How to fix the lower control limits to 0 for time based processes

[#1]

I'm creating a control chart for a time based process and I get a negative control limit.  Since this is a time variable with a minimum value of 0 then showing negative time is not appealing. How do I fix the LCL to 0 and not show negative control limits?

[#2 · Accepted solution]

You can set the LCL for your column by setting the Control Limit Column Property for that column to 0.  To do this, click on the the column header for the column of interest. Rt. mouse click and select Column Properties.  From the displayed list, select Control Limits.  When the Column Properties window pops up, select the control limit type that you want to set the LCL for......XBAR, Individual Measurement...etc.  Then just enter in the LCL value of 0.  You can set more than one type of control limit if you choose.

Exit out of the Column Properties window and then run your control chart.  The control chart platform will use the LCL from the column property.

[#3]

When I set the LCL to 0, the UCL gets recalculated. For the particular data set that I'm working with, the range between the UCL and LCL actually decreases, implying that there's more to it than shifting the UCL and LCL by the same value. What's happening on the back end to generate the new UCL?

[#4]

Another approach:  Double click on the Y-axis.  Set the minimum to 0.  Then add a reference line at 0 with the color and/or thickness of the line as you desire.  In these cases where negative numbers are impractical, you really don't have a lower control limit.  You have a lower boundary.

[#5]

Another thought:  Sometimes it is better to construct a Run Chart about the population median ( not the mean), especially when the data is bounded by an upper or lower value.  If the data is centered close to either boundary, the data is often skewed heavily and may take on the characteristics of a Poisson or Logarithmic distribution.  When using the median, then extreme values don't have undue influence.  I have used this thought process in analyzing safety data quite successfully.  I published a piece on this topic on QualityDigest.com.

[#6]

Would you mind posting the reference for article ? (I tried searching for "smoore2" but didn't find much)

[#7]

http://www.qualitydigest.com/inside/quality-insider-article/alleviating-tortured-data.html