Next: Solution to some problems
Up: Frequentists and Bayesian `sects'
Previous: Subjective or objective Bayesian
Contents
Finally, I would like to recall that Bayes' theorem is a
very important tool, but it can be used only when
the scheme of prior, likelihood, and final
is set up, and the distributions are properly
normalized.8.21This happens very often
in measurement uncertainty problems, but less frequently in other
applications, such as assessing the probabilities
of hypotheses. When Bayes' theorem
is not applicable,
conclusions may become
strongly dependent on individuals
and the only guidance remains the normative rule of the
hypothetical coherent bet.
Giulio D'Agostini
2003-05-15