ABSTRACT
All the topics that we have studied up to now are considered parametric methods of statistical inference. This is because many of these techniques rely on a specific collection of model assumptions. For instance, you may recall that in order to make reasonable inferences from an analysis of variance, we are assuming that the error component is normally distributed, each factor level has constant variance, and the observations are independent of each other. Also, in many of the analyses we have seen thus far, the presence of outliers can often have a profound impact on any inferences that we make.
