Suppose a smaller pool of subjects is available for a VBM/VBD study than one
would normally need for a two-groups t-contrast to be significant. Is it
possible to run a different sort of analysis in which one compares one
single subject against the rest of the group?
If 'yes', how would the statistical analysis be done in such a case? Our
common sense feeling is that one should not attempt the same two-group
procedure (with a design matrix having n_1 = 1 and n_2 = n-1, n being the
total group size).