A lot of work has been done with Shewhart’s Control Charts, extending it well beyond where it was meant to go. There are some high-toned, scientific and hyper-academic arguments over control charts out there, in part due to the deeply flawed theory of six sigma:
See here: http://demingcollaboration.com/whydislikename6sigma
A theory abandoned by its earliest adopters, Motorola.
See here: http://www.q-skills.com/Deming6sigma.htm
Truth is, life varies, and is very complex. When academic pursuits enter the realm of pure math they risk losing touch with the inherent reality of things. That is why Shewhart, and Deming after him, saw control charts as a simple tool, that was useful, in particular in a company using the Deming cycle, which is not PDCA but PDSA, where the S means study the results, a world away from just checking them.
See here: http://demingcollaboration.com/?page_id=1408
Rather than try to understand the nature of reality in a spreadsheet, think of the control chart as like a thermometer, like the one we use to take our temperature. Most are not exactly accurate, but for the most part, if you’ve got a fever it will tell you. Of course, that does nothing to tell you the disease, or how to treat it. Results can’t be managed directly, only their causes. Most of the time, if we have a fever we stop what we are doing and call the doctor; if we don’t have a fever, we just keep going. Great Very similar to a control chart, a tool that is effective in helping operators, those doing the work, to have some kind of guideline as to when what they were seeing was common or special cause variation, when its time to do nothing or its time to stop and call a doctor. How foolish would you look if you used a supercomputer to take your temperature?
All analyses are wrong, some are useful. Not everything that is true is useful, and not everything that is useful is true. When you drop a pan on your foot, remembering Newton the moment beforehand would be useful, even if his theories are not true.
As the Buddha said, don’t seek after entanglements as if they are real things. And remember, the true purpose of a control chart is to help you understand a process, not eliminate defects down to a percentage. In a system, nothing works in isolation but to the detriment of the whole. Optimize parts, go out of business. Great teams may have great players, but a team of great players will never be great, unless they work as a team, allowing themselves to run less than optimally when it makes sense for the good of the team. (This is, I think, a John Wooden quote.)
And before anyone joins the chorus declaring Shewhart and Deming passé, read them, with an open mind, you will be pleasantly surprised.
Links:
Great Book on the Subject by Dr. Wheeler, explains variation and the easy and proper use of XmR chart for counts: http://www.spcpress.com/book_understanding_variation.php
More Control Charts
A lot of work has been done with Shewhart’s Control Charts, extending it well beyond where it was meant to go. There are some high-toned, scientific and hyper-academic arguments over control charts out there, in part due to the deeply flawed theory of six sigma:
See here: http://demingcollaboration.com/whydislikename6sigma
A theory abandoned by its earliest adopters, Motorola.
See here: http://www.q-skills.com/Deming6sigma.htm
Truth is, life varies, and is very complex. When academic pursuits enter the realm of pure math they risk losing touch with the inherent reality of things. That is why Shewhart, and Deming after him, saw control charts as a simple tool, that was useful, in particular in a company using the Deming cycle, which is not PDCA but PDSA, where the S means study the results, a world away from just checking them.
See here: http://demingcollaboration.com/?page_id=1408
Rather than try to understand the nature of reality in a spreadsheet, think of the control chart as like a thermometer, like the one we use to take our temperature. Most are not exactly accurate, but for the most part, if you’ve got a fever it will tell you. Of course, that does nothing to tell you the disease, or how to treat it. Results can’t be managed directly, only their causes. Most of the time, if we have a fever we stop what we are doing and call the doctor; if we don’t have a fever, we just keep going. Great Very similar to a control chart, a tool that is effective in helping operators, those doing the work, to have some kind of guideline as to when what they were seeing was common or special cause variation, when its time to do nothing or its time to stop and call a doctor. How foolish would you look if you used a supercomputer to take your temperature?
All analyses are wrong, some are useful. Not everything that is true is useful, and not everything that is useful is true. When you drop a pan on your foot, remembering Newton the moment beforehand would be useful, even if his theories are not true.
As the Buddha said, don’t seek after entanglements as if they are real things. And remember, the true purpose of a control chart is to help you understand a process, not eliminate defects down to a percentage. In a system, nothing works in isolation but to the detriment of the whole. Optimize parts, go out of business. Great teams may have great players, but a team of great players will never be great, unless they work as a team, allowing themselves to run less than optimally when it makes sense for the good of the team. (This is, I think, a John Wooden quote.)
And before anyone joins the chorus declaring Shewhart and Deming passé, read them, with an open mind, you will be pleasantly surprised.
Links:
Great Book on the Subject by Dr. Wheeler, explains variation and the easy and proper use of XmR chart for counts: http://www.spcpress.com/book_understanding_variation.php