RE: Orthogonalising the covariates

[color=#000000]Dear Ramesh,[/color]

[color=#000000]There is no need to orthogonalize your covariates, as GLM can handle colinearity among predictors without any issues. Generally, if your design matrix includes effects A and B, and you imagine a Venn diagram representing the unique and shared variance explained by these two factors, then, orthogonalizing B with respect to A allows you to measure (using a standard "effect of A" contrast) the combination of unique and shared variance explained by A (while typically, without orthogonalization, the same "effect of A" contrast would only measure the unique variance explained by A)[/color]

Hope this helps
[color=#000000]Alfonso[/color]
[i]Originally posted by Ramesh Babu MG:[/i][quote]Dear experts, 


I have 3 groups and 3 nuisance covariates. Should I orthogonalise the nuisance covariates?
To orthogonalise, first I selected the nuisance covariates and clicked orthogonalise the selected covariates. Then a pop window appeared and asking to select orthogonalising factor. Here I am getting confusion. Which factor should I select? How to proceed further? Should I run the set up again? 

Please give your suggestions? 
Thanks
Ramesh[/quote]