# Minitab Estimates within-subgroup standard deviation of Normal Capability Analysis

#### philong

##### Registered
@Bev D My company blocks access to Google Docs. Once the OP has 5 posts and can attach their Excel file, I can take a look.
Hi Miner, Sorry but I cannot find where to attach file. Let me see after this fifth post

#### philong

##### Registered
Hi guys, finally I am able to attach the file.
First sheet "Data" include input data and Minitab snapshot for that data.
Second sheet "STD(within)" is my formulas to calculate STD
Third sheet "Unbiasing constants" is my formulas to calculate C4

#### Attachments

• Data HV.xlsx
426.4 KB · Views: 73

#### Semoi

##### Involved In Discussions
The uncertainty of the estimate is given by s/sqrt(2n-2). Thus, the uncertainty of the standard deviation for n=95 data points corresponds to approx 7%. In contrast, the difference between s1=0.0199385 and s2=0.01894227 corresponds to approx. 5%. Therefore, the difference is not negligible, but it is also not critical -- after all, we try to infer the population standard deviation from the sample standard deviation. However, if we forget the subgroups and we calculate the standard deviation of the total dataset, we get SD[total] = 0.01934139. Usually, this is an upper limit of the estimate -- if the data points are uncorrelated. So, it is weird (!) that minitab obtains a larger estimate for the subgrouped dataset. Are you able to reproduce the minitab result?
I'm sorry, I am not able to help you with this one.

#### Miner

##### Forum Moderator
I did not find an error. My results are attached.

#### philong

##### Registered
I did not find an error. My results are attached.

View attachment 30759
View attachment 30760
Hi Miner, thank you very much, this is great to know. I made again my Minitab result and realize that there is difference between 2 methods as below photos.
It seems we should re-arrange data instead of keep as original data.

Method 1: my current method, keep same as original data (5 sub groups) and use Capability Analysis Subgroup

Method 2: your method, re-arrange subgroup into column and use Capability Analysis with Data arrange

#### Attachments

• Minitab_STDwithinCompare.MPJ
73.7 KB · Views: 41

#### Miner

##### Forum Moderator
I use the stacked format to avoid any possible confusion over orientation. While I believe that this is counterintuitive, Minitab expected your data in a different format. It expects each subgroup to be in its own row. When you orient the data as shown, you get the correct result.

#### Johnnymo62

##### Haste Makes Waste
Miner, that's new for me but I'm barely a "user".

For your chart above then it's a subgroup size of 20, right?

I've been using a column of measurement data with a subgroup size. For a size of 5, the first 5 data points are the first subgroup, then the second 5 are another subgroup, and on to the end of the column. Is this incorrect?

#### Miner

##### Forum Moderator
The OP has 5 subgroups with 20 measurements in each subgroup. Their initial post was confusing on this point.

#### Johnnymo62

##### Haste Makes Waste
The OP has 5 subgroups with 20 measurements in each subgroup. Their initial post was confusing on this point.

I see that thanks.