I thought so, but I will check with the statistician next Monday, I'll get back to you as soon as I'm sure.
Date: Fri, 30 Jan 2015 14:12:32 -0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] average values per cluster
Actually, that looks correct. Is this what you fed to SPSS?
On 1/30/15 2:07 PM, maaike rive wrote:
Ok, I'm so sorry, I'm afraid I mixed things up. I
indeed used DODS, and assumed these were may regressors:
diagn1*state1*gender1
diagn1*state1*gender2
diagn1*state2*gender1
diagn1*state2*gender2
diagn2*state1*gender1
diagn2*state1*gender2
diagn2*state2*gender1
diagn2*state2*gender2
diagn1*state1*gender1*age
diagn1*state1*gender2*age
diagn1*state2*gender1*age
diagn1*state2*gender2*age
diagn2*state1*gender1*age
diagn2*state1*gender2*age
diagn2*state2*gender1*age
diagn2*state2*gender2*age
However
what I actually meant to do was testing for the effects of
diagnosis and interactions (e.g. diagnosis*age) by using
the appropriate
contrasts, so I thought that if I used the contrast for
the diagnosis*age
interaction, I was testing whether or not this interaction
was significant
regressing out the effects of state and gender. Another
example, I thought that if I used the contrast for
diagnosis1 versus diagnosis2, I was testing wether or not
there was a significant effect of diagnosis regressing out the
effects of state, gender and age.
But that
is not the case than? How should I build the model to
answer these questions?
Maaike
Date: Fri, 30 Jan 2015 13:33:40 -0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] average values per cluster
It depends on the actual covariates and whether you used
DOSS or
DODS. The default is DODS, which is a full interaction
model.
Assuming that state is a categorical variable with 2 levels,
you'd
have 4 regressors (gender by state) + 4 more regressors
(gender by
state by age). If the SPSS model did not have 8 regressors,
then it
is not the same.
doug
On 1/30/15 1:26 PM, maaike
rive wrote:
How should I have described the model?
Date: Fri, 30 Jan 2015
13:09:40 -0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] average values per cluster
>>>I told him it the model is a GLM with age,
gender
and state as covariates
This is not a sufficient description. Several design
matrices
could be derived from this description.
On 1/30/15 11:58 AM,
maaike
rive wrote:
The p value from SPSS is 0.156. I
checked
with a statistician, just to be sure, but he says
the SPSS
model is correct... (he doesn't know anything
about
freesurfer and FSGD files, but I told him it the
model is
a GLM with age, gender and state as covariates).
Date: Fri, 30 Jan 2015
11:46:37
-0500
From: gr...@nmr.mgh.harvard.edu
To: freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] average values per
cluster
You should report the clusterwise p-value. The
p-value
that you have computed is not interpretable. If
it did
not come out significant, then it would be
worrisome,
but it is only a check and cannot be used for
anything.
What p-value did you get from SPSS? I'd be
curious to
know what the source of the descrpancy is since
this has
happened several times now.
doug
On 1/30/15 3:25
AM,
maaike rive wrote:
Hi Douglas,
I used the matlab code and the p
value is
signifiant, although less so than the
p-value I
get from the clusterwise statistics (0.0027 vs
0.00010).
(Indeed I used an abs threshold for
the clusterwise statistics).
Which p-value should I
report?
Thanks,
Maaike
From: r_maa...@hotmail.com
To: freesurfer@nmr.mgh.harvard.edu
Date: Thu, 29 Jan 2015 19:48:44 +0100
Subject: Re: [Freesurfer] average
values per
cluster
No not yet; I will do
so!
> Date: Thu, 29 Jan 2015
13:44:33
-0500
> From: gr...@nmr.mgh.harvard.edu
> To: freesurfer@nmr.mgh.harvard.edu
> Subject: Re: [Freesurfer]
average
values per cluster
>
>
> The t is correct. When the
contrast
matrix only has one row, then the
F
> is just an unsigned t. I
don't know
anything about SPSS so I can't
tell
> from what you have sent
whether it is
the same model or not. Try to get
> SPSS to output the design
matrix. Did
you try the matlab code below?
>
> On 01/29/2015 01:40 PM,
maaike rive
wrote:
> > Hi Douglas,
> >
> > I checked but as far as
I see it
I used the same models. I attached
> > the SPSS model and
output as wel
as the FSGD file and contrast file
> > (to test a diagnosis x
age
interaction). Now that I come to
think of
> > it, I think the reason
for the
discrepant findings is that I used
a
> > t-contrast instead of an
F-contrast for the interaction in
Freesurfer.
> > Apologies I did not
think about
this earlier. But maybe there is
> > something else I did
completely
wrong.
> >
> > What do you think?
> >
> > If it is indeed the t vs
F
contrast, than how should I
specify that
> > the contrast is an
F-contrast?
Is it ok to just add another line
to
> > the contrastfile (so
> > 0 0 0 0 0 0 0 0 1 1 1 1
-1 -1 -1
-1
> > 0 0 0 0 0 0 0 0 -1 -1 -1
-1 1 1
1 1)?
