We
ran an experiment with 42 seniors at the University of Utah. Fourteen of them worked alone. The rest worked in
pairs doing pair-programming. All students completed the same assignments.
Pair-programming definitely
does not cost twice as much! In their first assignment (I call the "jelling-assignment") the pairs spent
60% more programmer hours than the individuals. In the second assignment, they had gotten used to this pair-programming
thing. The pairs spent only 20% more total time than the individuals. By the third assignment, the pairs spent
only 10% more time - so if an individual spent 10 hours on the assignment, the pair worked together for 5 hours
and 15 minutes.
In all cases, the pairs
passed about 15% more of the post-development test cases. |
![Laurie Williams](images/laurie.gif)
And, over 90% say they enjoy programming more and they feel more confident in their work when pairing.
As a posttest to the experiment,
all students worked individually to complete one assignment. One student said of going back to solo programming,
"Without my partner, I feel like I lost half my brain."![](../images/xplinktn.gif)
![](../images/pixel.gif) ![Laurie Williams](images/williams.gif)
University of Utah |