Dear Robert,
If you want to include Age and Sex as regressors in order to eliminate their
potential influence on your correlation (of FA and behaviour) then you just
need to insert them as additional, demeaned EVs, as you proposed.
The contrast for the FA - behavioural regression (correlation) does not change.
The statistics may or may not change much depending on how much effect
age and sex have on the data and how correlated these are to the behavioural
scores. The GLM will use the extra EVs to determine what is uniquely attributable
to behaviour in the FA data, making the results of your contrast such that you can
be sure that it was not driven by any age- or sex-related changes.
All the best,
Mark
On 4 May 2012, at 00:22, Robert wrote:
> Dear FSLers,
>
> I have a small questions reg. TBSS analysis. Considering previous discussions, I performed a simple linear correlation analysis with TBSS, creating GLM with the subjects (1 group) and 1 EV (=behavioural measures, demeaned), setting two contrasts (+1 = positive correlation, -1=neg. correlation).
>
> Now I would like to include Age and Sex as additional regressors? Do I just have to add EVs for these variables? What will the results look like? What contrasts whould I need if I would be still interesting in the correlation between FA (TBSS) and behaviour. How will the covariates influence my TBSS results. How will statistics consider these additional information?
>
> Thank you very much!
>
> Best, Robert
>
|