In my mind we could use MCMC to see reasonable acceptable results for y given the same x.

To that end, how could we use our chain results to produce a sequence of y? Or would we need to edit the functions to produce a chain of y values?

Please correct me if I have got this wrong!

]]>a) In general, a null model performs a significance test. And I would say that, although one can use NHST / MS / Parameter for asking whether a process is there, they are still asking this question in a slightly different way.

b) The second point is about the data-generating model. The randomization null models that are used in community analysis don’t really have an explicit data-generating model. Do we care? No idea, but there is a difference I feel to the parametric model that have a data-generating model.

]]>