The interaction term should literally be the product of the two main effect terms.Here is what you likely want based on my understanding of what you are trying to do:
EV1: column of 1'sEV2: MMSE (demeaned)EV3: HR response (demeaned)EV4: age (demeaned)EV5: genderEV6: product of EV2 and EV3
C: 0 0 0 0 0 1will test whether there is a positive interactionC: 0 0 0 0 0 -1will test whether there is a negative interaction
cheers,-MH
--Michael Harms, Ph.D.-----------------------------------------------------------Conte Center for the Neuroscience of Mental DisordersWashington University School of MedicineDepartment of Psychiatry, Box 8134660 South Euclid Ave. Tel: 314-747-6173
From: Paul Beach <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Tuesday, September 16, 2014 11:27 AM
Michael,
I understand that I want an MMSE*HR term, and looking at some information from that google and others let me see if my thoughts are on the right track. I might be way off, though.
Since I want to keep things in a within-subjects context I could create a 5th EV that is +1 when both MMSE and HR are positive (i.e. both are greater than the mean) and -1 when both are negative. For cases when there is a discrepancy (i.e. MMSE +; HR - or vice-versa) I would leave things as a zero for the 5th EV.
I'm not sure how this would translate to the contrast setup.
Am I way off track?
On Tue, Sep 16, 2014 at 11:39 AM, Harms, Michael <[log in to unmask]> wrote:
Hi,Google "interaction continuous variables" and post-back if it isn't clear.Make sure you demean MMSE, age, and HR response.
cheers,-MH
--Michael Harms, Ph.D.-----------------------------------------------------------Conte Center for the Neuroscience of Mental DisordersWashington University School of MedicineDepartment of Psychiatry, Box 8134660 South Euclid Ave. Tel: 314-747-6173
From: Paul Beach <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Tuesday, September 16, 2014 10:35 AM
To: FSL - FMRIB's Software Library <[log in to unmask]>
Subject: Re: [FSL] within-subjects GLM
Michael,
Thanks for your response. This is what I was suspecting, but I didn't want to separate my AD corhort into a specific set of groups. How could I create an interaction term without doing that?
On Tue, Sep 16, 2014 at 11:31 AM, Harms, Michael <[log in to unmask]> wrote:
No, you would need a model with an interaction term to get at that question.
cheers,-MH
--Michael Harms, Ph.D.-----------------------------------------------------------Conte Center for the Neuroscience of Mental DisordersWashington University School of MedicineDepartment of Psychiatry, Box 8134660 South Euclid Ave. Tel: 314-747-6173
From: Paul Beach <[log in to unmask]>
Reply-To: FSL - FMRIB's Software Library <[log in to unmask]>
Date: Tuesday, September 16, 2014 9:32 AM
To: FSL - FMRIB's Software Library <[log in to unmask]>
Subject: [FSL] within-subjects GLM
FSL folks,
I'm interested in performing a within-subjects analysis in an AD cohort looking at the relationship between cognitive status and a behavioral score on resting connectivity while adjusting for age and gender.
The general question I'm asking is: as Alzheimer's disease worsens (i.e. MMSE scores decrease) what are the resting neural correlates of changes in autonomic response to a pain stimulus?
I have a GLM setup and wanted to confirm the appropriate contrasts.
EVs are as follows:1 - MMSE2 - Age3 - Gender4 - Heart rate response
My current contrast setup:C1 PosMMSE_effect 1.0 0 00C2 NegMMSE_effect -1.0 0 00C3 PosMMSE_effect_hr 1.0 0 01.0C4 NegMMSE_effect_hr -1.0 001.0
Will the latter two contrasts get at the question I'm asking?
Thanks--
Paul Beach
DO/PhD candidate - Year VI
Michigan State University
- College of Osteopathic Medicine
- Neuroscience Program- MSU Cognitive and Geriatric Neurology Team (CoGeNT)
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.
--
Paul Beach
DO/PhD candidate - Year VI
Michigan State University
- College of Osteopathic Medicine
- Neuroscience Program- MSU Cognitive and Geriatric Neurology Team (CoGeNT)
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.
--
Paul Beach
DO/PhD candidate - Year VI
Michigan State University
- College of Osteopathic Medicine
- Neuroscience Program- MSU Cognitive and Geriatric Neurology Team (CoGeNT)
The materials in this message are private and may contain Protected Healthcare Information or other information of a sensitive nature. If you are not the intended recipient, be advised that any unauthorized use, disclosure, copying or the taking of any action in reliance on the contents of this information is strictly prohibited. If you have received this email in error, please immediately notify the sender via telephone or return mail.