Section 6.3: Non-parametric Tests
Recall that most of the statistical tests that we have discussed so far have several assumptions. Most common among these are the assumption that the data are normally distributed, the assumption of homogeneity of variance, and the assumption that the data are measured at least on the interval level. Because a normal distribution of the data is not an assumption, non-parametric tests are often referred to as distribution-free tests.
Parametric tests are so called because they make assumptions about population parameters; non-parametric tests are best used when those assumptions are violated.
Non-parametric tests are encountered in the social science literature much less frequently than their parametric counterparts. This is because the parametric statistics that we previously discussed are more powerful. That is, everything else being equal, they are more likely to allow the researcher to reject a false null hypothesis. Thus, non-parametric tests are usually only used when the assumptions of the appropriate parametric tests are violated to such a degree that the results would be erroneous or misleading.
Last Modified: 02/12/2019