Hi Rosalia - 

You shouldn't have +1 in the mean effect columns (1st and 2nd columns) for copes showing +ve or -correlations - you need to remove these.  Otherwise, it looks OK if the 5th column is simply an unrelated covariate of some sort.

Best

Eugene 

On 13 February 2017 at 13:48, Rosalia Dacosta Aguayo <[log in to unmask]> wrote:

---------- Forwarded message ----------
From: Rosalia Dacosta Aguayo <[log in to unmask]>
Date: 2017-02-13 9:36 GMT+01:00
Subject: Rifght model for correlations
To: FSL - FMRIB's Software Library <[log in to unmask]>


​Dear FSL experts,

I am trying to do separate correlations (+ or -) between two group of participants (HC = controls) and (PAT = patients) and a variable BDNF after covariating for X (COV). I would like to ensure if this model is correct.

Please, I would like more than grateful if any one of you coud give a look into it just to be sure all it is ok.

With my kind regards,

Rosalia​