Making Software

The Third Experiment: Patterns in Team Communication

Better team communication is the third major hypothesis about design patterns. The third experiment tests the hypothesis that if team members have common pattern knowledge, they can communicate more effectively than without. Obviously, there is no way around observing teams communicating when testing this hypothesis. The difficulty is how to elicit communication and how to evaluate it. Thinking-aloud or talking to an observer seemed unnatural for teamwork, so we decided to pair subjects and record their interactions (on audio and video) during design maintenance tasks, and then analyze the recordings.

The experimental setup was as follows. Pairs of subjects were given a design maintenance tasks. For eliciting an explanation phase, one subject of each pair was given extra time in advance. During this time, the subject studied requirements and design documents of the program. This member will be referred to as the expert. There was no time limit for preparation, but all subjects took about an hour. After preparation, the second team member joined as novice. The expert explained the program design to the novice. Subsequently, both team members collaborated on two maintenance tasks. This strategy provided a chance to observe two interesting communication phases: (1) the phase of explanation by expert to novice and (2) the teamwork phase between two maintainers.

The independent variable is design pattern knowledge. First, teams were ...