 |
 |
On Wed, Oct 21, 2015 at 6:44 PM, fMRI <[log in to unmask]> wrote: > Dear Donald > > Regarding the main effect or the average of a condition say condition A: > > Say if I have a ROI 1 as my seed: > > When performing a positive t tests (ie versus rest, ROI 2 is significant. > This means that ROI 1 and ROI 2 has a significant connectivity? > >> It means that you have a significant change in connectivity compared to the implicit baseline. > If this is correct, would it possible to comment about strong and weak > connectivity based on the higher t values? > >> No. I would use the con_ images. The t-statistics are the con_ images divided by the between-subject variance. > > My question is if I use a negative t tests, what does not mean if I get > ROI 3 being significant? > It means that the connectivity decreased relative to the implicit baseline. > > Regards, > > Aser > > On 20 Oct 2015, at 19:15, "MCLAREN, Donald" <[log in to unmask]> > wrote: > > See below. > > > On Tue, Oct 20, 2015 at 1:52 PM, Aser A <[log in to unmask]> wrote: > >> Thank you Donald.. >> >> In the second point, I meant the PM modulation (negative linear PPI). >> Does this mean that as the PM increases the connectivity decreases ? >> > > Yes. > > >> >> For the two group question, if there are no differences using two sample >> t tests but there are differences using one sample t tests (for example >> visually or when looking at common and specific areas), would this be still >> reportable and acceptable ? >> > > No. You can't conclude that the two groups are different unless you > statistically test that they are different. This is true of any analysis. > See: > Nieuwenhuis, S., Forstmann, B. U., & Wagenmakers, E.-J. (2011). Erroneous > analyses of interactions in neuroscience: a problem of significance. *Nature > Publishing Group*, *14*(9), 1105–1109. http://doi.org/10.1038/nn.2886 > >> >> Aser >> >> On Tue, Oct 20, 2015 at 6:22 PM, MCLAREN, Donald < >> [log in to unmask]> wrote: >> >>> See below. >>> >>> On Tue, Oct 20, 2015 at 1:11 PM, Aser A <[log in to unmask]> wrote: >>> >>>> Dear Donald and all, >>>> >>>> I have two PPI questions: >>>> >>>> 1- if I have a PM modulation (e.g. linear changes) and done +1 on the >>>> subjects levels and then did group analysis using +1 or -1 ? is this >>>> correct (i.e. do I have to return to the subjects level again and do the >>>> contrast -1 in order to perform one sample t tests group analysis ? >>>> >>> >>> >> Correct. Both directions at the group level can be tested with the >>> single contrast from the first level. >>> >>>> >>>> Now the PPI related question here is that what does it mean a negative >>>> first order linear PPI analysis between ROI (A (seed)) and ROI (B) ? >>>> >>> >>> >> Do you mean for a first level contrast of -1 over one PPI column? If >>> so, this would mean that the connectivity amplitude is less than during >>> baseline. >>> >>> >>>> >>>> 2- The second question is related to the group analysis. If I have two >>>> groups patients (A) and healthy (B), when performing two sample t tests to >>>> investigate PPI (A) > PPI (B) I do not get any significant even at very low >>>> threshdol. Is it always difficult to get differences between group and this >>>> need very high number of samples ? >>>> >>> >>> >> It depends on the effect size. Some effects will be greater than >>> others. Without knowing anything about the task or how the task was >>> modeled, its hard to say if you'd need a large number of subjects or not. >>> We are working on spatial analysis approaches that would be less dependent >>> on the actual effect size and more dependent on the spatial distribution of >>> the effects. >>> >>> >>>> >>>> Thanks >>>> >>>> Aser >>>> >>> >>> >> >