Hi, it certainly does not violate the assumptions of the t-test. However when those assumptions do not hold or hold approximately the statistic can be more vulnerable when the design is very unbalanced AND the sample sizes (one or both) are very small. You can see a short discussion here<http://www.basic.northwestern.edu/statguidefiles/ttest_unpaired_ass_viol.html#Special%20problems%20with%20imbalanced%20sample%20sizes>. Bw, Enzo On Mon, Mar 19, 2012 at 7:43 PM, Steve Masson <[log in to unmask]> wrote: > Dear SPM users, > > Is it statistically valid to compare two groups of very unequal size (eg, > group 1 = 12 subjects and group 2 = 24 subjects) with a two-sample t-test > in a second-level analysis in SPM? > > Thanks! > > Steve >