Thanks for your thoughts on this Donald. I've been checking through the SPM-list history looking for the discussion, but couldn't find much related to fMRI. But your comments are very helpful.

I take your point regarding two-sample tests. One other important thing you mention is whether to centre the covariate. When I look at the GUI options for a one-sample t-test, I don't see any options under Global calculation or Global normalisation to centre the mean. There is an option to do proportional scaling or ANCOVA - is this relevant?

Best,

Rich
 


On 01/05/2012, at 1:49 PM, MCLAREN, Donald wrote:

This question has come up a number of times. My thoughts are that you don't want to include the global activation as a covariate. The reason for this is that it changes the interpretation of the results.

Depending on if you center, or do not center the covariate, then you will be measuring the covariate-adjusted mean or intercept, respectively.

What does covariate-adjusted mean? It means that the effect is measured as if all subjects had the same global average effect. For a one-sample t-test, this will not change the mean. However, for two-sample t-tests, it can effect the results and interpretation.

Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Postdoctoral Research Fellow, GRECC, Bedford VA
Research Fellow, Department of Neurology, Massachusetts General Hospital and
Harvard Medical School
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 Fri, Apr 27, 2012 at 9:38 PM, Richard Morris <[log in to unmask]> wrote:
My bad, I was looking for subject effects in paired t-tests. And I can now see SPM5 and SPM8 are equivalent. Thanks Donald.

Including the global mean intensity as a covariate has me intrigued. In principle, wouldn't it make sense to include these values as a nuisance effect in fMRI, just as in PET? I guess in the paired t-test case, any subject differences in global mean intensity should be removed by the comparison, but for a one sample t-test or a two sample t-test it would seem to make sense…(?)

Ta,

Rich


On 28/04/2012, at 12:24 AM, MCLAREN, Donald wrote:

> (1) If you include a column for each subject in a one-sample t-test,
> then the group effect will always have a t-statistic of Inf. This
> results from the residual being 0. As far as I know, this was never
> done in SPM5.
>
> (2) I do not have a strong opinion on this, except to say that your
> contrast will no longer be comparing TaskA v. TaskB or TaskA v.
> baseline. Rather, you will be comparing the globally-adjusted
> contrast.
>
> (3) What subject-specific are you referring to?
>
> From the wording in your email, it seems like you might be using
> multiple measurements per subject. In this case, you should not be
> using a one-sample t-test.
>
> Best Regards, Donald McLaren
> =================
> D.G. McLaren, Ph.D.
> Postdoctoral Research Fellow, GRECC, Bedford VA
> Research Fellow, Department of Neurology, Massachusetts General Hospital and
> Harvard Medical School
> 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 Fri, Apr 27, 2012 at 6:43 AM, Richard Morris
> <[log in to unmask]> wrote:
>> Dear list,
>>
>> I've noticed a change in the one-sample t-test design matrices produced by SPM8 compared to SPM5. I seem to remember SPM5 used to (automagically?) include subject effects, with columns for each subject in the design matrix, but SPM8 doesn't do that anymore. Is there a setting change I can make to reintroduce subject effects in my design matrix?
>>
>> On a related note, is there any reason not to include the global covariate (ANCOVA) in a one-sample t-test  of fMRI data? I've noticed I often get more regionally specific activation with higher t-values in my results when I include the global covariate. I'm guessing this is because of subject-specific effects which I would like to remove/minimise.
>>
>> Best,
>>
>> Rich