Hi,
It is possible for this to happen, as combining measures together (such as RD and AD to form MD) can often reduce the noise and hence give you a boost in statistical power. To check to see if things really do make sense then look at your unthresholded statistics and see if the RD and AD statistics (in the areas where MD is significant) are just below threshold or whether they are far away from it. I would expect at least one of RD or AD to be near to the threshold, in which case this would all make sense. If neither are near threshold then something may have gone wrong somewhere.
I hope this helps.
All the best,
Mark
> On 26 Jul 2017, at 20:57, Walid Yassin <[log in to unmask]> wrote:
>
> Hello,
>
>
> I performed TBSS on my data and i am a little confused by the results.
> I checked the tfce_corrp files.
> FA was not significant and MD was between the 2 groups. This is okay.
> Then i did more analysis and found that AD and RD are both not significant.
> How can Md be significantly different between groups, yet its components are not?
>
> Is this logical results? if not, what might i have done wrong?
>
> I performed all the tests per the manual. Any help?
>
> Thank you in advance,
>
> Wald
|