BookmarkSubscribeSubscribe to RSS Feed

How to test for non-additivity in a 3 factors RCBD ANOVA?

Juno

Community Trekker

Joined:

Jan 4, 2017

As mentioned in the title. The three factors are entry, location, and year.

Don't know how to test for non-additivity in JMP? Thank you!

 

1 ACCEPTED SOLUTION

Accepted Solutions
markbailey

Staff

Joined:

Jun 23, 2011

Solution

Yes.

Learn it once, use it forever!
8 REPLIES
markbailey

Staff

Joined:

Jun 23, 2011

What do you mean by 'non-additivity?'

I suspect that it means the presences of effects in the response that are not only the additive main effects. If that guess is correct, then include cross-terms to test for interactions.

Learn it once, use it forever!
Juno

Community Trekker

Joined:

Jan 4, 2017

I am trying to see of the 3 factors are additive in the model. 

 

Do you  mean through running a full factorial ANOVA including 2way and 3way interactions, if any of the interactions terms are significant, the additivity of the three factors are not accepted ? 

markbailey

Staff

Joined:

Jun 23, 2011

Solution

Yes.

Learn it once, use it forever!
Juno

Community Trekker

Joined:

Jan 4, 2017

Thank you very much!

bordini

Occasional Contributor

Joined:

Feb 21, 2018

Please, would you be able to give me the steps to test for nonadditivity in JMP and how to interpret the results? I would appreciate if you could help me. I have data from a two-factorial experiment.

 

PS: I'm a beginner in Statistics and in JMP.

 

Thank you very much,

 

Isadora

cwillden

Community Trekker

Joined:

May 1, 2017

If your experiment has 2 factors (say X1 and X2), just add higher-order model terms such as X1*X2, X1*X1, and X2*X2 to your model as @markbailey suggested (I'm assuming you're using Fit Model for your analysis).  If the p-values for any of those effects are significant, then the relationship is not additive.  Do you need help adding higher order terms to your model?

-- Cameron Willden
bordini

Occasional Contributor

Joined:

Feb 21, 2018

Hi,

 

Thank you for your reply and explanation. And yes, I need help adding higher order terms to my model, please.

 

Thank you very much!

cwillden

Community Trekker

Joined:

May 1, 2017

In the Fit Model dialog window, you can add higher-order terms by doing the following: 

  • Adding interactions manually: Select a factor already in the model ("Construct Model Effects" box) and select a column in the list of columns in the left-hand pane ("Select Columns" box).  Press the "Cross" button.  You can cross a column with itself to make quadratic terms.AddInteraction.PNG
  •  Construct specific types of models using the Macros button: You can fully specify a specific type of model using the Macros button in the dialog.  Just highlight all the columns containing factors for your model and click the Macros button.  Hover over the options to see what each does.  Response Surface will add all 2-factor interactions and quadratic terms.  Full factorial will add all interactions up to the highest degree possible.  Factorial to Degree will add all interactions up to the specified degree (numeric input just below the Macros button).  The default setting for degree is 2, so you would get all 2-factor interactions.

-- Cameron Willden