 |
 |
Hi Eugene, Below you will find an outline of our communication from last week. Thanks so much for your time. One quick question, what did you mean by the “group level?” Is that the first EV that is labeled “group”? Where do I differentiate between MRI 1 and 2? Thanks Hi Fadel - OK, again, I think the paired t-test is the appropriate model, taking in the first-level feats. You can add additional regressors at the group level to account for variations in anxiety. Cheers, Eugene -- Eugene Duff, Phd Analysis Group, Centre for Functional MRI of the Brain (FMRIB) Nuffield Department of Clinical Neurosciences John Radcliffe Hospital University of Oxford, OX3 9DU Ph: +44 (0) 1865 222 523 -- On 19 May 2011 18:09, Fadel Zeidan <[log in to unmask]> wrote: Sorry for not being clearing. I am comparing meditation to simply attending to the breath across sessions. “The average effect” is the group mean for the difference between attending to the breath and rest at MRI 1 and meditating and rest at MRI 2. So, my “positive contrast” will show me greater activation for meditation as compared to attending to the breath and my “negative contrast” will show me greater activation for attending to the breath as compared to meditation. We also ran some a covariate of interest. Specifically, change in state anxiety corresponding to attending to the breath in MRI 1 and meditation in MRI 2. They came out beautifully when I just look at each session by itself. However, we want to directly compare meditation to attending to the breath and take into consideration the variance due to changes in anxiety. If you are in the States, my telephone number is 704 578 1271. Thanks again. This problem has been my life for the past two weeks. On 5/19/11 1:02 PM, "Eugene Duff" <[log in to unmask]> wrote: Hi Faldel - The Paired T-test seems the appropriate model. I'm not sure what you mean by " I have an average effect with two contrasts and an EV for 15 subjects." what are your two contrasts? Eugene On 19 May 2011 17:56, Fadel Zeidan <[log in to unmask]> wrote: Thanks again. OK. I just want to make triple sure. In MRI 1 Subjects either attended to the breath or were at rest. I have an average effect with two contrasts and an EV for 15 subjects. Intervention= meditation training In MRI 2 (after training) Subjects either meditated to breath or were at rest. I have an average effect with two contrasts and an EV for 15 subjects. I then took the two group “copes” from each MRI session and compared with two contrasts and 1 EV. I can’t do this? If not, how would you got about comparing these sessions? THANK YOU SO MUCH FOR YOUR TIME. Sincerely, Fadel On 5/19/11 12:51 PM, "Eugene Duff" <[log in to unmask]> wrote: Hi Fadel - So you mean a model with just the two group means as inputs, modelling the differences? Or just subtracting the copes? No these aren't appropriate approaches. Cheers, Eugene On 19 May 2011 17:28, Fadel Zeidan <[log in to unmask]> wrote: Hi. Thanks for your response. My main question is, can I compare to higher level analyses in a within subjects fashion. I have a group average for Time 1 before an intervention and another for after the intervention. I want compare them directly and I have by simply comparing their respective copes in FEAT. Can this be justified without being too liberal in my stats? Thanks so much! On 5/19/11 12:10 PM, "Eugene Duff" <[log in to unmask]> wrote: Hi Fadel - Are you saying your a comparing paired and unpaired t-tests? I suspect there is something wrong with the model if the paired t-test images are coming out speckled. Eugene On 19 May 2011 16:16, Fadel <[log in to unmask]> wrote: Simply put: Can I compare to third level analyses in a within subjects fashion. There were two conditions in MRI session 1. Subjects performed a task (manipulation) and did not (rest). I have the group effect for these contrasts corresponding to MRI 1. Subjects then participated in an intervention. There were two conditions in MRI session 2 (after the intervention). Subjects performed a task (manipulation) and did not (rest). I have the group effect for these contrasts corresponding to MRI 2. Am I being too liberal? Am I treating within subject variability as between subject variability? The analysis comes out beautifully. I also ran a paired-t test with the same data, and the images come out speckled. I've checked for outliers and have reran the analyses dozens of times to no avail. Thanks.