ABSTRACT
In this chapter we will examine the problem of reliably inferring causal relations from statistical data and fragmentary background knowledge. Such causal inference problems arise in many instances in statistics, sociology, economics and epidemiology, among other areas. The problem can also arise when building expert systems that use Bayes networks. In many cases such networks are constructed on the basis of some expert's background knowledge; in many other cases, however, our background knowledge is woefully inadequate for constructing a useful expert system.
