August 25, 2017 at 9:13 am #55804
MPBatistaParticipant@MBatista Include @MBatista in your post and this person will
be notified via email.
We know that there is a rule to check stability. We can notice if there is any special cause in SPC. Here we already have many rules to define special cause. However it’s a poor definition for me, mainly when we are talking about nonnormal distributions. Stability means that the process is predictable. Could we think that stability is a process whose distribution type and distribution parameters are constant?
In fact, I don’t use a lot of SPC but I see if there is any strange thing happenning in my process to understand if it’s stable or not.
Could you express your knowledge about it and criticize?
Thanks!0August 25, 2017 at 11:31 am #201784
Chris SeiderParticipant@cseider Include @cseider in your post and this person will
be notified via email.September 2, 2017 at 6:09 am #201805
Mike BonniceParticipant@mbonnice Include @mbonnice in your post and this person will
be notified via email.
A stable process is in control on the range chart. An unstable process is out of control on the range chart. It is possible for a stable process to be either in control or out of control on the X or Xbar chart. It is possible for an unstable process to be either in control or out of control on the X or Xbar chart. Stability means the range is acting now as it had in the past. In control means range and X are acting now as they had in the past (within limits).
A special cause of variation may show its impact on either of the charts (or not at all).
When Shewhart invented control charts, he demonstrated that their utility is not sensitive to the type of distribution. It was re-proven in Wheeler’s book on Statistical Process Control.
The purpose of control charts is to avoid over-reacting to variation. Shewhart set the control limits at three standard deviations because (among other reasons) the likelihood of mis-diagnosing out of control conditions is roughly the same for any distribution.
To investigate whether your process is unstable or out of control, spend time at the process making observations and mapping the sources of variation. Ask new questions, plot the data in new and different ways (including control charts with natural and transformed data), interview the operators.0
You must be logged in to reply to this topic.