ABSTRACT
THIS CHAPTER deals with (as the title suggests) nonparametric tests. Before we become involved in non parametric tests, it might be well to consider first, what is a parametric test? A parametric test is one that is based on a parameter, also known as “some descriptive statistic about a population.” The most commonly used (and useful) parameter is a mean-drawn from a population that we assume is normal. The standard deviation, close friends with the mean, is another commonly used parameter. Although some operations are based on other assumptions or parameters (e.g., binomial or Poisson distributions of data), the Nonparametric Tests procedure deals primarily with populations that are neither normally distributed nor based on continuous data (so means don’t mean anything) and considers how to conduct statistical tests if the assumption of normality is violated. For example, if you have 99 people that are listed in rank order, the mean rank will be 50. But that doesn’t tell you anything, except that half-way through the list of ranks is the number 50. So, if you want to do statistics on ranks, you need to use nonparametric procedures.
