Overview
Steven M. Lewis is a statistician (University of Washington) who collaborated with Adrian Raftery on Bayesian model comparison and MCMC methodology, and on their application to demographic event-history data.
Key Contributions / Features
- Laplace-Metropolis estimator (Lewis-Raftery 1997): marginal likelihoods / Bayes factors from posterior-simulation output without extra runs.
- MCMC convergence diagnostic (Raftery-Lewis 1992): the run-length diagnostic based on a two-state Markov chain for a posterior quantile.
- Event-history models with unobserved heterogeneity (Lewis-Raftery 1999): Bayesian discrete-time survival with frailty and Compound Laplace-Metropolis Bayes factors, applied to Iran's fertility decline.
Related