Nonparametric statistics does not mean you are testing every member of a population. It just means you are not assuming the population as a whole is normally distributed, as you do for parametric statistics. You are still using samples to draw conclusions about the larger population.
