 |
 |
Dear Margaret, The var(deviance)/2 formula used by the R2jags package has been around a long time. You can see a discussion of it from 2006 in Andrew Gelman's blog, including a rationale for why it was included in the R2WinBUGS package. http://andrewgelman.com/2006/07/06/number_of_param/ Thanks to Sumio Watanabe, we now know that the correct penalty is based on the sum of the variances of the individual contributions to the deviance and not the variance of the sum. This gives the "widely applicable information criterion", or WAIC. So the formula used in the R2jags package is "close, but no cigar". I wrote JAGS but not the R2jags package. I can't take any credit for R2jags so please be careful with your attributions. The native interface between R and JAGS is the rjags package and this gives you access to DIC statistics via the dic.samples() function. At the risk of muddying the waters further, this uses a third definition of pD which, nevertheless, has some justification. See the references on the help page for dic.samples. regards Martyn On Thu, 2015-09-03 at 15:41 +1000, Margaret Donald wrote: > Dear Gibbs samplers, > > > Does anyone have a reference which gives the reason for the definition > of pD in JAGS as deviance/2, versus the Spiegelhalter et al (2002) > definition as pD= Bbar-Dhat. > > > I have been struggling to initialize 3 chains in Brugs and have been > working in R2jags where parameter initilization ise easier, but I am > surprised and not happy with JAGS' pD (different) definition.. So > far I have not discovered the rationale... > > > Some thoughts would be most welcome. > > > Regards, > Margaret Donald > > > -- > Margaret Donald > Post Doctoral researcher > University of New South Wales > [log in to unmask] > 0405 834 550 ----------------------------------------------------------------------- This message and its attachments are strictly confidential. If you are not the intended recipient of this message, please immediately notify the sender and delete it. Since its integrity cannot be guaranteed, its content cannot involve the sender's responsibility. Any misuse, any disclosure or publication of its content, either whole or partial, is prohibited, exception made of formally approved use ----------------------------------------------------------------------- ------------------------------------------------------------------- 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