3 factor ANOVA with a Covariate

LX

Hi,

 

I am trying to run a 3 way ANOVA with a covariate, and I'm not sure how the data should be organized and how to run the test (my "full factorial" click doesn't respond either).

 

So far, I have one column for each factor and the covariate.

 

Thank you,

 

LX

gzmorgan0

LX,

From your description, I think your table is set up correctly.  A column for each factor a column for the covariate and a column for each response. To get JMP to create the Full Factorial model, you need to select the three factor columns then from the Macro menu select Full Factorial, then add your covariate.

 

When you have a Covariate, you should check if there is an interaction with the covariate and at least any main effect. The attached jmp table is an example with both the no interaction and interaction models.

 

I don't know if you are aware of Box-Cox transformations. They are specified as 0. If you want to check, change the argument of Box Cox Y Transformation ( 0 ) from 0 to 1.  [If you know this analysis, lambda is near 1, which means a power transformation is not needed].

 

Fit Model(
	Y( :Post ),
	Effects( :X1, :X2, :X1 * :X2, :X3, :X1 * :X3, :X2 * :X3, :X1 * :X2 * :X3, :Pre ),
	Personality( "Standard Least Squares" ),
	Emphasis( "Effect Leverage" ),
	Run(
		:Post << {Summary of Fit( 1 ), Analysis of Variance( 1 ),
		Parameter Estimates( 1 ), Plot Actual by Predicted( 1 ),
		Plot Regression( 0 ), Plot Residual by Predicted( 1 ),
		Plot Studentized Residuals( 0 ), Plot Effect Leverage( 1 ),
		Box Cox Y Transformation( 0 )}
	)
)

Hope that helps.

threefactorAncova.jmp