Hi all I'm hoping someone can help me with a problem I noticed when using the Control Chart template from the CD we received as part of doing the PuMP workshop. I work in education and am using Control Charts to monitor attendance of students. I found when I put data into the template for a school and allowed the automatic formulas to calculate the lcl and ucl, they were different than when I worked it out manually using the formulas in the PuMP workbook. Now I'm no statistician or anything close to it – so I know I've done something wrong somewhere either with the template or with calculating manually but I just can't work it out. The template gives me ucl & lcl's with a much greater variation than my manual calculations. When I look at the attendance data line on the template it tells me the attendance is within normal variation for the school. If I look at it in the manually calculated way, it tells me there are several points where attendance is outside the range of normal variability for the school. Obviously this makes a big difference to our analysis discussions & decisions around action.

If anyone could please help me with working out what I've done that would be just so appreciated!! I understand Stacey is away on leave until August – so am hoping for some collegiate advice before then.

Thanks anyone who reads this request!!

Cheers

Bec

Thanks for
sending through your control charts, Bec. I can see straight away what went
wrong…

The charts
you created using the spreadsheet is actually an old spreadsheet which I don’t
use anymore, since I learned about XmR charts (the chart type that the PMBW
workbook calculations are based on). Thanks to Donald Wheeler and his book, Understanding Variation.

With XmR
charts (one reason why they’re more appropriate than the charts using standard
deviation to calculate control limits) you can use as few as 6 or 7
points in your time series and still draw some useful conclusions. And as you noted, the control limits using the previous method are much wider, and that method actually overestimates the variation. The XmR chart is the perfect chart for KPIs of all kinds.

You can download the right XmR Excel template here.