84: 2 April 2001
Probability as a Guide to Life

Advisory Editors: Henry Kyburg, Rochester, and Mariam Thalos, Buffalo

Fifty percent of all college graduates, but only twenty percent of all churchgoers, drink alcohol. What is the probability that Alma, a church-going college graduate, drinks alcohol? There is, as yet, no general solution to the problem of the reference class, of which Alma's case is an instance. The problem is pervasive in the application of statistical knowledge. It is philosophical in nature: what grounds inference to single-case probabilities from knowledge about the classes in which they fall?

According to Richard von Mises, a founder of the philosophical study of probability, the very question is nonsense: probabilities apply only to classes. But the probabilities of classes are, in themselves, of no help in decision and action, for it is single cases we face, and we face them one at a time.

Will recent developments in the area of probabilistic causality help solve this problem? Or is the relationship between the reference class problem and causality entirely superficial? Is the problem purely a problem for epistemology, as Hans Reichenbach suggested? Does the Bayesian treat the reference class problem in a distinctive way? Can analysis of Simpson's paradox throw light on the problem? The present issue of The Monist will be devoted to these and related issues in the foundations of probability. Contributors will include Nancy Cartwright, Wesley Salmon, Clark Glymour and Deborah Mayo.

Table of Contents:

Henry E. Kyburg, Jr. 

Probability as a Guide to Life

Timothy McGrew 

Direct Inference and the Problem of Induction

Isaac Levi 

Objective Modality and Direct Inference

Wesley Salmon 

Explaining Things Probabilistically

Christopher Hitchcock 

Causal Generalization and Good Advice

Nancy Cartwright 

What is Wrong with Bayes Nets?

Gary Malinas 

Simpson's Paradox: A Logically Benign, Empirically Treacherous Hydra

Clark Glymour 

Instrumental Probability