Thank you very much for your reply. I ran the (0 0 1) and (0 0 -1) contrasts and the highest values were 0.5 and 0.3 respectively. Yes, I am referring to the contrast (1 -1 0). In this case the highest value is roughly 0.3. Although,when I perform the comparison without take under consideration the aging (1 -1) then I find higher values (0.9-0.93). Is the model that I used wrong?
Cheers,
Ioannis
>Hi,
>I'm not familiar with the conversion that you applied for age, but that's fine. Your first contrast doesn't really make any sense, so presumably you're referring to the >inference for contrast 2 ( 1 -1 0)? It sounds like your effect is described better by age than by group membership. If you run the (0 0 1) and (0 0 -1) contrasts you can >test the effect of your transformed age covariate.
>Cheers,
>Jeanette
>On Sun, Jul 15, 2012 at 2:35 PM, Ioannis Giapitzakis <[log in to unmask]> wrote:
>Thank you very much for your reply. I know that in terms of statistics a p-value of 0.1 is not so useful since a lower p-value is needed. Although, these are some >preliminary results and I am just looking for methods to draw better results, I want also to perform a VBM analysis in order to use it as an additional confound in the >design matrix hoping that this will improve the results.
>Ioannis
>The problem may be your definition of significant results. p=0.1 is not significant, p=0.05 (or less) is.
>Christine
> Date: Sun, 15 Jul 2012 19:52:57 +0100
> From: [log in to unmask]
> Subject: [FSL] Query regarding design matrix -age as an EV
> To: [log in to unmask]
>
> Dear FSLs,
>
> I have a question regarding the design matrix. In particular, I have 15 healthy controls (mean age 31), 8 patients (mean age 61) and I have performed a concat MELODIC on them. Using fslsplit I kept only the DMN component and then I ran the dual regression script. The results were good since there were some significant differences (p=0.1) between these groups. Although, in order to demonstrate that these differences arise from the disease and not from the aging I wanted to use the age as a third EV. For this purpose, I used a quadratic model EV(x)=a*x^2+b*x+c and I defined the 3 parameters a,b,c using 3 conditions-equations: EV(Max age)= 1, EV(Min age)=-1, EV(Median age)=0. Afterwards, I calculated the EV value for each patient and I used it for the design matrix.
>
> design matrix:
>
> /Matrix
> 1.000000e+00 0.000000e+00 -6.800000e-01
> 1.000000e+00 0.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00 -4.000000e-02
> 1.000000e+00 0.000000e+00 -4.200000e-01
> 1.000000e+00 0.000000e+00 -1.300000e-01
> 1.000000e+00 0.000000e+00 -4.000000e-02
> 1.000000e+00 0.000000e+00 -4.700000e-01
> 1.000000e+00 0.000000e+00 -4.200000e-01
> 1.000000e+00 0.000000e+00 -4.700000e-01
> 1.000000e+00 0.000000e+00 -6.800000e-01
> 1.000000e+00 0.000000e+00 -1.000000e+00
> 1.000000e+00 0.000000e+00 -6.800000e-01
> 1.000000e+00 0.000000e+00 -1.700000e-01
> 1.000000e+00 0.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00 -2.100000e-01
> 0.000000e+00 1.000000e+00 6.300000e-01
> 0.000000e+00 1.000000e+00 9.700000e-01
> 0.000000e+00 1.000000e+00 8.200000e-01
> 0.000000e+00 1.000000e+00 8.200000e-01
> 0.000000e+00 1.000000e+00 9.700000e-01
> 0.000000e+00 1.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00 1.000000e+00
> 0.000000e+00 1.000000e+00 6.600000e-01
>
> Contrasts
> /Matrix
> 1.000000e+00 -1.000000e+00 1.000000e+00
> -1.000000e+00 1.000000e+00 0.000000e+00
> 1.000000e+00 0.000000e+00 0.000000e+00
> 0.000000e+00 1.000000e+00 0.000000e+00
>
>
> However, using this method I could not find any significant difference between the two groups. Could you give me any advice or correction for the method that I mentioned above since I do not have any previous experience on that?
> Thank you very much for your time.
>
> Kind Regards,
> Ioannis
|