Print

Print


Dear Pari,

Your first way of labeling is correct. With the second you will only label the first 3 trials. DCM results depend on the data so if you use different data you'll get a different answer, especially if you are using one trial per condition compared to 20. Remember that this is all about statistical estimation of some quantities rather than the absolute truth.

Best,

Vladimir

On Mon, Oct 26, 2015 at 1:29 PM, Pari Eghbali <[log in to unmask]> wrote:
Dear SPM experts,
I have three different kinds of datasets recorded from 64 channels with 199 datapoints. I want to repeat each of dataset for 20 times (overall 60) and concatenate them to have 60 trials like:
Data (timepoint, channels, 1:3:60) = data1
Data (timepoint, channel, 2:3:60) = data2
Data (timepoint, channels, 3:3:60) = data3
 I am going to estimate connectivity modulation by DCM between these three datasets.
1)      If I define condition as like as:
D = conditions(D, 1:3:Ntrials, 'Condition 1')
   D = conditions(D, 2:3:Ntrials, 'Condition 2')
   D = conditions(D, 3:3:Ntrials, 'Condition 3')

I obtain different result with another definition of condition like :
D = conditions(D, 1, 'Condition 1')
   D = conditions(D, 2, 'Condition 2')
   D = conditions(D, 3, 'Condition 3')


Why DCM gives differen B matrix for the two labelling? I suppose to have same results from those two labelling since if DCM estimate connectivity by averaging of same trials, averaging of 20 same signals is as same as one signal.

Thank you for any response,
Pari