As recently discussed on the list, it's better to do statistics on the log so you can report the mean of the A and B matrices directly and do stats on it. No need to exponentiate. Best, Vladimir On Tue, Jun 28, 2011 at 2:43 PM, Pierre Larigneux <[log in to unmask]> wrote: > Dear Vladimir, > thank you very much once again for your rapidity. > Happily, I asked you! > Just another question : > I have 15 subjects. to calculate the mean matrix of connectivity should I do > the mean of the exponentiel of the connectivity for each subject or should I > take the exponential of the mean of the connectivity ? > ie mean(exp(B(i)) or exp(mean(B(i)) ? > for the statistic on the coefficient, I have done it on the connectivity ( > in fact their logarithm) should I do it on their exponential? > thank you for your precious help. > Gaetan >> Date: Tue, 28 Jun 2011 14:04:52 +0100 >> From: [log in to unmask] >> Subject: Re: [SPM] DCM parameters >> To: [log in to unmask] >> >> Dear Gaetan, >> >> It will be exp(A+B) >> >> Best, >> >> Vladimir >> >> On Tue, Jun 28, 2011 at 1:06 PM, Gaetan Yvert <[log in to unmask]> wrote: >> > Dear SPM users, >> > >> > I have done a DCM modeling with two conditions. >> > For the between trial effect, I put 0 to condition 1 and 1 to condition >> > 2. >> > >> > I have a doubt. >> > >> > Is the connectivity matrix for the condition 1 is exp(DCM.Ep.A) and for >> > condition 2 exp(DCM.Ep.A)+exp(DCM.Ep.B) ? >> > >> > Thank you >> > >> > Gaetan >> > >