No in the null hypothesis, for the distribution, don't have to care, can be distribution free, except assume that the expected values on the measurements are th

No in the null hypothesis, for the distribution, don't have to care, can be distribution free, except assume that the expected values on the measurements are the same. For details, see my post https://news.ycombinator.com/item?id=10484602 https://news.ycombinator.com/item?id=10484602