Do Your Inspections Work?

[article]
  • (an inspection that goes much faster than average probably missed some defects).
  • Correlate pairs of metrics, such as defect density and preparation time (slower preparation generally finds more defects).
  • Use statistical process control to monitor key parameters, such as defect density (an unusually low defect density could indicate that inspectors missed some defects).

Inspection data can also indicate whether you are catching defects in the same lifecycle phase in which they were created. Because the cost of dealing with defects in subsequent stages increases rapidly, your goal should be 100 percent defect containment, with none leaking into downstream work products.

Some Measurement Caveats
The prime directive of software measurement is that a manager must neither reward nor punish individuals for their metrics results. The first time a practitioner is punished for some data he reported is the last time the manager will get accurate data. Aggregate the data from multiple inspections to monitor averages and trends in your inspection process without compromising the privacy of individual authors.

Beware of measurement dysfunction, which arises when the measurement process leads to counterproductive behaviors. Forms of inspection measurement dysfunction include defect severity inflation or deflation, and distorted defect densities, preparation times, and defect discovery rates. Inspectors who are rated on how many defects they find will report many defects, even if it means debating whether every small issue truly is a defect. Authors whose performance evaluation depends on how many defects inspectors find in their products will avoid inspections, fudge the data, or spend excessive time perfecting a product before inspecting it.

Your Bottom Line
I don't know how many defects your inspectors will find. However, many organizations have saved considerable effort through software inspections, which provides an excellent counter-argument to resisters who fear that inspections will slow the project down. The ROI only has to be a little higher than 1.0 to make inspections worth doing.

About the author

Karl E. Wiegers's picture Karl E. Wiegers

Karl Wiegers, Ph.D., is the Principal Consultant at Process Impact in Portland, Oregon, and the author of Software Requirements (Microsoft Press, 1999) and Creating a Software Engineering Culture (Dorset House, 1996). You can reach him at www.processimpact.com. You can find more than 35 of Karl's articles at www.processimpact.com/pubs.shtml

AgileConnection is one of the growing communities of the TechWell network.

Featuring fresh, insightful stories, TechWell.com is the place to go for what is happening in software development and delivery.  Join the conversation now!

Upcoming Events

Oct 12
Oct 15
Nov 09
Nov 09