Thank you very much.
In fact I have read your paper about rfx, however
there are still two problems I cannot understand.
1. Is the one sample t-test applied to residuals of
first level results? According to the literatures, the
first level outputs are modelled as a_i = a + e_i, e_i
~ N(0,sigma). So it seems that the one sample t-test
should be applied to the resuidual. Since mean of the
residuals is zero, can we still used zero hypothesis
mean?
2. Besides beta value, can I use input other measures
from first level analysis to second level analysis? My
understanding of rfx is that the rfx is a framework
for random effect analysis, so any measures from first
level analysis can be input to rfx.
Thanks in advance,
Cliff
-----Original Message-----
From: SPM (Statistical Parametric Mapping)
[mailto:[log in to unmask]]
Sent: Friday, November 04, 2005 11:38 AM
To: [log in to unmask]
Subject: Re: [SPM] second level analysis and one
sample t-test
cliff macnab wrote:
> Dear All,
>
> I would like to know how the one sample t-test
is implemented for
> second level analysis. Suppose the "raw" data for a
one sample t-test
> are from 5
> subjects: a1, a2, a3, a4, a5, for the one sample
> t-test, a hypotheses mean has to be given. My
question
> is how to set the hypothese mean?
>
Usually a hypothesis mean of zero is used. In fact, in
the classical inference procedures implemented in SPM
there is no other option.
In SPM's Bayesian inference you can produce posterior
probability maps of the population effect being
greater than a user-specified size eg. 1% of the
global mean.
Best,
Will.
> Thanks,
>
> Cliff
>
>
>
> __________________________________
> Yahoo! FareChase: Search multiple travel sites in
one click.
> http://farechase.yahoo.com
>
>
--
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/
__________________________________
Yahoo! Mail - PC Magazine Editors' Choice 2005
http://mail.yahoo.com
|