ARNOLD ZELLNER: SCIENTIST, LEADER, MENTOR, AND FRIEND

摘要

This Special Issue is a tribute to Arnold Zellner as a scientist, as a leader in statistics and econometrics, as a mentor for young researchers, and as a wonderful friend of many. The idea of this issue was conceived in October 2010. The theme was clear: Bayesian Inference and Information, topics about which Arnold was quite passionate. Arnold was attracted by Jeffreys' work to the Bayesian paradigm for learning from data for inference about unknown quantities such as unobservable parameters and unobserved outcomes. He viewed Bayesian paradigm more broadly and Bayes rule as a learning mechanism that enables one to learn from data through the likelihood model when a prior distribution for the model's parameters can be formulated. But desire to learn from data does not necessarily endow one with a complete prior distribution that represents predata information, nor a likelihood model that represents data information. Zellner (1971) suggests the use of Jeffreys' prior "to represent knowing little about the value" of a parameter.