On Wed, 27 Jan 2010 10:36:45 +0000, Sophie Green <Sophie.Green-
[log in to unmask]> wrote:
>Dear SPMers,
>
>I have another question regarding PPI analyses..
>
>Is it possible to see significant PPI effects in voxels that are not
>significantly present in a standard analysis?
>
>I would have thought that this could be the case since you are testing
>for a change in regression slope between two conditions, rather than
>activation of an area above a certain threshold per se. Presumably,
>you could still get coupling of areas, which may not have 'great
>enough' activation in a standard analysis, but if the contribution of
>one area explains the activation in another area under a given
>condition and this changes with condition, this could be significant?
As a theoretical matter, yes, it's certainly possible. More abstractly, you can
have significant interaction terms even if main effects aren't significant,
assuming the former isn't highly confounded with the latter (which is the case
in a truly multifactorial experiment).
One could ask the empirical question---how often does this happen. I'm not
sure about that.
>
>Very best wishes,
>Sophie
>
>
>
>
>
>--
>Sophie Green
>PhD Student
>Neuroscience and Aphasia Research Unit (NARU)
>Zochonis Building, 3rd Floor
>Brunswick Street
>University of Manchester
>Manchester M13 9PL
>
>Tel: +44 (0)161 306 0451
|