Dear Experts,
Daily we are monitoring 6 medical records to see if the medical assessment is documented or not, yes/no data.
And we are monitoring the process by using the np chart; my question is in relation to how to interpret the results.
For example for a given month the chart gave us a UCL=2.78, CL=0.59, and a LCL=0.00.
All the figures are not integer numbers, and we are dealing records.
So how can I interpret the above figures, can I see the maximum expected number of incomplete records is 2.78 records?
Many thanks in advance.

Dear Experts,
Daily we are monitoring 6 medical records to see if the medical assessment is documented or not, yes/no data.
And we are monitoring the process by using the np chart; my question is in relation to how to interpret the results.
For example for a given month the chart gave us a UCL=2.78, CL=0.59, and a LCL=0.00.
All the figures are not integer numbers, and we are dealing records.
So how can I interpret the above figures, can I see the maximum expected number of incomplete records is 2.78 records?
Many thanks in advance.

IF your system is stable (I don't know how much data you used to make your baseline) and nothing perturbs the system in the future, then, yes, the 2.78 represents a maximum expected. In a given day, I assume the possible results are 0, 1, 2, 3, 4, 5, or 6 records not documented. From how you describe your data, on a typical day I would expect to get a 0, 1, or 2; but not a 3 or above.

Personally, I prefer straight p-charts as being easier to interpret, and you are not limited to a fixed sample size. I suspect your results may have more meaning if you rolled up the results on a weekly basis, the sample size of six is a little small for a p-chart (or np chart).

IF your system is stable (I don't know how much data you used to make your baseline) and nothing perturbs the system in the future, then, yes, the 2.78 represents a maximum expected. In a given day, I assume the possible results are 0, 1, 2, 3, 4, 5, or 6 records not documented. From how you describe your data, on a typical day I would expect to get a 0, 1, or 2; but not a 3 or above.

Personally, I prefer straight p-charts as being easier to interpret, and you are not limited to a fixed sample size. I suspect your results may have more meaning if you rolled up the results on a weekly basis, the sample size of six is a little small for a p-chart (or np chart).

Dear Mr. Steve,
Thanks for the information.
The number of points is 22.
Yes the possible results are 0, 1, 2, 3, 4, 5, or 6 records and usually we have between 0 and 3 defective records per day.
Regarding the use of p-chart instead of np, in our case the daily sample size is constant (6 records per day).
Kindly, what about the minimum sample size to use p or np?
Many thanks.

Kindly, what about the minimum sample size to use p or np?
Many thanks.

I'd suggest at least 20. That would give 21 possible results rather than only 7. Assuming you are sampling 5 days a week, I'd suggest trying weekly updates (sample size of 30) and see how that looks.

A few posts about Overall Equipment Efficiency (OEE) were moved here:
http://elsmar.com/Forums/showthread.php?t=21771

Tim Folkerts

I'd suggest at least 20. That would give 21 possible results rather than only 7. Assuming you are sampling 5 days a week, I'd suggest trying weekly updates (sample size of 30) and see how that looks.

Dear Mr. Steve,
I didn’t understand what you mean of 21 possible results rather than only 7?
Thanks.

Dear Mr. Steve,
I didn’t understand what you mean of 21 possible results rather than only 7?
Thanks.

With the original sample size of 6, the possible results are {0,1,2,3,4,5,6}, which is 7 different values.

With an increased sample size of 20, the possible values are {0,1 ... 20}, which is 21 different values. This gives more resolution in the results.

The larger sample sample size is better at detecting smaller shifts. The downside is that you might have to wait 3 times as long to detect a major sift (since you would only be updating the chart about 1/3 as often.


Tim F