> >
> > Thanks again,
> >
> > Maaike
> >
> > > Date: Mon, 26 Jan
2015
11:23:59 -0500
> > > From: gr...@nmr.mgh.harvard.edu
> > > To: freesurfer@nmr.mgh.harvard.edu
> > > Subject: Re:
[Freesurfer]
average values per cluster
> > >
> > >
> > > We get these kind
of
reports occasionally. When I ask
people to
confirm
> > > that they use
exactly the
same design matrix in SPSS, I
never hear
> > back,
> > > so I assume that it
gets
resolved. So please check. The
other thing
you
> > > can do is to run in
matlab,
something like
> > >
> > > cd glmdir/contrast
> > > X = load('Xg.dat');
> > > y =
load('ocn.dat');
> > > C = load('C.dat');
> > > [beta rvar] =
fast_glmfit(y,X);
> > > [F p] =
fast_fratio(beta,X,rvar,C);
> > > p will be the
p-value
> > >
> > > If you used an
unsigned
cluster-forming threshold (ie,
abs), then
it is
> > > possible that some
of the
voxels are pos and some are neg so
that
they
> > > average out
> > >
> > > doug
> > >
> > >
> > >
> > >
> > > On 01/26/2015 09:03
AM,
maaike rive wrote:
> > > > Dear
Freesurfer
experts,
> > > >
> > > > Sorry to
bother you
again, but I have two more
questions about
> > > > extracting
(thickness/surface/GI) values from
a
certain cluster.
> > > >
> > > > As I
understood, the
abs.y.ocn.dat file gives the
average
values
> > for a
> > > > given
significant
cluster (e.g. a cluster where
there is a
> > significant
> > > > AxB
interaction).
> > > >
> > > > I may be
completely
misunderstanding things, but if I
use
these
> > values
> > > > in SPSS for
further
statistics and test the same
interaction
(AxB),
> > > > than according
to SPSS
this interaction is /not
/significant
> > > > (corrected for
the
same covariates as in the FSGD
file).
> > > >
> > > > Could you tell
me what
is going wrong here? I do not
trust my
> > results now.
> > > >
> > > > Furthermore,
is it
possible (and if so, how?) to
extract the
average
> > > > values of
exactly the
same cluster, but in an
independent group
not
> > > > used in the
analysis,
for post-hoc comparisons in SPSS?
> > > >
> > > > Thanks,
> > > >
> > > > Maaike
> > > >
> > > >
> > > >
_______________________________________________
> > > > Freesurfer
mailing
list
> > > > Freesurfer@nmr.mgh.harvard.edu
> > > >
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
> > >
> > > --
> > > Douglas N. Greve,
Ph.D.
> > > MGH-NMR Center
> > > gr...@nmr.mgh.harvard.edu
> > > Phone Number:
617-724-2358
> > > Fax: 617-726-7422
> > >
> > > Bugs:
surfer.nmr.mgh.harvard.edu/fswiki/BugReporting
> > > FileDrop:
https://gate.nmr.mgh.harvard.edu/filedrop2
> > >
www.nmr.mgh.harvard.edu/facility/filedrop/index.html
> > > Outgoing:
ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
> > >
> > >
_______________________________________________
> > > Freesurfer mailing
list
> > > Freesurfer@nmr.mgh.harvard.edu
> > >
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
> > >
> > >
> > > The information in
this
e-mail is intended only for the
person to
> > whom it is
> > > addressed. If you
believe
this e-mail was sent to you in
error and
> > the e-mail
> > > contains patient
information, please contact the
Partners
Compliance
> > HelpLine at
> > >
http://www.partners.org/complianceline
. If the e-mail was sent to
> > you in error
> > > but does not
contain
patient information, please
contact the
sender
> > and properly
> > > dispose of the
e-mail.
> > >
> >
> >
> >
_______________________________________________
> > Freesurfer mailing list
> > Freesurfer@nmr.mgh.harvard.edu
> >
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
>
> --
> Douglas N. Greve, Ph.D.
> MGH-NMR Center
> gr...@nmr.mgh.harvard.edu
> Phone Number: 617-724-2358
> Fax: 617-726-7422
>
> Bugs:
surfer.nmr.mgh.harvard.edu/fswiki/BugReporting
> FileDrop:
https://gate.nmr.mgh.harvard.edu/filedrop2
>
www.nmr.mgh.harvard.edu/facility/filedrop/index.html
> Outgoing:
ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/
>
>
_______________________________________________
> Freesurfer mailing list
> Freesurfer@nmr.mgh.harvard.edu
>
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer
mailing list Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is
intended
only for the person to whom it is
addressed.
If you believe this e-mail was sent to
you in
error and the e-mail contains patient
information, please contact the
Partners
Compliance HelpLine at
http://www.partners.org/complianceline
. If the e-mail was sent to you in
error but
does not contain patient information,
please
contact the sender and properly
dispose of the
e-mail.
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only
for the
person to whom it is addressed. If you believe
this
e-mail was sent to you in error and the e-mail
contains
patient information, please contact the Partners
Compliance HelpLine at
http://www.partners.org/complianceline
. If the e-mail was sent to you in error but
does not
contain patient information, please contact the
sender
and properly dispose of the e-mail.
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for
the person
to whom it is
addressed. If you believe this e-mail was sent to you
in error
and the e-mail
contains patient information, please contact the
Partners
Compliance HelpLine at
http://www.partners.org/complianceline
. If the e-mail was
sent to you in error
but does not contain patient information, please
contact the
sender and properly
dispose of the e-mail.
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the
person to whom it is
addressed. If you believe this e-mail was sent to you in
error and the e-mail
contains patient information, please contact the Partners
Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was
sent to you in error
but does not contain patient information, please contact the
sender and properly
dispose of the e-mail.
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
The information in this e-mail is intended only for the person to whom it is
addressed. If you believe this e-mail was sent to you in error and the e-mail
contains patient information, please contact the Partners Compliance HelpLine at
http://www.partners.org/complianceline . If the e-mail was sent to you in error
but does not contain patient information, please contact the sender and properly
dispose of the e-mail. _______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
