Martijn Mulder wrote:
> Thank you Rik, that is helpful!
> I wasn't aware of that page.
>
> However, as mentioned by you:
> " Note that, if your resulting modulators are linearly-dependent, this
> will mean that you cannot estimate certain contrasts (namely those
> that don't sum to zero) - but this doesn't matter if you are always
> interested in *differences* between conditions, rather than the unique
> effect of each."
>
> Indeed, when I add each condition seperately -- [1 1 0 0 0 ... ], [0 0
> 1 1 1], ... etc. -- I can't estimate the effect of each condition
> seperately, but that's what I want to do actually (option 2 in my
> original post).
>
> Note that I didn't add RT as a modulator yet. I first want to check
> whether condition effects are the same as when I use a categorical
> design.
>
> I workaround would be option a) in this post from Tobias Egner:
> https://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind05&L=SPM&D=0&P=321362
> In which he suggest to get rt from the design where all trials are
> collapsed, and rt is added as parametric modulator.
>
> What do you think?
Yes, sounds possible. Good luck!
R
>
>
> On Jan 5, 2009, at 10:13 AM, Rik Henson wrote:
>
>> Martijn -
>>
>> I think this WIKIpage should help:
>>
>> http://imaging.mrc-cbu.cam.ac.uk/imaging/ParametricModulations
>>
>> Happy new year
>> Rik
>>
>>> Hello Rik Henson and/or other SPM'rs that have time and knowledge to
>>> share an answer,
>>>
>>> I want to control for (between condition) reaction time in a 2x2
>>> factorial design.
>>> I collapsed factors as parametric modulations, and add rt as
>>> paramatric modulator to control for global rt effects according to:
>>> https://www.jiscmail.ac.uk/cgi-bin/wa.exe?A2=ind06&L=SPM&P=R750063
>>>
>>> I can create a regressor for all trials (factor A + B), and add each
>>> factor as a parametric modulator by:
>>>
>>> 1) using 2 modulations 1) factor A: A1 as [1 1 1 1 ...] and A2
>>> as [-1 -1 -1 -1 ...]
>>> 2) factor B: B1 as [1 1 1 1 ...] and B2 as [-1 -1 -1 -1 ...]
>>> or
>>>
>>> 2) using 4 modulations 1) A1 as [ 1 1 0 0 0 0 0 0]
>>> 2) A2 as [ 0 0 1 1 0 0 0 0]
>>> 3) B1 as [ 0 0 0 0 1 1 0 0]
>>> 4) B2 as [ 0 0 0 0 0 0 1 1]
>>>
>>>
>>>
>>> I prefer the second option because I want to take effects of each
>>> seperate condition (A1, A2, B1, B2) to a second level factorial design.
>>>
>>> Obviously, this won't work with option 1. However, using option 2
>>> gives some problems in the SPM design: one of the parametric
>>> modulators are zeros (i.e. black in the design). I guess this is due
>>> to the redundancy Rik is talking about in the thread mentioned
>>> above. I tried 3 param mod's (N-1), but I'm not sure which contrasts
>>> to use to capture the effects for each condition: Maybe [0 1 0 0 0],
>>> [0 0 1 0 0 ], [0 0 0 1 0 ] and [1 -1/4 -1/4 -1/4 0] ?
>>> These effects doesn't look quite the same as in the original
>>> categorical design though.
>>>
>>> In short: Can somebody help me to get beta's for each seperate
>>> condition (A1, A2, B1, B2) in a parametrical design, reflecting the
>>> same effects as if I used a categorical design with A1, A2, B1, B2
>>> as seperate regressors?
>>>
>>> Thanks,
>>>
>>> Martijn
>>>
>>
>>
>> --
>>
>> -------------------------------------------------------
>>
>> DR RICHARD HENSON
>> MRC Cognition & Brain Sciences Unit
>> 15 Chaucer Road
>> Cambridge, CB2 7EF
>> England
>> EMAIL: [log in to unmask]
>> URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal
>>
>> TEL +44 (0)1223 355 294 x522
>> FAX +44 (0)1223 359 062
>> MOB +44 (0)794 1377 345
>>
>> -------------------------------------------------------
>
--
-------------------------------------------------------
DR RICHARD HENSON
MRC Cognition & Brain Sciences Unit
15 Chaucer Road
Cambridge, CB2 7EF
England
EMAIL: [log in to unmask]
URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal
TEL +44 (0)1223 355 294 x522
FAX +44 (0)1223 359 062
MOB +44 (0)794 1377 345
-------------------------------------------------------
|