I don't believe OP is trying to conclude 2 distributions are different, but that one is similar to another. The way I learned such goodness of fit tests, the null hypothesis always assumes equality. You do not have a choice on that. There is a separate section in the wikipedia article which states how to construct a confidence interval, but I cannot quite understand how to apply it. "Setting_confidence_limits_for_the_shape_of_a_distribution_function"
On Thu, Jul 16, 2020 at 2:56 PM Raul Miller <[email protected]> wrote: > The distributions are not equal. > > The ?n$q distribution lacks any elements which exceed q (or are equal > to q) while the ?m$p distribution would contain such values. > > Thanks, > > -- > Raul > > > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
