Inferences about the efficacy of treatment are weakened when large differences exist between groups.  The Impact of this bias increases when sample size is small.

Statistical techniques (controlling for differences) can allow adjustment of study results to account for such differences.