Variability investigation

MikeKim — Fri, 09 Jun 2023 00:50:05 GMT

Dear all

This is Mike.

I am now investigating Variation Component.

It looks like that easiest way to split variation is, using Variability chart.

However, it looks like I cannot split several variation individually.

In other words, I have several variation source of interest.

1. Between day

2. Between operator

3. Between Lot

4. Between sampling point

5. within Lot

6. within repeatability condition (between replicate)

However, if I put [day, operator, lot, point] as Grouping Factor, X, and put [vial] as Part,

JMP gives me a no reasonable answer (R&R= 91%), whereas when I put only [operator] as X, and put [vial] as Part, JMP gives me 35% (which is reasonable).

So... is there any other method to split all of these specific source of variability?

How can I make JMP to understand that

[day, operator] as reproducibility

[lot, point, vial(part)] as part variability

[replicates variability for individual part] as repeatability

Additionally I know the theory of split-plot design, nested design and how to figure out their Var.Comp to define respective Variance(each source).

But, this is only theory and I practically cannot do this by my hand.

I cannot find the proper JMP menu...

If it is there, please let me know.

*Note that individual vial can be tested 3 times by one person, but destructive sothat cannot be tested by the other person. So I am very confused whether I can apply crossed design to this. Personnally, I think I can suppose whole vials can be regarded as ONE Lot since our process is supposed to be similar and comparable to their reportable value.

I appreciate you to read this far.

Re: Variability investigation

Mark_Bailey — Wed, 25 May 2022 12:35:06 GMT

You can use Analyze > Fit Model instead to estimate the variance components. See the Fitting Linear Models guide in JMP Help to understand how to enter the data columns in the launch dialog. The dialog supports nesting (it assumes crossed factor levels). It supports random effects.

topic Variability investigation in Discussions

Variability investigation

Re: Variability investigation