Robert K. Tsutakawa

Overview

Robert K. Tsutakawa was a statistician at the University of Missouri, Columbia. His research focused on Bayesian hierarchical models for geographic and disease data, including Generalized Linear Mixed Models (GLMMs) for spatially structured count observations.

Key Contributions

• Tsutakawa (1988): Developed a Bayesian hierarchical model for estimating geographic mortality rates. The observation model is Binomial, with log-odds modelled as a linear function of covariates plus region-specific random effects with a conjugate gamma-mixture prior. Estimation proceeds via an Empirical Bayes (EB) iterative algorithm; applied to Missouri county-level stomach-cancer mortality 1972–1981. See Conditional Autoregressive Model and Generalized Linear Mixed Model.
• Sun-Tsutakawa-Speckman (1999): Co-established necessary and sufficient conditions (Theorem 2: rank condition $\mathrm{rank}(X_2'R_1X_2+B)=q$; Theorem 3: necessity; Theorem 4: GLMM extension) for the posterior to be proper in hierarchical models with Conditional Autoregressive (CAR) spatial random effects — extending Hobert-Casella (1996) to the singular precision-matrix case that arises with the standard Besag-York-Mollié intrinsic CAR prior.

Related

• Dongchu Sun
• Conditional Autoregressive Model
• Bayesian Hierarchical Model
• Generalized Linear Mixed Model
• Tsutakawa (1988)
• Sun-Tsutakawa-Speckman (1999)