I know I am likely to re-open a discussion about non-informative priors for
precision parameters, but I would still value your help related to a more
specific problem I have encountered.
I am fitting a hierarchical model in which parameters are distributed
according to a gamma distribution, parameterised by mu and tau as in the
Bugs manual. I need to put some vague priors on mu and tau. I know that the
parameters to be fitted should be positive and not very large but otherwise
I have not other prior belief (or rather I do not want my belief in the
parameters to influence the results).
Choosing:
...
phi1[i, k] ~ dgamma(tau1[k],mu1[k])
...
mu1[k] ~ dgamma(1.0E-3, 1.0E-3)
tau1[k] ~ dgamma(1.0E-3, 1.0E-3)
...
or similar leads to an extremely fat-tailed (and wiggly) distribution of a
mean and variance of a phi1 and convergence problems.
Any clues? Thanks in advance,
Adam Kleczkowski
--
Adam Kleczkowski
[log in to unmask] (work), [log in to unmask] (teaching)
[log in to unmask] (private)
http://mathbio.com (work), http://kleczkowski.net (private)
-------------------------------------------------------------------
This list is for discussion of modelling issues and the BUGS software.
For help with crashes and error messages, first mail [log in to unmask]
To mail the BUGS list, mail to [log in to unmask]
Before mailing, please check the archive at www.jiscmail.ac.uk/lists/bugs.html
Please do not mail attachments to the list.
To leave the BUGS list, send LEAVE BUGS to [log in to unmask]
If this fails, mail [log in to unmask], NOT the whole list
|