Time to leave hypothesis testing and parameter-based methods behind, and move to what we’ve always wanted. Which is a proper understanding of uncertainty.
I’ve uploaded some new software, thus far only based on MCMCpack. But we have hopes of further extensions once more get involved.
All people want, but no classic method delivers, is
Pr( Y | new X, old X&Y, Model & Assumptions)
where Y is the proposition of interest, and the rest is obvious (or not!). Hypothesis testing and parameter estimates (including confidence/credible intervals) do NOT give you this, nor anything like it.
The software is available at this link.
So far, all that’s up is the software itself and a general introduction. Beginning tomorrow (and roughly two times a week) there will be explanatory articles at the same link which work through the various model types (the examples code at the bottom of the first article).
Guess you could call this tooting my own horn. But I’ve never mastered the embochure. In any case, it’s well past the time to eschew p-values and the like.