Home > Statistics Every Writer Should Know > The Stats Board > Discusssion
SPC Charts vs. Rolling Average Charts
Message posted by EC on November 18, 2000 at 12:00 AM (ET)
I plotted some safety data (injuries per million manhours per month) on a rolling average chart, and the chart showed a dramatic performance improvement over nine months. I charted the same data on a 3 sigma statistical process control chart that showed no change in system performance. Why?
READERS RESPOND:
(In chronological order. Most recent at the bottom.)
Re: SPC Charts vs. Rolling Average Charts
Message posted by JG on November 19, 2000 at 12:00 AM (ET)
How is the mean and standard deviation for you control chart calculated. Are they updated ? How many data points does your moving average use ?
Re: SPC Charts vs. Rolling Average Charts
Message posted by Tommy VH on November 21, 2000 at 12:00 AM (ET)
SPC charts have specific formulas that determine the control limits. For the averages chart, the mean of the ranges is multipled by factor A2 to establish the UCL and LCL. Two major conditions must be considered: 1) factor A2 is dependent upon the subgroup size. The smaller the subgroup, the wider the control limits on the averages chart. 2) what was the rational basis for subgrouping? The range chart represents the within subgroup variation and the averages represents the between subgroup variation. If you are looking far a significant change between subgroups, how the subgrouping was handled may well be the aanswer to your question.
Re: SPC Charts vs. Rolling Average Charts
Message posted by EC on November 21, 2000 at 12:00 AM (ET)
The rolling average chart had 27 data points representing 27 months. There was a sharp downward trend the last nine months which I expected due to a successful improvement project in progress.In the initial SPC analysis, I used the 27 points to calculate the control limits.
The upper and lower control limits were calculated by adding and subracting 2.66 times the average of the moving range of the 27 points.
I now see that by doing the SPC analysis on the final nine data points (my subgroup), some of the earlier data points fall outside the uppper and lower control limits indicating a change did occur in the process; this corresponds to the change shown on the moving average chart.
Thanks.
Your $5 contribution helps cover part the $500 annual cost of keeping this site online.
