This is slightly off-topic, but probably still the most appropriate forum
to discuss this matter. I calculate DIC as a means of weighing different
models. I find for my purposes, however, that DIC seems to overfit. For
instance, I have several models (all a priori defined) which are comprised
of several variables that turn out to have credibility intervals
overlapping zero. The model containing much fewer covariates, each of
which do not overlap zero, is rarely the best model as identified by DIC.
Since I am involved in mapping my model results, a consequence of an
overfitted model is that I map a model which is basically a bunch of random
junk plus a couple informative covariates. It seems the covariates which
overlap zero actually interject noise, obfuscating the signal. Strict
interpretation of information-theoretic approaches seems to dictate that I
include these random signals. I'm tempted, however, to go with the model
whose parameters possess credibility intervals not bounding zero. I'd love
to here participants thoughts on this matter.
I've not seen any literature discussing the propensity of DIC to overfit.
Is this a generally observed phenomena?
Thanks,
Wayne Thogmartin
Wayne E. Thogmartin, PhD
USGS Upper Midwest Environmental Sciences Center
2630 Fanta Reed Road
La Crosse, WI 54603
608.781.6309
[log in to unmask]
-------------------------------------------------------------------
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
|