EMP III Donald J. Wheeler Differences in behavior chart

[New to statistics]

Hello lads.

Photo from the book

I-MR from minitab with Median Moving Range

Why the MR is different?
I have found in other cases of the book also!! What's the cause?

Image from the Book: EMP III Evaluating the measurement process & using imperfect data by Donald J. Wheeler
Thank you

 

[Miner]

If you exclude the out-of-control moving range values, you will get an MR_Bar = 26.7 or 27. When calculating the average range, first calculate the limits with all the values, then exclude any values that are out-of-control and recalculate the limits. Otherwise, all of your control limits will be inflated.

 

[New to statistics]

If i exclude all out of control (o.o.f) values i get this graph

Now i have different UCL & LCL/averages in both diagrams.

Questions:
I remove all the o.o.f from moving range and i leave the I-chart as it is?

But if i remove points out of moving range cc the control limits will change too right (as above)?

 

[brionhurley]

Hello lads.

Photo from the book
View attachment 31322
I-MR from minitab with Median Moving Range
View attachment 31321

Why the MR is different?
I have found in other cases of the book also!! What's the cause?

Image from the Book: EMP III Evaluating the measurement process & using imperfect data by Donald J. Wheeler
Thank you

Wheeler is using the median MR, Minitab uses the average MR.

Wheeler might be using the median to avoid having to remove the outliers.

The MR is used to determine the control limits, so it does make a big difference when the MR chart shows outliers.

 

[New to statistics]

Wheeler is using the median MR, Minitab uses the average MR.

Wheeler might be using the median to avoid having to remove the outliers.

The MR is used to determine the control limits, so it does make a big difference when the MR chart shows outliers.

I use Median Also, the only difference is MR (wheelers=27 & mine=31.9)


See the graphs with average and median

Now i tried the appoach of @Miner to exlude the values out of limits but that changed all limits

Can the computation of the book is wrong maybe?

 

[Bev D]

First let me say that not everyone here are ‘lads’. (I even use the last known picture of me in a dress to show that I am female)
I can’t speak to the math or the purpose of this specific chart set as I no longer have my copy of EMP.

If you exclude MR points from the MR central line calculation then you must also exclude the related Individual points from the average Individual value as they are directly mathematically related. They should still be charted.

The limits won’t be ‘too tight’. They will be ‘true’; it is your process that is not stable and must be fixed. There is no mathematical trick that will fix the real life stability of any process…

If your process is still out of control after excluding obvious Out Of Control points then your process is just not stable. That is the entire point of control charts…you either have an unstable process or you have irrational subgrouing. It’s not about the math - it’s about the insight you gain regarding the process…

 

[Tidge]

First let me say that not everyone here are ‘lads’. (I even use the last known picture of me in a dress to show that I am female)

? LAD =?= 'Last Authenticated Dress' ?

 

[Bev D]

Well alrighty then I’ll go with that…

 

[ChrisM]

(I even use the last known picture of me in a dress to show that I am female)

Unfortunately these days, that proves nothing......