Powers, Probability and Statistics

Abstract

Regularity is often taken as the starting point of our causal knowledge. But pure constant conjunctions are not what science finds. Even in randomised controlled trials, we do not discover a regular frequency of occurrence of some effect. The dispositionalist is able to explain the evidence of science in terms of the ontology of real causal powers exhibiting an irreducibly tendential nature: less than necessity but more than pure contingency. Much evidence of this kind has to be understood probabilistically and there is a frequentist interpretation in which the facts of frequency of occurrence fix all the facts of probability. However, the dispositionalist has a stronger propensity interpretation of probability at their disposal in which the facts of probability are determined by the individual powers of things. The dispositional approach allows us to make sense of large-scale population data in which different individuals within the same sub-group can have different probabilities of being affected by a cause. On this view, individual propensities can compose to make an overall chance of an effect for a group. But from a starting point of general facts of probability for groups, we cannot decompose those chances back to individuals.