Bayesian Analysis of Linear Models

Lyle Broemeling is a recently retired Professor of Statistics from the MD Anderson School of Medicine at the University of Texas in Houston. There he worked with other Bayesian statisticians including Don Berry in the application of Bayesian methods in medical research. When this book was published in the mid 1980s he was Professor of Statistics at Oklahoma State University. Bromemeling is an excellent teacher and author and in this book he presents the methodology needed to apply Bayesian methods to the wealth of linear models that are applicable in a wide variety of statistical problems including man problems in the analysis of clinical trials data. The only drawback of the book is that it was written prior to the development of MCMC methods as an approach to producing Bayesian posterior distributions. So at that time for computational reasons conjugate priors were often used to allow closed form solutions. Such restrictions are no longer needed although the new methods require a lot of computing and diagnostic checking. The Bayesian concepts are timeless and this book is as good as any for learning the Bayesian principles and how they can be applied to linear models.

Lyle Broemeling is a recently retired Professor of Statistics from the MD Anderson School of Medicine at the University of Texas in Houston. There he worked with other Bayesian statisticians including Don Berry in the application of Bayesian methods in medical research. When this book was published in the mid 1980s he was Professor of Statistics at Oklahoma State University. Bromemeling is an excellent teacher and author and in this book he presents the methodology needed to apply Bayesian methods to the wealth of linear models that are applicable in a wide variety of statistical problems including man problems in the analysis of clinical trials data. The only drawback of the book is that it was written prior to the development of MCMC methods as an approach to producing Bayesian posterior distributions. So at that time for computational reasons conjugate priors were often used to allow closed form solutions. Such restrictions are no longer needed although the new methods require a lot of computing and diagnostic checking. The Bayesian concepts are timeless and this book is as good as any for learning the Bayesian principles and how they can be applied to linear models.