Statistical Analysis V: Non-parametric Statistics

Abstract

The NPAR procedure allows the user to carry out a variety of non-parametric statistical tests. Unlike parametric statistics (including the t-test, anova, the product-moment correlation, etc.) non-parametric statistics make few assumptions about the nature of the data. They may be particularly suitable where there are relatively few cases in the population we wish to examine, or where a frequency distribution is badly skewed (rather than symmetrical). Whereas a procedure like ANOVA demands that the dependent variable be truly numerical, non-parametric tests can handle data at other levels of measurement. In particular they allow us to analyze nominal and ordinal data.