Negative pD is a serious problem. This means that you have overdisersion
in the sampler and essentially you cannot use
WinBUGS DIC for meaningful comparisons.
I would suggest either :
use the pD computed from sd(Deviance )/2 as is proposed by Gelman OR use a
predictive measure such as MSPE. However MSPE is not adjusted for
parameterization
Another possibility is that you should reparameterize the model
best
Andrew Lawson
On 1/19/2012 2:19 AM, 近藤文代 wrote:
> Hi, everyone
>
> I have a problem. I got a minus pD in a Bayesian hierarchical model. This
> means that Dhat is greater than Dbar.
>
> Is there anybody who has an opinion what I should do?
>
>
> Fumiyo Kondo
> Division of Policy and Planning Sciences
> Faculty of Engineering, Information and Systems
> University of Tsukuba
> Tel & Fax: 029 (853) 5223
> Email: [log in to unmask]
> http://shakosv.sk.tsukuba.ac.jp/~kondo/
>
> -------------------------------------------------------------------
> 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
-------------------------------------------------------------------
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
|