Hi all,
I conduct, with SPM5, an analysis of MRI structural data in one group of subjects. I wanted to correlate behavioural score (BS) with Gray matter volumes. My statistical model was Multiple Regression = 1 covariate of interest (BS), 3 covariates of non interest(sex, age, TIV) = 1 or -1, 0, 0, 0.
I posted few weeks ago a question about Preprocessing. With a reply of the mailing list, i had interesting results. Now, i want to test gender difference in the correlation between BS and regional gray matter volume.
I have a question about statistical model, i have to use 2-sample t-test OR Multiple regression ?
1)2-sample t: 2 condition (Male, Female), 1 covariate of interest (BS, interaction with Factor 1, centering Factor 1 mean), 2 covariates of non interest = 6 columns : Male, Female, BS@sF2Male, BS@sF2Female, age, TIV.
t-contrast: 0, 0, 1, -1, 0, 0 AND 0, 0, -1, 1, 0, 0.
Do you confirm: Statistical Model (2-sample t) ? Design Matrix with 6 columns ? These t-contrasts ?
What should be the signification of contrasts: 1, -1, 1, -1, 0, 0 AND -1, 1, -1, 1, 0, 0 ?
2)Multiple Regression with the same model i explained at first, BUT Modifying Covariate sex with Interactions with Facor 1, and centering Factor 1. I will have 4 columns in Design Matrix= sex@sF2, age, TIV, BS.
So, i don't understand signification of sex@sF2 covariate and contrast (1, 0, 0, 0).
Out of curiosity, why the regressor name's ...F2.. when interaction with Factor 1 ?
Thanks a lot in advance !
Maxime freton
____________________________________________________
Toute l’actu Sport est sur http://sports.voila.fr
|