Dear FSL Experts,
I am reanalyzing a dataset and am hoping you can clear up a question I have about setting up the GLM. The study involves participants seeing three types of images that have been rated on a scale of 1-50. The ratings have a bipolar distribution (they are usually very high or very low with a few exceptions), so I originally looked at differences in BOLD response to "highly rated" vs. "low rated" instances of each image type. In other words, I used a categorical approach (3x2 design with 3 types of images and 2 rating levels - high vs low - for each image).
Now, I am interested in using a parametric approach and modeling the rating variable as continuous. So, I think I'll need to create 3 EVs -- one for each image type -- containing not 1s but rather the rating that each image received. My question is: should I ALSO include additional EVs that model image type only (with a 1 whenever the image type appears and 0s otherwise)? Or should I only use the continuous EVs?
It is the difference between this (continuous EVs only):
4 0 0
0 3 0
0 0 2
5 0 0
9 0 0
0 0 7
0 0 6
and this (a categorical EV and a continuous EV for each of the three image types):
1 0 0 4 0 0
0 1 0 0 3 0
0 0 1 0 0 2
1 0 0 5 0 0
1 0 0 9 0 0
0 0 1 0 0 7
0 0 1 0 0 6
I'm having a hard time wrapping my head around what each of these models would represent. I do know that I'll need to demean my continuous variable, even though I haven't done that here.
Thanks very much,
Lauren
|