Dear SPM experts,
I am interested in the parametric modulation effect associated with a subjective rating on a 0-10 scale, as well as in the effect of the difference between this subjective rating and a fixed intensity value of the stimulus on which the subjective evaluation was made (these parametric modulators refer to specific task phases, of course). Naturally, participants differed in both the absolute values they used within the scale, as well as in the range of possible values they exploited in their respones - so the parmetric regressors differ between participants. I would also like to associate a between-subject covariate (anxiety score) with the parametric modulation effects. I was wondering about the following issues:
1. Should the subjective rating values be re-scaled? the absolute scores made by participants are meaningful, if this might affect this descision.
2. Should I orthogonalize the parametric modulators?
3. Should I orthogonalize the covariate?
4. Is a mixed effects model, or perhaps the sandwich toolbox, more suitable in this case?
Looking forward to your advice,