Bonjour,
Je voulais aussi vous signaler une conférence très intéressante qui aura lieu le jeudi 19 février 2015 à 12:00 à la salle Extranef 126, donnée par Stephen Senn (Competence Center in Methodology and Statistics, Luxembourg Institute of Health) (voir http://www.hec.unil.ch/ob/seminars/brownbag?session=1669 et ci-dessous)
Valentin
You May Believe You Are a Bayesian But You Are Probably Wrong
Stephen Senn (Competence Center in Methodology and Statistics, Luxembourg Institute of Health)
19 février 2015 - 12:00-13:00, salle Extranef 126
I provide brief outlines of what George Barnard considered were the four great systems of statistical inference, which might be labelled Fisherian, automatic Bayesian, subjective Bayesian and Neyman-Pearson.
At the danger of overlooking many subtleties, these can be thought of in terms of the four combinations of two factors at two levels. The first factor is fundamental purpose (decision or inference) and the second factor probability argument (direct or inverse). Of these four systems the ‘fully Bayesian’ approach of decision-making particularly associated with Ramsay, De Finetti, Savage and Lindley and using not only inverse probability but also utility and personal belief has some claims to be the most impressive. I argue, however, and illustrate by example, that this approach seems to be impossible to follow.
I speculate that there may be some advantage to the practising statistician of following George Barnard’s advice of being familiar with all four systems. This is very much in agreement with Gigerenzer’s and Marewski’s recent plea that a narrow automation of statistical inference should be avoided.
