Choose Language Hide Translation Bar
Highlighted

How can I calculate Pooled Variance Estimate in JMP?

Hello JMP Community,

 

How can I calculate Pooled Variance in JMP?

 

I am hoping, as with all things JMP, this exists and I have just not found the magic pathways to produce the desired results quickly and efficiently.  I can of course do this by going through the actual maths myself, but I have frequently run into this situation so it would be great to learn if this already exists in JMP.

 

I have a data set that is grouped by process conditions which have differing means as a result of process centering (tuning) activities.  I would like to generate a pooled variance estimate for this process for use in specifications/capability estimates in support of this new process launch.  My total run count is rather low so for initial process variability/capability discussions I would like to do this to increase my N count.

 

Technical Situation

- Process 1 Condition A, was running at lower end of our desired process window [n=8]

- Process 1 Condition B, changed our tuning lever to center the process (new) to the desired target [n=10]

- The process change resulted in a mean shift, but did not change the process variability (and fundamentally was not expected to)

- I would like to generate the "Pooled Variance" estimate for the P1CA + P1CB

- As this is a new process, my data set is limited, so the desire is to use this to enable a increase n count during initial capability discussions

 

This is the simply case.  I have much more complex cases but they all follow this same scenario where variance is unaffected when a lever was pulled to create a mean shift.

 

Thank You,

WanderedProcess

 

***sorry if a thread exists detailing this.  I did try searching key terms I am familiar with. =-] ***

1 ACCEPTED SOLUTION

Accepted Solutions
Highlighted
cwillden
Super User

Re: How can I calculate Pooled Variance Estimate in JMP?

If you only have 2 groups, can't you just do the One-way Analysis in Fit Y by X (red-arrow > Mean/ANOVA/Pooled t)?  The Mean Square Error (MSE) is the pooled variance estimate.  You can grab that from the ANOVA table or by squaring the RMSE in the summary.

-- Cameron Willden

View solution in original post

2 REPLIES 2
Highlighted
cwillden
Super User

Re: How can I calculate Pooled Variance Estimate in JMP?

If you only have 2 groups, can't you just do the One-way Analysis in Fit Y by X (red-arrow > Mean/ANOVA/Pooled t)?  The Mean Square Error (MSE) is the pooled variance estimate.  You can grab that from the ANOVA table or by squaring the RMSE in the summary.

-- Cameron Willden

View solution in original post

Highlighted

Re: How can I calculate Pooled Variance Estimate in JMP?

Thank You very much. I knew it was hiding in there somewhere!
Article Labels

    There are no labels assigned to this post.