Likelihood principle: difference between weak and strong version

Question

Does anyone understand the difference between weak likelihood principle and strong likelihood principle?

The hits in [Google searches for "weak likelihood principle"](https://www.google.com/search?q="weak+likelihood+principle") show that the answer is yes. — whuber, Jan 15 '14 at 18:54
... then the answer may be more useful to you (& perhaps to others, perhaps even to the answerer) than one that trots out the usual explanation. — Scortchi - Reinstate Monica, Jan 15 '14 at 19:04

score 2 · Answer 1 · answered Sep 10 '14 at 02:26

2

Yes, it's straightforward. The weak LP is within a given model and distribution (it's essentially just the sufficiency principle within a model), whereas the strong LP, SLP, makes the claim (of equivalent evidential import) for pairs of models. Easiest to refer to my paper here: http://errorstatistics.com/2014/09/06/statistical-science-the-likelihood-principle-issue-is-out

fortunately, it turns out the the SLP does not follow from sufficiency and conditionality, as had long been thought.

answered Sep 10 '14 at 02:26

user55501

4

Could you please also give a complete reference? Links have a habit of dying. – Glen_b Sep 10 '14 at 02:42
1

You say that the SLP does not follow from sufficiency and conditionality, but in fact Mayo's disproof of Birnbaum's proof has been seriously challenged and may well be flawed. https://arxiv.org/abs/1711.08093 – Michael Lew Mar 22 '18 at 23:52

1 Answers1