Dear Martin,
the suggestion of Cyril to rather work with differences will be ease your analyses. You can use the attached function to calculate differences between surfaces that will be added to CAT12 in newer releases. Please note that the difference between the 2nd and 1st image or surface will be calculated. Thus you obtain for an increase from TP1 to TP2 also positive differences.
Best,
Christian
On Wed, 3 May 2017 15:04:03 -0700, Martin Juneja <[log in to unmask]> wrote:
>Hi everyone,
>
>I have a question for stat experts.
>
>I calculated cortical thickness for a region X for a group of patients
>(Group A: At baseline and after treatment A). Then I calculated cortical
>thickness for same region X for a group of patients (Group B: At baseline
>and after treatment B).
>
>I have behavioral data: At baseline and after placebo treatment A and
> At baseline and after treatment B.
>
>I found that:
>*For Behavioral Data:*
>- Paired t-test between baseline and after treatment A is not significant
>(magnitude after treatment > baseline).
>- Paired t-test between baseline and after treatment B is significant
>(magnitude after treatment > baseline).
>- Using repeated measures ANOVA, I found that the interaction between Time
>(baseline/post-treatment) and Group (A/B) is not significant.
>
>*For Cortical Thickness:*
>- Paired t-test between baseline and after treatment A is not significant
>(magnitude after treatment < baseline).
>- Paired t-test between baseline and after treatment B is also not
>significant (but magnitude after treatment > baseline).
>- Using repeated measures ANOVA, I found that the interaction between Time
>(baseline/post-treatment) and Group (A/B) is significant.
>
>Lastly, correlation between changes in behavioral data and changes in
>cortical thickness is not significant for group A but is significant and
>positive for group B.
>
>I was wondering how can I interpret these results because in one case I am
>having significant interaction (for cortical thickness) between groups and
>in other case there is no significant interaction (for behavioral data) but
>paired t-test showed significant difference. In addition, there was a
>correlation between changes in behavioral data and changes in cortical
>thickness for group B.
>
>So I am not sure how can I proceed from here onwards and justify my
>findings that interaction is important for cortical thickness but not for
>behavioral data and paired t-test is important for behavioral data but not
>for cortical thickness.
>
>Any help would be really appreciated.
>
>Thanks.
>
|