Erin Mazerolle wrote:
> Hello,
> In the SPM99 manual it states that to do second-level analyses, first-level
> analyses must be the same. I am trying to analyze event-related data in which
> different subjects were presented stimuli in a different order so that their
> design matrices are slightly different. Will this prevent me from doing
> second-level analyses?
Strictly, Yes.
But in practice, No.
The Summary Statistic (SS) approach (eg.
doing t-tests on con images) assumes the
precision with which these contrasts are estimated
is the same for all subjects.
In practice the SS approach is robust to
violations of this assumption. You need
precisions which vary an order of magnitude
to throw off SS.
A different order of events is unlikely to
cause this. In fact, in the RFX exemplar
analysis on the SPM data web page, the order
of events was different for different subjects.
The only reliable way to throw SS off
in practice is to have an orders of magnitude
difference in the number of events (which
could perhaps occur if this isnt under
experimental control eg. in epilepsy research).
Best wishes,
Will.
> Thanks,
> -Erin Mazerolle
>
>
>
--
William D. Penny
Wellcome Department of Imaging Neuroscience
University College London
12 Queen Square
London WC1N 3BG
Tel: 020 7833 7478
FAX: 020 7813 1420
Email: [log in to unmask]
URL: http://www.fil.ion.ucl.ac.uk/~wpenny/
|