Dear Experts,
In case if the control chart showed that the process is stable (no special cause variations), but the average baseline is unacceptable (slight high), the question is, can we see that the process is in-control?
Thanks.

Yes,
the process is in-control.
Your question actually is about (i believe) one of the greatest misconceptions about control charts, that is:
within stastical control means process is capable?

The answer is no. Process being acceptable actually means two things:
1. Process is within control
2. Process is capable

Your question also outlines the natuaral flow of process improvement
1. Make the process within control
2. Make the process nominal to be acceptable
3. Back to #1

>>rey

Dear Experts,
In case if the control chart showed that the process is stable (no special cause variations), but the average baseline is unacceptable (slight high), the question is, can we see that the process is in-control?
Thanks.

Reynald is correct. It's possible for a process to be in a state of statistical control (subject only to random variation) and produce nothing but out-of-tolerance output. What your situation is telling you is that the process itself must be changed in order to provide the nominal baseline you require. If your numbers are accurate, the statistical analysis has done its job--it's told you that your process is functioning "normally," but won't provide the output you expect.

Hi guys,

Another thing I want to add is we must evaluate the R chart first as it must be in control to interpret the bar X chart. It is because the bar X control limits are calculated considering both process variation center.

If R chart in not in control, bar X chart is inaccurate.The lack of the control will due to unstable variation rather than the actual changes in process center.

To study the process capability, the data of the sample must in the normal distribution and there are come from a process control.:biglaugh:

Assuming all the calculations and interepretations of the original chart were correct, then I'd answer the original question as "Yes". You can certainly be stable and "in control" but have a process that is not capable of meeting the aim of the corporation. And the data do not have to be normal to determine that a process is not capable. Dr. Deming's Red Bead Experiment is a perfect example of this. The data are not "normal" (they are counts) but the results are stable and predictable on a np or c chart. The key is that if you want to cause improvement to the results, you can't just work on the most recent results. You must work on the overall system, using all of the historical data, and determine how to "remove the red beads" from the system.

Dear Experts,
Thanks for the valuable information.
Kindly I have two cases related to this issue and I will be highly appreciated if you send me your feedback.
1.According to one of our policy the medical assessment should be done within the first 24 hours , the auditing process for the previous month have showed that the average baselines is 04:02 hours (no special cause variation) , my question is , can we see that the process is capable?
2.The second case is about the needle stick incidents, we are using c chart to monitor the number of incidents on a monthly basis, the process is stable but the problem is we don’t have a target number of incidents, so in this case how can we know if the process is capable?
Many thanks in advance.

Dear Experts,
1.According to one of our policy the medical assessment should be done within the first 24 hours , the auditing process for the previous month have showed that the average baselines is 04:02 hours (no special cause variation) , my question is , can we see that the process is capable?

Sometimes the average can be misleading, which is why variation around it is important. If you have enough data, you should be able to predict with reasonable assurance the likelihood of the 24-hour mark being violated, and then address the process if it's necessary. You say there's no special-cause variation, but it's still possible that the process won't perform as required (although since the mean seems to be comfortably distant from the upper limit, and the process is in control, it's probably unlikely). The question you need to ask yourself is, "Is there anything in the data that makes me believe that the process needs to be adjusted?" Given the probable number of variables involved, it would be difficult for anyone here to give you definitive advice.


The second case is about the needle stick incidents, we are using c chart to monitor the number of incidents on a monthly basis, the process is stable but the problem is we don’t have a target number of incidents, so in this case how can we know if the process is capable?


At first blush, there's an urge to say that the target number should be zero, but I'm not a fan of the zero defects approach when it's known that defects are inevitable. What number of fatal car accidents is "acceptable"? That's a loaded question, somewhat like the classic "Have you stopped beating your wife" query. The intuitive answer is "none," but we still know that fatal car accidents are inevitable, so it makes no sense to have zero as a goal (imo).
I think that maybe you've skipped a step, and need to do a little PDCA, rather than worrying about a target at this point. Here are some questions you might like to ask:

How many needle sticks are happening now, over some period of time x?
Why are they happening? Is there a training issue? Is there any pattern? For example, do needle sticks happen more on a particular shift? Are they more likely to be perpetrated by certain people, or in certain situations (or some combination of the two)?
Outside of needlessly inflicting pain, what are the risks? We know that transmission of pathogens is a concern, for example.There are other potential questions, but you get the idea. Once you know something about the problem, you will be in a better position to develop methods of mitigating it. The reason you are collecting data should be to give you knowledge of how, when and why the problem occurs, and what might be acceptable means of dealing with it. Setting a target at this point probably won't be helpful, unless the target is a meaningful and pragmatic reduction.

Sometimes the average can be misleading, which is why variation around it is important.

Exactly. As an analogy with my height i can easily walk on a river bed with waters 5 feet deep. Given the data that a certain river has the depth of 4 feet on the average is not enough for me to say if im capable to walk its depth. The river depth could be ranging from 1feet to 7feet! (in that case the average would be most likely 4feet). That is why variation, not only the average is important.

Dear Experts,
Thanks for the valuable information.
Kindly I have two cases related to this issue and I will be highly appreciated if you send me your feedback.
1.According to one of our policy the medical assessment should be done within the first 24 hours , the auditing process for the previous month have showed that the average baselines is 04:02 hours (no special cause variation) , my question is , can we see that the process is capable?
2.The second case is about the needle stick incidents, we are using c chart to monitor the number of incidents on a monthly basis, the process is stable but the problem is we don’t have a target number of incidents, so in this case how can we know if the process is capable?
Many thanks in advance.

Number 1 has already been answered (must know the variability as well as the average).

For number 2, it seems like zero is the correct goal. Needle sticks can at least generate significant emotional impact, if not physical. But - as with anything that is stable, you need to ask yourself are you happy with this current level? Could you continue to operate indefinitely at that level and not have any significant negative results? If the answer is - I need to improve - then the goal becomes to generate an improvement per whatever control chart trending rules you are using (such as 7 in a row below average). Once you have that improvement, and you stabilize, then you ask yourself if you need another improvement.

As a Deming person, I am against numerical goal setting. I'd much rather see a group use the logic of do I need to improve or not.