Hi Anderson,
Thank you for your reply. In the case that I want to compare the patient's timepoint1 with the controls and I also want to put all image in one design, could I use this design:
EB EV1 EV2 EV3 EV4 EV5 EV6 EV7 EV8
Con 1 1 0 0 0 0 0 0 0
Con 1 1 0 0 0 0 0 0 0
Con 1 1 0 0 0 0 0 0 0
Con 1 1 0 0 0 0 0 0 0
Con 1 1 0 0 0 0 0 0 0
Con 1 1 0 0 0 0 0 0 0
Pat_t1 2 0 1 0 0 0 0 0 0
Pat_t2 2 0 1 1 0 0 0 0 0
Pat_t1 3 0 1 0 0 0 0 0 0
Pat_t2 3 0 1 0 1 0 0 0 0
Pat_t1 4 0 1 0 0 0 0 0 0
Pat_t2 4 0 1 0 0 1 0 0 0
Pat_t1 5 0 1 0 0 0 0 0 0
Pat_t2 5 0 1 0 0 0 1 0 0
Pat_t1 6 0 1 0 0 0 0 0 0
Pat_t2 6 0 1 0 0 0 0 1 0
Pat_t1 7 0 1 0 0 0 0 0 0
Pat_t2 7 0 1 0 0 0 0 0 1
contrast: Control > Patient timepoint 1:
1 -1 0 0 0 0 0 0
Thank you so much,
Best regard,
Kim
On Sat, 11 Jun 2016 07:55:39 +0100, Anderson M. Winkler <[log in to unmask]> wrote:
>Hi Kim,
>
>The test statistic will be the same, although the p-values may be
>different. To have identical p-values, for designs 2 and 3, would need to
>drop the -within, leaving -whole only. However, the p-values would still
>differ a bit, even if using the same seed, as with designs of different
>sizes, the random permutations will rarely be equivalent.
>
>All the best,
>
>Anderson
>
>
>On 10 June 2016 at 13:50, Kim TRAN <[log in to unmask]> wrote:
>
>> Hi Anderson,
>>
>> As one timepoint is knocked out from each subject in design 2 and 3, if I
>> enter in model only the data of timepoint 1 or 2, the design 2/3 and this
>> one should give the same results?
>>
>> EB EV1 EV2
>> 1 1 0
>> 1 1 0
>> 1 1 0
>> 1 1 0
>> 1 1 0
>> 1 0 1
>> 1 0 1
>> 1 0 1
>>
>> Thanks a lots,
>>
>> Best regard,
>>
>> Kim TD
>>
>> On Wed, 20 Jan 2016 08:58:10 +0000, Anderson M. Winkler <
>> [log in to unmask]> wrote:
>>
>> >Hi Kim,
>> >
>> >Design 1 tests within-subject effects (if there is a difference between
>> >timepoints, for instance, and interactions). Design 4 tests
>> between-subject
>> >effects (if there is a group difference across both timepoints).
>> >
>> >In design 1, the subject-specific intercepts are entered as additional,
>> >confound EVs. In design 4, the subject-specific slopes are entered as
>> >additional, confound EVs.
>> >
>> >Designs 2 and 3 do something else: for these designs either of the two
>> >timepoints is knocked out from each subject, such that it's as if these
>> >marked timepoints didn't exist. Design 2 causes t2 to be ignored, such
>> that
>> >comparisons are between groups on what concerns timepoint 1 only. Design 3
>> >causes t1 to be ignored, such that comparisons are between groups on what
>> >concerns timepoint 2 only.
>> >
>> >So, please note that for designs 2 and 3, it's the opposite than what you
>> >wrote below, i.e., design 2 tests timepoint 1, whereas design 3 tests
>> >timepoint 2.
>> >
>> >All the best,
>> >
>> >Anderson
>> >
>> >
>> >On 19 January 2016 at 13:07, Kim TRAN <[log in to unmask]> wrote:
>> >
>> >> Hi Anderson,
>> >>
>> >> I have a small question, what is the difference between design1 et
>> design4
>> >> in the file that you showed me? I think they're kind of similar.
>> Design2 is
>> >> used for testing 2nd scan because values for EV3-Ev22 are in the 2nd
>> scan,
>> >> and design3 is used for the 1st scan, isn't it? Could you check this for
>> >> me? Thank you very much,
>> >>
>> >> Best regard,
>> >>
>> >> Kim
>> >>
>> >> On Sat, 14 Nov 2015 07:44:22 +0000, Anderson M. Winkler <
>> >> [log in to unmask]> wrote:
>> >>
>> >> >Hi Kim,
>> >> >
>> >> >These tests need to go in separate designs. You can still correct
>> across
>> >> >all with -corrcon in PALM. Or prune your hypotheses to what is more
>> >> >relevant (presumably the effect of the treatment, which is in design
>> 1),
>> >> >and run that in randomise or FEAT.
>> >> >
>> >> >All the best,
>> >> >
>> >> >Anderson
>> >> >
>> >> >On 13 November 2015 at 09:20, Kim TRAN <[log in to unmask]> wrote:
>> >> >
>> >> >> Hi Anderson,
>> >> >>
>> >> >> I am so sorry for having bothered you. I've found the response in
>> others
>> >> >> sheets of your file. It's all that I want, but I cannot make it just
>> >> with
>> >> >> one design?
