Influence of philosophical concepts of causality on causal modelling in statistical research

Abstract

In this paper causality is seen from a pluralist point of view: About its physical reality one can not claim to know anything as it is seen to be not a matter of fact but rather a matter of human perception. Different philosophical concepts of causality are presented and their philosophical impact and their limitations are sketched. They are shown to underlie different statistical causal modelling. For Pearl's (1995) causal graphical models, the decision theoretic causal graphical models of Heckermann and Shachter (1995) and the potential-response model of Rubin(1974) it is indicated how the advantages and disadvantages of their underlying causal concepts influence their applicability in certain scientific contexts and their acceptancy by individual researchers.