How to display the value of alpha (statistical significance) that is used in the Fit Model graphs? Is it possible with JSL JMP scripting?

[#1]

Graphs in the Fit Model platform show confidence intervals.  For instance, "Actual by Predicted Plot" and "Prediction Profiler".  How can I add to the display to show the value of significance or confidence that that interval represents.  For instance, if I changed the alpha to 0.1 I'd like to see "alpha = 0.1" somewhere in the platform.  Then I can compare the graph to one where the alpha is 0.15 and I'll know which is which.

I suppose I could just add some static text, and that's a last resort.  I'd rather it be automatic so I can look at any graph that has confidence intervals and know the alpha.

If I add a script button to the report, is it possible to access the value of alpha that is being used?  I could use the gray triangle to "Append Item > Add Script Button", but then what is the code in the script to access the value of alpha?  What is the context for the script code from which I could find the avialable variables to access?

By the way, the same applies to the color coding for pvalue.  I'd like an automatic way to know the alpha that is in the conditional formatting, without having to go to the dialogs for re-defining the conditional formatting.

[#2]

Hi @MiBo ,

  I'm not sure I can help about adding alpha to a graph, but yes, as last resort, you could always enter it manually.

  On the other hand, you can add as many conditional formatting color levels as you see fit for the p-value. It's a little deep in the preferences, but if you go:

File > Preference > Reports, next to Show conditional formatting there's a Manage Rules button. Click that and select PValue, then Edit. This will take you to a window that you can change the default and then add in your own upper and lower limits for conditional formatting. I've done this for my p-values adding levels for 0.1 and 0.15. This will then be the conditional formatting for all of JMP's platforms.

Hope this helps!,

DS

[#3]

@SDF1,

I like the idea of coloring a variety of levels of p value, that's a good suggestion.

I've done this in the past with a heat map of p-values. The goal was to have a quick report visual to attract the eye to significant factors. I chose a red-yellow-green scheme, where green means significant, red means not significant, yellow means maybe. What colors scheme do you use?

Mike

[#4]

Hi @MiBo ,

  For the conditional formatting, I use the default for alpha <0.01, which I believe is orange and then I use a couple different shades of red for when the p-value increases to about 0.15, and then it is the regular black if p > 0.15. I do that because some of the studies and evaluations we do need to be a bit more "lenient" -- they can have a slightly higher false positive rate than JMP's default of 0.05. It helps to see when some factors are still significant for our work, even when their p-values might be around 0.1, which would otherwise be black in JMP's default settings.

  For heatmaps, I'd imagine either the  "chromatic" (which you described as using) or "diverging" color theme might be best to be a good visual for p-values. The "sequential" might not really show the differences between significant and non-significant as easily.

Good luck!,

DS