Most of the hypothesis testing procedures we have investigated so far depend on some assumption about the underlying distribution of the data, the normality of the data, for example. Distribution-free methods, while not completely "assumption-less" relax some of those distributional assumptions. In this lesson, we'll investigate three such hypothesis tests:
- the sign test for a median
- the signed rank test for a median
- the Wilcoxon test for testing the equality of two distributions
In each case, the assumptions made are much less stringent than those made when we learned how to conduct hypothesis tests about population parameters, such as the mean \(\mu\) and proportion p.