Hi Sam

do the following:

- split per group and zscore cov1 cov2 cov3

- add 0 to cov1A cov2A cov3A cov11B cov2B cov3B so that it has the right length

- now do you 2 groups + 6 covariates

- [1 -1] and [-1 1] for groups as usual

- [0 0 1] or any combo like [0 0 1 1 1 1 1 1] to look at covariates per group or across groups

- [0 0 1 0 0 -1 0 0] would be how covariate differ for group1 > group2 after removing the group difference

--> if you were to look only at interactions, setup a multiple regression with the 6 covariates (this model assumes no group differences on average, only modulations by covariates)

best

cyril

From: SPM (Statistical Parametric Mapping) <[log in to unmask]> on behalf of Sam W. <[log in to unmask]>
Sent: 30 October 2018 18:07
To: [log in to unmask]
Subject: [SPM] interaction group with covariate

Hi,

I have two groups and created a contrast for each group at first level. At the second level I'm using a two sample t-test with three covariates.

My design matrix is therefore

G1 G2 Cov1 Cov2 Cov3

How can I test for an interaction between group and cov1 for example?

SPM accepts the contrast [1 -1 1 0 0] but I'm not sure if this is correct.

Thanks

Sam