topic How do I find the "between run standard deviation percent of total standard deviation" in JMP? in Discussions

How do I find the "between run standard deviation percent of total standard deviation" in JMP?

SerialDesign687 — Thu, 07 Aug 2025 18:46:01 GMT

I am trying to find the Between-run standard deviation percent of total standartd deviation on some data in JMP. I have an old example of the output I am looking for from previous data analysis that was completed in minitab (see attached screenshot). How do I get the same type of results in JMP?

Re: How do I find the "between run standard deviation percent of total standard deviation" in JMP?

jthi — Thu, 07 Aug 2025 19:08:37 GMT

You are maybe looking for Oneway analysis in JMP Basic Analysis > Oneway Analysis > Oneway Analysis Reports > Analysis of Means Reports

Or Fit Model Fitting Linear Models > Standard Least Squares Models > Regression Reports > Analysis of Variance

or maybe Quality and Process Methods > Variability Gauge Charts > Variability Gauge Analysis Report Options > Variance Components

Re: How do I find the "between run standard deviation percent of total standard deviation" in JMP?

SerialDesign687 — Thu, 07 Aug 2025 19:38:30 GMT

These all get close to what I am looking for but I still cant seem to find the standard deviation percent of total. I can only seem to find either the percentage of the total variation each component accounts for or the standard error of each variance component. I am looking for the standard deviation % of total of the run vs the error.

Re: How do I find the "between run standard deviation percent of total standard deviation" in JMP?

yusuke_ono — Thu, 07 Aug 2025 22:33:04 GMT

Hello, I am Yusuke Ono, Senior Tester at JMP Japan.

It seems that your model is a variance component model (or a random effect model). You can fit a (one-way) variance component model on JMP by the following steps.

(1) You need to prepare two variables in the input data.
The one is the continuous response variable which contain measurments. The other is the nominal group variable, which is "Run" in the example. Note that you need to set the "Run" column to Nominal.

(2) Select [Analyze] > [Fit Model]. The Fit Model window is launched.
(3) Set the response variable (or the measurment variable) to [Y].
(4) Select "Run" variable, and click [Add] button in the Construct Model Effects panel.
Select the "Run" variable in the Construct Model Effects, and click the red triangle icon for the [Attributes] and select [Random Effect].

(5) Click [Run] button.
(6) Go to the REML Variance Component Estimates table. Rihgt-click somewhere in the table,and select [Columns] > [Sqrt Variance Component]. This shows standard deviations of variance components.

Note that these variance estimates are estimated by REML (restricted maximum likelihood, residual maximum likelihood). If your data is not balanced, these results are different from EMS(Expected Means Squares, or moment method). The difference between REML and EMS is summarized in the following document page.
https://www.jmp.com/support/help/en/18.2/index.shtml#page/jmp/mixed-models-and-random-effect-models.shtml

If you want to do MSA (measurment system analysis), JMP has a specific platform for MSA.
The Measrument System Analysis platform can be invoked by [Analyze] > [Quality and Process] > [Measrument System Analysis].