This topic has been discussed on numerous occasions on the listserv,
but I'm still unsure about a few things. When extracting a VOI
timecourse (e.g., for a PPI analysis):
1. Can adjusting for different contrasts produce large differences in
results? In other words, is the decision important? And if so, why
don't folks typically report this in empirical articles using PPI?
2. Is there ever a case in which it is desirable to adjust for a
t-test contrast (e.g., the comparison of only two conditions)? In
spm_regions the default is to only allow adjustment for F-contrasts,
but of course adjustment can be made for t-contrasts if desired.
3. My understanding is that the F-test omnibus (coding for all effects
of interest) is the favored option for adjustment. But there is some
ambiguity in what "all effects of interest" refers to. Should all
effects expected to relate to neural activity be included (so that
only motion regressors/constants are removed), or only effects "of
interest"? For example, if I've modeled response time and skipped
trials as covariates of no interest, should these be included in the
omnibus (because they are expected to relate to neural activity, not
to artifactual changes in signal intensity)?
Many thanks in advance for any tips.
Department of Psychology
University of California, Los Angeles