If inferential statistics allows us to use samples to make generalizations about the population the samples represent, then how can Nonparametric statistics (which do not rely on assumptions about the population) be a type of interferential statistics?
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.