Jason Craggs wrote:
> Greetings,
>
> I have a 'simple' question about the contrasts listed below. -1 0 1 --> group 1 no = group 3.
>
> Does this intrinsically mean that both groups are different from, say, rest?
No.
If 'rest' is defined as anything that's not conditions 1, 2 or 3 then eg. a [1 0 0]
F contrast will tell you that condition 1 is different from rest.
Its possible that the response to condition 1 is slightly negative and the
response to condition 3 is slightly positive - so the difference could be
significant without each being significantly non-zero.
Similarly, condition 1 may show no response, condition 3 may be significantly
non-zero resulting in a significant difference etc..
You can mask the result of one contrast with another to test for exactly the
effect you are after - see eg. Rik Henson's email from a couple of days ago.
Best,
Will.
> So that we can make
> the statement that the voxels in both group significantly predict voxel intensities/unit time,
> and that those significant voxles are now also significantly different from each other?
>
> Thanks Jason
>
>
>
> -----Original Message----- From: SPM (Statistical Parametric Mapping) on behalf of Will Penny
> Sent: Thu 2/3/2005 8:00 AM To: [log in to unmask] Subject: [SPM] [Fwd: Re: spm contrasts]
>
>
>> I wanted to do contrasts for 3 groups in which i want to look at gross differences between the
>> 3 groups simultaneously. I specified the F-contrast in one-way ANOVA test as such: -1 0 1
>> (Group1 < Group3) -1 1 0 (Group1 < Group2) 0 -1 1 (Group2 < Group3)
>>
>> I have one scan/one condition per patient. I was hoping this would show me hypo-active areas in
>> Group1. When I change the contrast to 1 0 -1 1 -1 0 0 1 -1
>>
>> I get the same map. Should not the second one show the hyperactive areas? Thank you so much,
>>
>
>
> No.
>
> F-contrasts look for two-sided effects.
>
> With both contrasts you are testing
>
> (Group1 not = Group 3) or (Group 1 not = Group 2) or (Group2 not = Group 3)
>
> Best,
>
> Will.
>
>
> -- William D. Penny Wellcome Department of Imaging Neuroscience University College London 12
> Queen Square London WC1N 3BG
>
> Tel: 020 7833 7475 FAX: 020 7813 1420 Email: [log in to unmask] URL:
> http://www.fil.ion.ucl.ac.uk/~wpenny/
>
>
>
--
William D. Penny
Wellcome Department of Imaging Neuroscience
University College London
12 Queen Square
London WC1N 3BG
Tel: 020 7833 7475
FAX: 020 7813 1420
Email: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|