Skip to content
Prev 307417 / 398506 Next

car::linearHypothesis Sum of Sqaures Error?

Dear John

On Tue, 9 Oct 2012 02:07:07 +0000
"John Jay Wiley Jr." <jwileyjr at syr.edu> wrote:
For type III tests, you should use contrasts that are orthogonal in the row basis of the design. Perhaps you've done that (by setting the contrasts for the factors directly), but I suspect not. Why not just use type II tests? They're hard to screw up. 

As well, I assume that the variables that enter additively are the covariates. If not, and a covariate is involved in the interaction, the type III tests aren't sensible unless the 0 point of the covariate is where you want to test a "main effect" or lower-order interaction.
Only if the data are balanced.

I hope this helps,
 John

------------------------------------------------
John Fox
Sen. William McMaster Prof. of Social Statistics
Department of Sociology
McMaster University
Hamilton, Ontario, Canada
http://socserv.mcmaster.ca/jfox/