# How to determine what should be the sub-group size

Six Sigma – iSixSigma Forums Old Forums General How to determine what should be the sub-group size

Viewing 14 posts - 1 through 14 (of 14 total)
• Author
Posts
• #39483

Raju
Participant

Hi everyone:
Could somebody some light on how to determine the sub-group
size for the control charts, say for example, Xbar, R charts,
why sub-group size cannot be more than 12,

Thanks

0
#120197

Mikel
Member

Who says the subgroup size can’t be more than 12?

0
#120199

Issa
Participant

For X-bar and R control charts , I expect the sub groups to be at least 20.
Where did you get your info from?

0
#120200

Dayton
Member

Here is a pretty good rule of thumb:
Variable Charts

Chart Type
Minimum Subgroup Size

Maximum Subgroup Size
LCL
UCL
LSL
USL

X-bar Chart
2
30 (if the limits are to be based on the ranges)Infinite (if /stdev option is used)
Negative Infinite
Positive Infinite
Negative Infinite
Positive Infinite

R-Chart
2
30
0
Positive Infinite
0
Positive Infinite

s Chart
2
Positive Infinite
0
Positive Infinite
0
Positive Infinite

Individual Chart
2
30
Negative Infinite
Positive Infinite
Negative Infinite
Positive Infinite

Median Chart
2
25
Negative Infinite
Positive Infinite
Negative Infinite
Positive Infinite

Moving Average Chart
2
30
Negative Infinite
Positive Infinite
Negative Infinite
Positive Infinite

Moving Range Chart
2
30
0
Positive Infinite
0
Positive Infinite

Cumulative Sum Chart
1
Depends on number of cases of variable. (This chart requires at least two subgroups.)
N/A
N/A
N/A
N/A

Process Capability Chart
1
<=30 unless sigma is based on the subgroup Standard Deviations
Negative Infinite
Positive Infinite
Negative Infinite
Positive Infinite

Process Capability Analysis
N/A
N/A
N/A
N/A
< Mean within group variation
Depends on ‘Tail Fraction of data to use’

Group Measures Chart
N/A
N/A
N/A
N/A
N/A
N/A

0
#120201

Mikel
Member

Do you know what a subgroup is?

0
#120203

Ken Feldman
Participant

Typically subgroup sizes are in the range of 4-5.  Shewhart used 4 for his research.  It is usually best to use more frequent small subgroups until you understand the process.  For subgroups above 8-10 you probably want to shift to an Xbar/S chart.  How frequently the process generates the data might help you decide on subgroup size.

0
#120204

Mikel
Member

My process generates products, not data. What is my subgroup for this unique situation?

0
#120209

Ken Feldman
Participant

Ah, a thought provoking question.  I had assumed that the poster was inferring that there was a product/service/output that had some critical measurement characteristic that they were interested in monitoring.  It is the values associated with that ctq that might be subgrouped.
I hope your ears weren’t burning for the past few hours.  I just had dinner with a disciple of yours and your name came up now and then…..always in a good way of course.  You have apparently not been in touch for a while.  Came from the old days at a former employer of yours and later in your consulting company.  Guess he also worked with Mrs. Stan.  Initials are RT if that helps.

0
#120215

Mikel
Member

Be careful what you believe from those accountant types.
Can h still whistle tunes from he Wizard of Oz?

0
#120248

Ken Feldman
Participant

That be he.   Didn’t get him drunk enough to do the whistle thing but I will get that next time we visit.  Seems like a good guy and he spoke highly of you so I question his judgement of character.

0
#120249

Mikel
Member

He doesn’t whistle when he is drunk, he does it in the presence of stupidity (think “if I only had a brain” from the Wizard of Oz). This means you must have been on your good behavior.
He speaks highly of both me and Carnell – his judgement really suxxs and there is your data to support it.

0
#120252

Ken Feldman
Participant

And he speaks highly of Mrs. Stan.  Both her and Carnell I can understand but YOU…….I don’t know.  He did say he wasn’t crazy about Greg, so maybe there is hope after all.

0
#120491

Tired of you
Member

Don’t you guys have e-mail?  Why do you insist on diluting the discussiono with your inane blather about the boffo butter buds you drink with and their inane and arrogant insults of others.
Gawd, Jim Conklin!  That you would to specter be and leave inquiry at peace in your absence.
“Jee, Stan.  Have you met any heads of state this week?  Yeah, me too, but they were stupid.  HAHAHA.”
TOY

0
#120503

Mikel
Member

Email?
Bothered by something posted a week ago? Don’t have much to do?

0
Viewing 14 posts - 1 through 14 (of 14 total)

The forum ‘General’ is closed to new topics and replies.