Nonparametric Tests

Nonparametric tests are relatively simple to calculate. Their speed and convenience offers a distinct advantage over the parametric alternatives discussed in the previous chapters. Therefore, as investigators we can use these procedures as a quick method for evaluating data. Use of Nonparametric Tests Nonparametric tests usually involve ranking or categorizing the data and by doing so we decrease the accuracy of our information (changing from the raw data to a relative ranking). We may obscure the true differences and make it difficult to identify differences that are significant. In other words, nonparametric tests require differences to be larger if they are to be found significant. We increase the risk that we will accept a false null hypothesis (Type II error). It may be to the researcher’s advantage to tolerate minor doubts about normality and homogeneity associated with

a given parametric test, rather than to risk the greater error possible with a nonparametric procedure.