posted 08 August 1998 06:19 AM               

Subject: Re: Q: SPC Sampling for Software Manufacturers/Botta/Grout
Date: Mon, 3 Aug 1998 08:56:35 -0600
From: ISO Standards Discussion

From: jgrout@campbell.berry.edu (John Grout)
Subject: Re: Q: SPC Sampling for Software Manufacturers/Botta/Grout

> My company and I are going to branch out into new (for us) areas of SPC.
> Areas that are beyond just running charts strictly from customers AFTER
> the software has been sent. I have done research into SPC and sampling,
> but keep butting my head against a wall as to how/where I can sample the
> software in the design mode. I was hoping that someone out there,
> preferably a software company, could give me some feedback. Please keep
> in mind that I am not a CQE......

Tanya,

I wish the best in implementing SPC for software quality. I hope you can achieve great success using it. However, there are other approaches you might consider. I've been reading recently that single occurrences of human errors tend to be rare and intermittent. They are often thought of as "common causes" of variation, and not easily managed using control charts.

My suspicion is that even though single occurrences are rare that as a
group human errors are a significant source of defects in designing software. Hewlett-Packard and Motorola are using a technique developed in Japanese manufacturing (poka-yoke) to deal with software development errors. (It can be used in concert with SPC.) A paper written by an HP systems engineer on this topic is available at:

http://www.campbell.berry.edu/faculty/jgrout/pokasoft.html

While this suggestion is not exactly what you asked for, I hope you'll find it of interest.