Statistics for People Who (Think They) Hate Statistics

by Neil J. Salkind

On the other hand, you may very well find yourself dealing with samples that are very small (or at least fewer than 30 cases) or data that violate some of the important assumptions underlying parametric tests.