It depends on your question:
(1) Do you want to know how the group means compare? If so, then
average within each group and use separate regressors for each group.
(2) Do you want to know how the group means would compare if the
performance was the same? This is the covariate adjusted means of each
group. Demean all the covariates at once. Use a regressor for each
group and test if you have an interaction. If you have an interaction,
then you shouldn't interpret the adjusted means because they are based
on the performance of the current set of subjects.
Hope this helps.
Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Research Fellow, Department of Neurology, Massachusetts General Hospital and
Harvard Medical School
Postdoctoral Research Fellow, GRECC, Bedford VA
Website: http://www.martinos.org/~mclaren
Office: (773) 406-2464
=====================
This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED
HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is
intended only for the use of the individual or entity named above. If the
reader of the e-mail is not the intended recipient or the employee or agent
responsible for delivering it to the intended recipient, you are hereby
notified that you are in possession of confidential and privileged
information. Any unauthorized use, disclosure, copying or the taking of any
action in reliance on the contents of this information is strictly
prohibited and may be unlawful. If you have received this e-mail
unintentionally, please immediately notify the sender via telephone at (773)
406-2464 or email.
On Tue, Jul 24, 2012 at 6:04 AM, Giulia Preti
<[log in to unmask]> wrote:
>
> Hi Donald and all,
>
> thank you very much for your answer.
> The website you suggested was great to correct some mistakes in the mean
> centering I was still doing (like demeaning within groups in the second
> case).
>
> However, I still have some doubts about the design which is most suitable
> for my work.
> In my case the covariate is the task performance, a continuous variable
> which is constant for controls (at the maximum value: everyone perform well)
> and generally increases with the increasing of the activation (and vice
> versa) for the two group of patients (one group with mild impairment has
> high activation and acceptable performance, the other has bad performance
> and small activation).
> Considering this, which model should I choose? What should I look at in
> taking this decision?
>
> Thank you very much-
>
>
> Giulia
>
>
>
>> Date: Mon, 23 Jul 2012 15:36:19 -0400
>> From: [log in to unmask]
>> Subject: Re: [FSL] GLM - ANOVA with covariates
>> To: [log in to unmask]
>
>>
>> The designs are different and have different assumptions about the
>> data. The following website has a great explanation of your two
>> models:
>>
>> http://mumford.fmripower.org/mean_centering/
>>
>> Best Regards, Donald McLaren
>> =================
>> D.G. McLaren, Ph.D.
>> Research Fellow, Department of Neurology, Massachusetts General Hospital
>> and
>> Harvard Medical School
>> Postdoctoral Research Fellow, GRECC, Bedford VA
>> Website: http://www.martinos.org/~mclaren
>> Office: (773) 406-2464
>> =====================
>> This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED
>> HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is
>> intended only for the use of the individual or entity named above. If the
>> reader of the e-mail is not the intended recipient or the employee or
>> agent
>> responsible for delivering it to the intended recipient, you are hereby
>> notified that you are in possession of confidential and privileged
>> information. Any unauthorized use, disclosure, copying or the taking of
>> any
>> action in reliance on the contents of this information is strictly
>> prohibited and may be unlawful. If you have received this e-mail
>> unintentionally, please immediately notify the sender via telephone at
>> (773)
>> 406-2464 or email.
>>
>>
>> On Mon, Jul 23, 2012 at 5:52 AM, Giulia Preti <[log in to unmask]>
>> wrote:
>> > Hi there,
>> >
>> > I'm running an fMRI analysis and I'm trying to set up a design matrix to
>> > perform an ANOVA with 3 groups of subjects (controls, two groups of
>> > patients) and one covariate (task performance).
>> >
>> > I set up the design matrix like in the example below:
>> >
>> >
>> > Group EV1 EV2 EV3 EV4
>> > 1 1 0 0 9,70
>> > 1 1 0 0 9,70
>> > 1 1 0 0 9,70
>> > 1 1 0 0 9,70
>> > 1 0 1 0 7,70
>> > 1 0 1 0 4,14
>> > 1 0 1 0 -12,52
>> > 1 0 0 1 -10,30
>> > 1 0 0 1 -20,86
>> > 1 0 0 1 -6,97
>> >
>> > CONTRASTS:
>> >
>> > 1 -1 0 0
>> > 1 0 -1 0
>> > 0 1 -1 0
>> > -1 1 0 0
>> > -1 0 1 0
>> > 0 -1 1 0
>> >
>> >
>> >
>> >
>> > Is this correct? Is it the same if I specify the value 1,2,3 for the
>> > column "Group" and split the performance column (EV4) in 2 different columns
>> > like below (demeaning for each column separately and deleting the column of
>> > the controls'performances, since it woul be all 0)?
>> >
>> >
>> > Group EV1 EV2 EV3 EV4 EV5
>> > 1 1 0 0 0 0
>> > 1 1 0 0 0 0
>> > 1 1 0 0 0 0
>> > 1 1 0 0 0 0
>> > 2 0 1 0 7,3 0
>> > 2 0 1 0 3,75 0
>> > 2 0 1 0 -12,9 0
>> > 3 0 0 1 0 2.41
>> > 3 0 0 1 0 -8.15
>> > 3 0 0 1 0 5.74
>> >
>> >
>> >
>> > What is the difference between the two designs, and which is better?
>> >
>> >
>> > Thank you in advance for your help.
>> >
>> >
>> > Giulia
|