In social science research, sampling is virtually always done "without" replacement; that is, after a particular unit has been selected for observation, it is not put back into the pool so it can be selected again.