger extent. However, the expert pairs were also slower than the novice pairs. The pairs within both groups switched keyboard and mouse possession frequently. Furthermore, most pairs did not share the input devices equally but rather had one partner who is more active than the other.