 |
 |
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? 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 > > -------------------------------------------------------