>> >> >>
>> >> >> Thanks a lots,
>> >> >> Kim
>> >> >>
>> >> >> On Fri, 13 Nov 2015 08:58:05 +0000, Kim TRAN <[log in to unmask]>
>> >> >> wrote:
>> >> >>
>> >> >> >Hi Anderson,
>> >> >> >
>> >> >> >Thank you very much for your reply. The contrasts C1 and C2 in your
>> >> >> attached file are the differences (t1-t2) for each group? Because I
>> >> want to
>> >> >> run a one side t-test, could I add these 2 more contrasts for a
>> complete
>> >> >> approach:
>> >> >> >C5: -1 0 0 0 0 ... (t2-t1 in group 1)
>> >> >> >C6: 0 -1 0 0 0 ... (t2-t1 in group 2)
>> >> >> >
>> >> >> >Might they change the interpretation of C3 and C4?
>> >> >> >
>> >> >> >Thanks a lots,
>> >> >> >
>> >> >> >Kim
>> >> >> >
>> >> >> >
>> >> >> >On Fri, 13 Nov 2015 07:46:47 +0000, Anderson M. Winkler <
>> >> >> [log in to unmask]> wrote:
>> >> >> >
>> >> >> >>Hi Kim,
>> >> >> >>
>> >> >> >>Design is fine, but the contrasts need to be changed. Please,
>> delete
>> >> all
>> >> >> of
>> >> >> >>them and instead use:
>> >> >> >>
>> >> >> >>C1: 1 -1 0 0 0 0 ....
>> >> >> >>C2: -1 1 0 0 0 0 ....
>> >> >> >>
>> >> >> >>These will test the effect of the treatment, and can be used in
>> FEAT,
>> >> >> >>randomise, or PALM.
>> >> >> >>
>> >> >> >>However, it seems you also want to compare groups in the 1st and
>> 2nd
>> >> >> scans,
>> >> >> >>is this correct? If yes, please see at this link
>> >> >> >><
>> >> >>
>> >>
>> https://dl.dropboxusercontent.com/u/2785709/outbox/mailinglist/design_kimtran.ods
>> >> >> >
>> >> >> >>the relevant designs. You'd need to use PALM, in which case you can
>> >> also
>> >> >> >>correct across all of them in the same run (option -corrcon).
>> Include
>> >> the
>> >> >> >>options -whole and -within (both) if you take this route.
>> >> >> >>
>> >> >> >>All the best,
>> >> >> >>
>> >> >> >>Anderson
>> >> >> >>
>> >> >> >>
>> >> >> >>
>> >> >> >>On 12 November 2015 at 16:24, Kim TRAN <[log in to unmask]>
>> wrote:
>> >> >> >>
>> >> >> >>> Dear FSL experts,
>> >> >> >>>
>> >> >> >>> I have 8 subjects, each scanned 2 times at 2 times point. The
>> fisrt
>> >> 5
>> >> >> >>> subjects are in drug group, and another 3 are in placebo group. I
>> >> want
>> >> >> to
>> >> >> >>> compare results of those 2 groups. Are the design and contrasts
>> >> below
>> >> >> >>> correct?
>> >> >> >>>
>> >> >> >>> EB EV1 EV2 EV3 EV4 EV5 EV6 EV7
>> EV8
>> >> >> >>> EV9 EV10
>> >> >> >>> 1 1 0 1 0 0 0 0 0
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 1 -1 0 1 0 0 0 0 0
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 2 1 0 0 1 0 0 0 0
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 2 -1 0 0 1 0 0 0 0
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 3 1 0 0 0 1 0 0 0
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 3 -1 0 0 0 1 0 0 0
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 4 1 0 0 0 0 1 0 0
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 4 -1 0 0 0 0 1 0 0
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 5 1 0 0 0 0 0 1 0
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 5 -1 0 0 0 0 0 1 0
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 6 0 1 0 0 0 0 0 1
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 6 0 -1 0 0 0 0 0 1
>> >> >> 0
>> >> >> >>> 0
>> >> >> >>> 7 0 1 0 0 0 0 0 0
>> >> >> 1
>> >> >> >>> 0
>> >> >> >>> 7 0 -1 0 0 0 0 0 0
>> >> >> 1
>> >> >> >>> 0
>> >> >> >>> 8 0 1 0 0 0 0 0 0
>> >> >> 0
>> >> >> >>> 1
>> >> >> >>> 8 0 -1 0 0 0 0 0 0
>> >> >> 0
>> >> >> >>> 1
>> >> >> >>>
>> >> >> >>> Contrast:
>> >> >> >>> mean G1, 1st scan 5 0 1 1 1 1
>> >> >> 1
>> >> >> >>> 0 0 0
>> >> >> >>> mean G1, 2nd scan -5 0 1 1 1 1
>> >> >> 1
>> >> >> >>> 0 0 0
>> >> >> >>> mean G2, 1st scan 0 3 0 0 0 0
>> >> >> 0
>> >> >> >>> 1 1 1
>> >> >> >>> mean G2, 2nd scan 0 -3 0 0 0 0
>> >> >> 0
>> >> >> >>> 1 1 1
>> >> >> >>> G1 > G2, 1st scan 5 -3 1 1 1 1
>> >> >> 1
>> >> >> >>> -1 -1 -1
>> >> >> >>> G1 > G2, 2nd scan -5 3 1 1 1 1
>> >> >> 1
>> >> >> >>> -1 -1 -1
>> >> >> >>>
>> >> >> >>> Any help you can provide will be greatly appreciated.
>> >> >> >>> Best regard,
>> >> >> >>> Kim
>> >> >> >>>
>> >> >> >>
>> >> >>
>> >> >
>> >>
>> >
>>
>
|
