Tom
Sorry for being confusing. The p values are not exactly one but they
are very close. The range for A-B and B- A are as follows for the 1-9
grp file
NAR_tfce_corrp_tstat1.nii.gz 0.000000 0.078125
NAR_tfce_corrp_tstat2.nii.gz 0.000000 0.003906
and for the col of 1s, my range is:
tfce_corrp_tstat1.nii.gz 0.000000 0.656400
tfce_coorp_tstat2.nii.gz 0.000000 0.538800
So they don't seem funny to me other than the fact that the first
group file should give results with much greater intensities ie lower
p values.
Jeff
On 6/28/10, Thomas Nichols <[log in to unmask]> wrote:
> Dear Jeff,
>
> By using a group file of all ones, you're letting randomise flip everything
> around unrestricted, including permutations where the 1st 8 scans are four
> subjects, and the last 8 scans are four different subjects. Such a
> (unrestricted) relabeling may generate very large differences, i.e. large T
> scores in permutation, which will deflate significances.
>
> Now, getting all 1's for the 1-corrected P image (i.e. all zero FWE
> P-values!) doesn't sound right either. Can I just confirm you're not
> getting exactly 1.0 (it is impossible to get 0 p-values with permutation)?
> e.g. with fslstats blah_corrp -R, which should show a maximum of at most
> 1-1/nPerm. What does your distribution of t-values look like? Are they
> funny? (I.e. shifted away from zero?)
>
> -Tom
>
>
>
> On Mon, Jun 28, 2010 at 6:32 PM, Jeff Churchill <[log in to unmask]> wrote:
>
>> Tom and all other VBM experts
>> I set up grp files two ways. The first way was just the output of the
>> GLM which had one group (a column of all one's) which I ran but did
>> not think that this would make for a paired ttest. So I ran what I
>> thought would make for a paired ttest by having my group file as shown
>> below and for some reason I have yet to figure out how the 1-9 group
>> file yielded a result with corrected p values all 1s but the column of
>> 1s group file yielded p values as low as 0.4. Also the uncorrected
>> intensity values (ie 1-P) showed up with a lesser intensity for the
>> 1-9 group than the simple group uncorrected version. Which in my
>> understanding means that the program worked but when corrected the 1-9
>> group version yielded much
>> higher p values and less significant results. What could be the cause
>> for these strange findings? Could it have anything to do with a messed
>> up left right orientation?
>> Thank you,
>> Jeff
>>
>> 1
>> 2
>> 3
>> 4
>> 5
>> 6
>> 7
>> 8
>> 9
>> 1
>> 2
>> 3
>> 4
>> 5
>> 6
>> 7
>> 8
>> 9
>>
>> On 6/22/10, Stephen Smith <[log in to unmask]> wrote:
>> > You also need to delete your EV1.
>> > Cheers.
>> >
>> >
>> > On 22 Jun 2010, at 14:30, Thomas Nichols wrote:
>> >
>> >> Dear Jeffrey,
>> >>
>> >> Yes, you do need the group file. See the "Repeated measures ANOVA"
>> >> section of the randomise docs for more info.
>> >>
>> >> -Tom
>> >>
>> >>
>> >> On Tue, Jun 22, 2010 at 2:03 AM, Jeff Churchill <[log in to unmask]>
>> >> wrote:
>> >> Thank you for your response and expertise,
>> >> I attempted to follow your advice and was hoping that this was the
>> correct
>> >> format for my design files. I would very much appreciate it if you
>> >> could
>> >> take a look and let me know. Also is it necessary to make a design.grp
>> >> file for this paired t-test analysis?
>> >> Thank you very much,
>> >> Jeffrey
>> >>
>> >> Design.mat:
>> >> %! VEST-Waveform File
>> >> /NumWaves 11
>> >> /NumPoints 18
>> >> /Skip 0
>> >> /PPheights 1 1 1 1 1 1 1 1 1 1 1
>> >> /Matrix
>> >> 1 1 1 0 0 0 0 0 0 0 0
>> >> 1 1 0 1 0 0 0 0 0 0 0
>> >> 1 1 0 0 1 0 0 0 0 0 0
>> >> 1 1 0 0 0 1 0 0 0 0 0
>> >> 1 1 0 0 0 0 1 0 0 0 0
>> >> 1 1 0 0 0 0 0 1 0 0 0
>> >> 1 1 0 0 0 0 0 0 1 0 0
>> >> 1 1 0 0 0 0 0 0 0 1 0
>> >> 1 1 0 0 0 0 0 0 0 0 1
>> >> 1 -1 1 0 0 0 0 0 0 0 0
>> >> 1 -1 0 1 0 0 0 0 0 0 0
>> >> 1 -1 0 0 1 0 0 0 0 0 0
>> >> 1 -1 0 0 0 1 0 0 0 0 0
>> >> 1 -1 0 0 0 0 1 0 0 0 0
>> >> 1 -1 0 0 0 0 0 1 0 0 0
>> >> 1 -1 0 0 0 0 0 0 1 0 0
>> >> 1 -1 0 0 0 0 0 0 0 1 0
>> >> 1 -1 0 0 0 0 0 0 0 0 1
>> >>
>> >> Design.con
>> >> %! VEST-Waveform File
>> >> /NumWaves 10
>> >> /NumPoints 2
>> >> /Skip 0
>> >> /PPheights 1 1 1 1 1 1 1 1 1 1
>> >> /Matrix
>> >> 1 0 0 0 0 0 0 0 0 0
>> >> -1 0 0 0 0 0 0 0 0 0
>> >>
>> >>
>> >> On Mon, Jun 21, 2010 at 4:17 AM, Stephen Smith <[log in to unmask]>
>> >> wrote:
>> >> Hi - this isn't a valid design for the paired t-test case - see the
>> >> FEAT
>> >> higher-level examples in the manual for a correct example.
