ELICITOR addon: eliciting expert priors for Generalised Linear Models

Mary Kynn (Please address all correspondence, including bug reports, to Mary, not the WinBUGS team.)

Elicitor was originally designed to elicit normal priors for logistic regression presence/absence models in ecology. It has now been extended and should allow you to elicit a more flexible generalised linear model (with independent priors). The links that are programmed in already are the identity, log and logit; and the priors that are programmed in are normal and laplace (double exponential). The modules for writing new link and prior functions are open source and you are welcome to play with them (you’ll need BlackBox Component Pascal, which is free, but see the user manual or email me for more information). In any case, I haven’t had time to test extensively or update more than the basics in the user manual – so if you are serious about doing some elicitation you are very welcome to contact me for more help.

For a quick overview, see my poster from the 2006 Valencia Meeting.

For a more detailed description read this (unpublished) paper on Designing Elicitor (figure).

Download the add-on here: elicitor_v35_addon.txt. You need to have either WinBUGS or BlackBox already on your computer to do this. I no longer compile standalone executables (easier to patch the add-on version). Instructions for decoding are given at the top of the file.

Please EMAIL ME and let me know if you find it useful, or if you need to ask for help.

Any bugs should be reported to Mary Kynn (NOT the WinBUGS team).

Site maintained by:
Dave Lunn
Dept. Epidemiology & Public Health,
Imperial College School of Medicine,
Norfolk Place,
London W2 1PG