Minitab: Normal Plot of Standardized Effects - Interpretation

[VisibleConfusion]

"Calling all Six Sigma Black Belts!"

Hi everyone,

I am trying to evaluate testing results using Minitab. I would like to use the graphical method "Normal Plot of Standardized Effects" to vizualize how the test result is affected by a number of factors. Here's an example of what this plot looks like:

https://support.minitab.com/en-us/minitab/18/normal_efects_plot_def.png

What I need now is how to interpret the graph, more specifically: the vertical coordinate of the dots. I know that the line through the plot is a cumulated standard normal distribution. The horizontal coordinate corresponds to each factor's effect which is calculated from the results. If effects lie on the line, that means they are likely to be coincidental and to follow a standard normal distribution. If their horizontal distance to the line is above a certain threshold (p-value greater than significance level), they are considered non-coincidental.
I think I understood all this pretty well. What I can't get my mind around is what the vertical 'percentage' coordinate means in regard to the Experiment factors. It is calculated via the Benard rank, which in turn depends on the effects' t-values. As I understand, the t-value also states by how much a observed result differs from a standard normal distribution.

Because I have no background in statistics whatsoever, I tried to understand how the plot is generated from start to finish. I contacted Minitab's support for this, and while they were being very helpful, they could only tell me how the graph mathematically comes together. I also asked People in my company, but they don't seem to use Minitab all that much, and also not this kind of graph. The Minitab guide book by Quentin Brook also does not go into detail on this topic.

 

[Hansac]

I don't quite understand the plot you shared but test results, by statistics, are "fit" into equations, either linear, quadratic, or more complex equations, to check for predictability.

This equation is then drawn through the test data results (or samples).

This is called statistical regression and with that you can check whether your R-square and this R-square will tell you whether your test results is reproducible ie valid.

Statistics is fun and an easier branch of mathematics. Unlike algebra.