Probabilistic Explanation

Conceptual Overview and Discussion

Probability has become an increasingly important way of thinking about knowledge that is often uncertain and incomplete. Probabilistic explanation attributes to probability a helpful role by ascribing either a numerical value or a degree of credence to the likelihood of random events. It produces helpful causal descriptions of indeterminate phenomena by examining the relationship between cause and chance. Probabilistic explanation allows us to clarify why a random event happened and attempts to make sense of potential events by unifying diverse phenomena through limited patterns and principles. This form of explanation raises questions about our ability to reliably know what is uncertain and brings attention to the inherent limitations associated with case study research and chance phenomena.

There is ongoing debate with regard to the nature of probability. Views are sometimes radically varied and opposed to one another, for example, with the separation of radical belief and radical frequency theorists. Today, probability is most often associated with quantitative expressions and has become central to modern notions of scientific evidence. It has proven to be especially useful as an established field of applied mathematics, yet disciplines as diverse as medicine, biology, and politics have made use of probability and statistical analysis in order to deduce and understand causal relationships.

Causal inference is an important form of reasoning in case study research. Probabilistic explanation has become almost synonymous with causation, because it allows us to speak meaningfully about uncertain causal claims. With it, we may offer general accounts of causation in terms of contributing causes and factors, exceptions, and variations (e.g., “X will probably cause Y in most cases”) rather than having to either affirm or deny a specific and fixed determining cause. This is important, because it allows us to make [Page 730]informed decisions when we are researching either irreducibly complex systems for which there are multiple casual factors (when phenomena are overdetermined) or incomplete situations for which we have limited sets of case study data (when phenomena are underdetermined).