Hello CONN experts,
Is CONN suitable for a 2x2 within-subjects / repeated measures ANOVA interaction with covariate control (ANCOVA). From the toolbox manual, it was not clear whether CONN could handle a repeated measures ANCOVA. I read that including categorical covariates in an ANCOVA might be problematic.
I have combined 51 subjects with 4 scans each from two studies and wish to look at contrast and interaction effects.
Subject effects:
AllSubjects
Age: continuous
BMI: continuous
Sex: 0/1
Study#: 0/1
Effect of all subjects [1 0 0 0 0]
Sessions:
drug-condition1
drug-condition2
placebo- condition 1
Placebo- condition 2
Drug-condition interaction: User-defined contrast: [1 -1 -1
1]
Main drug effect: [1 1 -1 1]
Is my setup correct, and can CONN control for covariates in a repeated measures ANOVA (including categorical covariates)?
Kind regards,
Peter
Hi Peter,
Yes, that looks perfectly fine, the only recommendation here would be to make sure all of your control covariates are centered (otherwise the 'AllSubjects' effect will be estimated at the zero-level of each control covariate, which may not be all that meaningful depending on each covariate specific encoding).
Best
Alfonso
Originally posted by Peter Skov-Andersen:
Hello CONN experts,
Is CONN suitable for a 2x2 within-subjects / repeated measures ANOVA interaction with covariate control (ANCOVA). From the toolbox manual, it was not clear whether CONN could handle a repeated measures ANCOVA. I read that including categorical covariates in an ANCOVA might be problematic.
I have combined 51 subjects with 4 scans each from two studies and wish to look at contrast and interaction effects.
Subject effects:
AllSubjects
Age: continuous
BMI: continuous
Sex: 0/1
Study#: 0/1
Effect of all subjects [1 0 0 0 0]
Sessions:
drug-condition1
drug-condition2
placebo- condition 1
Placebo- condition 2
Drug-condition interaction: User-defined contrast: [1 -1 -1 1]
Main drug effect: [1 1 -1 1]
Is my setup correct, and can CONN control for covariates in a repeated measures ANOVA (including categorical covariates)?
Kind regards,
Peter