Steele, Russell

Research Interests

Currently I am working on Bayesian approaches to mixture models and multiple imputation. Some of my work with mixtures suggests using importance sampling methods to calculate Bayes factors for determining the number of components. Another topic I’m interested in is the effect of prior specification on selecting the number of components for mixture models, especially for mixtures of multivariate normal distributions. Finally, I’ve done some preliminary work exploring the advantages of using importance sampling for multiple imputation in certain settings involving large amounts of missing data and restricted numbers of copies.

Publications