Designing Elicitor: Software to Graphically Elicit Expert Priors for Logistic Regression Models in Ecology. 2006

ELICITOR is graphical elicitation software created to elicit normal prior distributions for a Bayesian logistic regression model. Motivated by a real need to include expert knowledge in presence–absence models in ecology, this research describes a synthesis of theory from statistics, psychology and ecology. The aim was to build elicitation software that would be user friendly to environmental scientists, aiding the formal quantification of knowledge about species’ distributions and providing informative priors for a Bayesian logistic regression model. The software was written as an add-on to winBUGS, and automatically generates appropriate code so that both the model can be taken from prior to posterior within the one program. The software and expert elicitation process is demonstrated with a case study on the Brush Tailed Rock Wallaby Petrogale penicillata.