Dear All,
I’m setting up a GLM model for my ICA dual regression analysis and I have two questions about my model.
To summarize the design, I have 4 groups: 1) Patients age1, 2) Controls age1, 3) Patients age2, 4) Controls age2. Patients and controls within an age group are matched for age and gender. There is not the same number of participants in age 1 and age 2 and there is not the same number of male and female in age1. I am interested by the interaction between age 1 and age 2 and by the effect in each age group (patients vs controls).
When I run a GLM with only one age group in the model (Patients age1 vs controls age1 – contrast 1 -1 and -1 1) and an unpaired t-test, I have not the same results than when I run the GLM with the 4 groups in the same model and assess the result of the same exact contrast (Patients age1 vs controls age1: contrast 1 -1 and -1 1). I have read on the forum that when you put all groups together in the same model, other groups contribute to the final result through error term that is pooled across all groups. I am then wondering what is the best solution in my case?
I have a second question regarding demeaning. I have read a lot of comments on the forum about this question but I have not found an answer to my particular question yet. I would like to put gender in nuisance factor in my model. As I have set up 4 different groups, I have defined 4 EV for gender and I have demeaned gender within groups (with males =1 and females = 0). Is that correct or should I rather demean across groups first and separate the 4 EV for gender after?
If I have 3 participants in each group and all groups in the same model, my model is like this:
A B C D E F G H I
1 1 0 0 0 -0.3 0 0 0
1 1 0 0 0 0.7 0 0 0
1 1 0 0 0 0.7 0 0 0
2 0 1 0 0 0 0.5 0 0
2 0 1 0 0 0 -0.5 0 0
2 0 1 0 0 0 0.5 0 0
3 0 0 1 0 0 0 -0.3 0
3 0 0 1 0 0 0 0.7 0
3 0 0 1 0 0 0 0.7 0
4 0 0 0 1 0 0 0 0.5
4 0 0 0 1 0 0 0 -0.5
4 0 0 0 1 0 0 0 0.5
(A=group, B=patients age1, C=patients age2, D=controls age1, E=controls age2, F,G,H,I = the 4 EVs for gender
And my contrasts
A B C D E F G H
1 -1 -1 1 0 0 0 0
-1 1 1 -1 0 0 0 0
1 1 -1 -1 0 0 0 0
-1 -1 1 1 0 0 0 0
1 0 -1 0 0 0 0 0
-1 0 1 0 0 0 0 0
0 1 0 -1 0 0 0 0
0 -1 0 1 0 0 0 0
Many thanks in advance!
Best regards,
Elisa
|