>. So, I'd like to see more studies on this, but my experience and intuition still tells me that the benefits are there
> to be had.

I like to think that I'm a rational person. And when presented with the data, I base my behavior and decision making on the data that's presented rather than my beliefs. To do otherwise is irrational. What concerns me is that TDD has been practiced for well over 10 years … and yet there is still no conclusive data on the benefits of TDD (with the exception of the reduction of defects). Nor is there a multitude of data points to show a clear trend.

The absence of evidence is not evidence of absence, so ultimately I'd have to agree. I think the real problem that I'm trying to address here is that there are no well designed studies on TDD, and there is very, very little data from which to draw meaningful conclusions.

>. So, I'd like to see more studies on this, but my experience and intuition still tells me that the benefits are there
> to be had.

I like to think that I'm a rational person. And when presented with the data, I base my behavior and decision making on the data that's presented rather than my beliefs. To do otherwise is irrational. What concerns me is that TDD has been practiced for well over 10 years … and yet there is still no conclusive data on the benefits of TDD (with the exception of the reduction of defects). Nor is there a multitude of data points to show a clear trend.

The absence of evidence is not evidence of absence, so ultimately I'd have to agree. I think the real problem that I'm trying to address here is that there are no well designed studies on TDD, and there is very, very little data from which to draw meaningful conclusions.

My personal experience shows that TDD helps provide self-direction and self-guidance for less experienced programmers, thereby increasing their productivity. I don't really see any productivity impact with seasoned programmers.

Companies with highly leveraged models might be interested in the findings.

Nice article. I've read those studies too, and I globally agree with your assessment of what they say. However, I find these studies seem to go against my gut feeling and practical experience with TDD – when you get into a good TDD flow, it just *feels* more productive. And there are less bugs to fix. Studies are hard – these ones were with teams that were new to TDD, and didn't take into account maintenance costs over time, so I wonder if they give the whole picture of the benefits of TDD. I find, and I've heard from other TDDers, that the extremely low number of bugs that get through to production has a huge impact on maintenance costs. And, as you rightly imply, there are the overall quality aspects too, that are hard to quantify: executable documentation, safer changes, etc. So, I'd like to see more studies on this, but my experience and intuition still tells me that the benefits are there to be had.

- Cheers,

John Ferguson Smart