Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[no subject]

"If you are a Bayesian you can never check your prior assumptions,
however much data you collect. I don't like this; I think it is
unscientific. In science you try something and if it doesn't work very
well you try something else. Bayesians do in practice change their
minds but in principle they shouldn't. It is necessary to define a
prior probability for every parameter in the model. If this is done
subjectively then the rest of the analysis has this subjective
element. Occasionally it may be possible to find an objective
definition, which others can agree on, but this is rare. A lot of
people don't realise that you don't just fit models by plonking the
thing down, turning the handle, getting the estimates -- you then look
and see if the fit is internally consistent, and if it isn't, you go
back and do something else. So I don't like the rigidity of

"[Youngjo] Lee and I have introduced likelihood methods for dealing
with many models that are often treated by Bayesian methods. These
have an objectivity that Bayesian methods lack. Our likelihood models
have the property that all aspects of them can be checked, given
enough data. Bayesian models lack this desirable property". (p77)

Joyce H (2005) "A Life in Statistics: Synthesis and Quality",
_Significance_ 2(2): 75-7.

Clive Nicholas

[Please DO NOT mail me personally here, but at
<>. Please respond to contributions I make in
a list thread here. Thanks!]

"My colleagues in the social sciences talk a great deal about
methodology. I prefer to call it style." -- Freeman J. Dyson

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index