>> >> Cheers.
>> >>
>> >>
>> >> On 18 Jun 2010, at 20:31, Jeff Churchill wrote:
>> >>
>> >>> Dear FSL Experts,
>> >>>
>> >>> I have run FSL-VBM for a longitudinal study and found significant
>> results
>> >>> in
>> >>> the forward direction (before - after) and no significant results for
>> >>> the reverse direction (after - before). I wanted to make sure that
>> >>> there were no errors when i ran the program so I re-ran it and got the
>> >>> same
>> >>> results. I then switched the before and after groups and re-ran it
>> >>> with
>> >>> the
>> >>> same matrix and contrast files. However, this time my results didn't
>> >>> match
>> >>> up with what I already had when I ran it in the forward direction.
>> Given
>> >>> my
>> >>> matrix I thought that I should see the same results but flipped (ie I
>> >>> should see no
>> >>> significant results for the forwards direction and have the same
>> >>> significant
>> >>> results for the backwards direction).
>> >>> Should this difference be a concern? or am I misunderstanding
>> something?
>> >>> Thank You
>> >>> Jeff
>> >>>
>> >>> Here are my matrix and contrast files:
>> >>> Design.mat:
>> >>> %! VEST-Waveform File
>> >>> /NumWaves 11
>> >>> /NumPoints 18
>> >>> /Skip 0
>> >>> /PPheights 1 1 1 1 1 1 1 1 1 1 1
>> >>>
>> >>> /Matrix
>> >>> 1 0 0 0 0 0 0 0 0 1 0
>> >>> 0 1 0 0 0 0 0 0 0 1 0
>> >>> 0 0 1 0 0 0 0 0 0 1 0
>> >>> 0 0 0 1 0 0 0 0 0 1 0
>> >>> 0 0 0 0 1 0 0 0 0 1 0
>> >>> 0 0 0 0 0 1 0 0 0 1 0
>> >>> 0 0 0 0 0 0 1 0 0 1 0
>> >>> 0 0 0 0 0 0 0 1 0 1 0
>> >>> 0 0 0 0 0 0 0 0 1 1 0
>> >>> 1 0 0 0 0 0 0 0 0 0 1
>> >>> 0 1 0 0 0 0 0 0 0 0 1
>> >>> 0 0 1 0 0 0 0 0 0 0 1
>> >>> 0 0 0 1 0 0 0 0 0 0 1
>> >>> 0 0 0 0 1 0 0 0 0 0 1
>> >>> 0 0 0 0 0 1 0 0 0 0 1
>> >>> 0 0 0 0 0 0 1 0 0 0 1
>> >>> 0 0 0 0 0 0 0 1 0 0 1
>> >>> 0 0 0 0 0 0 0 0 1 0 1
>> >>>
>> >>>
>> >>> Design.con
>> >>> %! VEST-Waveform File
>> >>> /NumWaves 11
>> >>> /NumPoints 2
>> >>> /Skip 0
>> >>> /PPheights 1 1 1 1 1 1 1 1 1 1 1
>> >>>
>> >>> /Matrix
>> >>> 0 0 0 0 0 0 0 0 0 1 -1
>> >>> 0 0 0 0 0 0 0 0 0 -1 1
>> >>>
>> >>
>> >>
>> >>
>> ---------------------------------------------------------------------------
>> >> Stephen M. Smith, Professor of Biomedical Engineering
>> >> Associate Director, Oxford University FMRIB Centre
>> >>
>> >> FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
>> >> +44 (0) 1865 222726 (fax 222717)
>> >> [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
>> >>
>> ---------------------------------------------------------------------------
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> --
>> >> ____________________________________________
>> >> Thomas Nichols, PhD
>> >> Principal Research Fellow, Head of Neuroimaging Statistics
>> >> Department of Statistics & Warwick Manufacturing Group
>> >> University of Warwick
>> >> Coventry CV4 7AL
>> >> United Kingdom
>> >>
>> >> Email: [log in to unmask]
>> >> Phone, Stats: +44 24761 51086, WMG: +44 24761 50752
>> >> Fax: +44 24 7652 4532
>> >>
>> >
>> >
>> >
>> ---------------------------------------------------------------------------
>> > Stephen M. Smith, Professor of Biomedical Engineering
>> > Associate Director, Oxford University FMRIB Centre
>> >
>> > FMRIB, JR Hospital, Headington, Oxford OX3 9DU, UK
>> > +44 (0) 1865 222726 (fax 222717)
>> > [log in to unmask] http://www.fmrib.ox.ac.uk/~steve
>> >
>> ---------------------------------------------------------------------------
>> >
>> >
>> >
>> >
>>
>
>
>
> --
> ____________________________________________
> Thomas Nichols, PhD
> Principal Research Fellow, Head of Neuroimaging Statistics
> Department of Statistics & Warwick Manufacturing Group
> University of Warwick
> Coventry CV4 7AL
> United Kingdom
>
> Email: [log in to unmask]
> Phone, Stats: +44 24761 51086, WMG: +44 24761 50752
> Fax: +44 24 7652 4532
>
|