P-Chart with variable sample size - How to update the chart?

[pagnonig]

Maybe it is a trivial question, but I need some help...

I've plotted a p-chart with 20 points (last 20 months). The process is predictable. Each point refers to a sample size different from the others. Therefore the UCL and LCL for each point are calculated using the follow stdev with sample size n:

(pavg x (1-pavg) / n)


Now, I've to add one more point to my chart (data from a new month). Which option should I choose:

1. plot the value on the chart. Do not change the average baseline computed on the previous 20 points (pavg_20). Plot the UCL and LCL using the pavg_20 value in the stdev formula. Check if there are out of control signals.

2. plot the value on the chart. change the average baseline including all 21 points (pavg_21). Plot the UCL and LCL using the pavg_21 value in the stdev formula. Check if there are out of control signals.

I would use option 1. Your comments are highly appreciated. Ciao.
Giuseppe

 

[Coury Ferguson]

SPC experts: Any feedback?

 

[Miner]

Use Option 1. The only time that you would use Option 2 is when you know that a process change has occurred. The new p-bar should be calculated from the point that the process change occurred.

 

[Bev D]

agreed. option 1 is the correct first choice. as Miner says the average line is only changed / recalculated when a known permanent change has occurred. and the new average is calculated only usign the points that represent the new stable level.

 

[Steve Prevette]

Davis Ballestracci provided a good phrase to me during one of his classes:

"A baseline is innocent until proven guilty"

 

[luckyboyvn85]

Hi Miner if Min/ Max <= 75% UCL and LCL can't use option 1. Can you help me for this case? Thank

Use Option 1. The only time that you would use Option 2 is when you know that a process change has occurred. The new p-bar should be calculated from the point that the process change occurred.

 

[Bev D]

I’m sorry can you explain your situation in more detail? I’ve never heard of the situation you refer to...the only time you change the control limits is when there is a known improvement that was made to change the process....