I recently attended a lecture by an industrial statistician. Part of the lecture included a summary of Dorian Shainin’s body of work. I had to smile as the lecturer spoke about Shainin’s “exaggerated claims” of the results attributed to his methods and his infamous “pre-control.” This really piqued my interest because I’ve had good results using this technique, although reviews are mixed.
Why haven’t we heard much of Shainin’s methods since the Six Sigma/lean operations rage has taken hold? That’s not easy to pinpoint. One reason may be that users who are aware of his philosophies aren’t willing to accept product that’s “good enough.”
Dr. Shainin came from a different era, before the use of powerful computers. He accomplished much, developed techniques that worked well, and helped users solve difficult problems.
Although a statistician himself, his approach was not focused on applying statistics so much as to solving problems, which appealed to industrial leaders. What came to be known as the Shainin System (SS) was developed for problem-solving in medium to high volume processes where data are readily available, statistical methods are widely used, intervention in the process is difficult, and ‘conformance to specification thinking’ was expected.
A fundamental tenet of SS is that, in any problem, there is at least one dominant cause of variation in the process which leads to that problem. This presumption is based on the application of the Pareto principle (Juran’s vital few, useful many).
Within SS there is a distinct recognition that, in addition to the dominant cause, there is usually a second or third cause that must be dealt with before the problem can be solved. Actually, this is the reason many problems aren’t completely solved. If there is not a single dominant cause, reducing variation becomes more difficult. Multiple significant causes have to be addressed to substantially reduce the overall output variation.
Certainly, there’s more to the SS than the dominant cause theory, but the whole approach relies mainly on simple observational studies and small experiments. Again, the approach is not about getting detailed technical answers, but solving the problem.
The application of Shainin techniques relies heavily on observational studies, which if understood and planned well provide a good, quick approach to solving the problem. The question is whether the SS qualifies as real “design of experiments.”
A hidden benefit of SS just might be that it prevents a lot of impulsive, poorly designed experiments which are wasteful and expensive. One reason why certain problems are unsolved is because people often take ineffective approaches. Their approach is usually based on a set of assumptions, tools and techniques that are less effective for solving tough chronic quality problems.
Not once during the lecture did this ‘expert’ speaker mention the simple technique of plotting the observational data in its naturally occurring time order (e.g. process capability study). If planned correctly, simple plotting allows one to mine a hidden wealth of information. With SS, the user quickly gets to the data, then to the histogram and, finally, to Pareto the data into significant causes so they can be resolved.
As mentioned earlier, there is a high risk that multiple failure modes contribute to a problem. Therefore, the result could likely be confounded with different dominant causes for each failure mode. The Shainin methodology uses a process of elimination, called “progressive search,” to identify the dominant causes. Progressive search uses a strategy much like Kepner-Tregoe or “20 questions” where users attempt to identify the answer using a series of questions that increasingly divide the search space into smaller and smaller regions.
As a quality engineer, I was working with a team to reduce leaks in sealing surfaces of engine cylinder blocks. The team had been working hard for some time but had made little progress until we realized there were three categories of leaks, defined by location. When the team considered leaks at each location as separate problems, a dominant cause and related remedy was identified for each location. Hence, a long-standing issue was resolved.
My experience is that the cause of a chronic problem is often related to something that no one thought about in the beginning; therefore, the obvious continues to be overlooked. In the current environment, the tendency is to focus on the complexity and not simplicity, The Shainin System, however, focuses on simplicity. We may be statistically more sophisticated, but is that what we really seek? I think not. As Leonardo DaVinci said, “Simplicity is the ultimate sophistication.”