Interaction Profiles - automated display of panels based on factor inclusion

[#1]

My question is how JMP controls the display of panels of Interaction Profiles plots that correspond to factors not included in the associated model.

Does JMP consistently display such panels differently from panels corresponding to factors that are included?

I would like to distinguish factors included and omitted (within the model) on JMP Interaction Profiles. In my experience this operation is done automatically in some cases but not others.

In the first screenshot(1-No_distinction) I've  manually indicated the terms included in the model with asterisks (*). 

JMP did not distinguish the display of the panels based on inclusion of factors.

In a different JMP workbook the plots corresponding to factors not included in the model are dashed lines and appear with dimmed intensity (2-With_distinction).

[#2]

1-No_distinction2-With_distinction

embedding screenshots instead of attaching

[#3]

I may not understand your situation, but JMP will only give plots for the terms in the model.  If terms are confounded (aliased), then you will get dashed lines to indicate this.

[#4]

Thank you. In the top example, the three plots without asterisks are for interaction terms not in the model. Yet, they are not displayed in dashed lines. Could  you please elaborate on the aliasing of terms?

[#5]

Can you post your data table (anonymize it).  You should not be getting any plots for terms not in the model.

If, for example, you run a fractional factorial design and do the analysis with a saturated model, the interaction plots will show dashes indicating there are aliased terms.  Here are the interaction plots from a 2^4-1 res IV experiment.  Since 2nd order effects are aliased, you see the dashed lines for the aliased interactions.

 

[#6]

The screenshot is from the top row (pertaining to factor "A") of the interaction profiler for the attached workbook – accessed via the embedded script Interaction Profiles. The script contains the model.

Only A*B and A*F are interaction terms included in the model for factor A. Yet, the interaction term plots are not greyed out or dashed for A*C and A*D nor A* E, as I have discussed in the preceding posts.

[#7]

If you look at the Model it appears all 2nd order factor interactions are in your model.  Hence you will see the interactions.

If I modify your model with just AB and AF, this is what you get:

 

[#8]

Thank you! in response - can you please address why A*C-E do not appear in the Effect Summary? 

What I have intended in the model derivation/analysis was to include these terms initially, but then they were eliminated based on significance/contribution.

In other  work, eliminated interaction terms not appearing in Effect Summary  are grayed out/dashed in Interaction Profiles.

[#9]

Well if I wager a guess it's because you removed them.  Here is the effect summary with your model (except I removed the D*D since it was aliased) before any terms are removed:

I'm not sure what exactly you are doing, but I removed AE from the Effects Summary and here is the Interaction Profile.  Notice the AE interaction is greyed out.

 What version JMP are you running? 

 

[#10]

OK -- the graying-out circumstance dependent on the inclusion status of interaction term is behaving as expected when I regenerate the model with all terms and then individually remove the undesired terms. 

But, in the initial script I shared, the display properties on the interaction plots are still unresponsive to the removal of terms via the Effect Summary -- i.e., the inconsistency persists between Effect Summary and Interaction Profiler plotting properties in the initial script I shared. I don’t believe we need a solution for this anomaly.

I am running JMP 17.2. It is possible that the initial script I shared and am referring to was generated with JMP 16.2.