In the May/June 2007 issue of IEEE Software magazine there was an awesome series of articles about Test Driven Development and I’ve just discovered that one of them – the main one – is now freely available: Guest Editors’ Introduction: TDD–The Art of Fearless Programming by Ron Jeffries and Grigori Melnik.
Go get it, read it all and spend some time studying Table 1 on page 28:
In particular the last two columns: Productivity effect and Quality effect
Looking at the tables i see very little reason to be happy.
“Tried it for 3 hours and did not see a significant improvement” can hardly be called science.
Wim, also considering only a single “data point” and not reading the whole article can hardly be called science
It says: “many studies don’t have the statistical power to allow for generalizations. [...] So, we advise readers to consider empirical findings within each study’s context and environment.”
I find interesting that at different scales of time, people, size, language the majority of the industry studies ended up improving the quality considerably investing only a relatively small amount of additional effort.