ABSTRACT
The preceding chapters have dealt with procedures that are sometimes referred to as classical statistics. In each case an underlying distribution was known (or assumed) and the requisite population parameters were known or estimated. There is a growing body of literature commonly referred to as nonparametric statistics. Some authors prefer the term distribution-free statistics. Currently, the two terms are often used interchangeably. The important aspect is that some of the restrictive assumptions are relaxed. In actuality, some of these tests deal with parameters without specifying an underlying distribution. Consequently, they should be classed as distribution-free rather than nonparametric. However, such fine distinctions of definition are not important in practice.
