What are the chances?
  An explication of single-case chance

Project funded by NWO, the Dutch organisation for scientific research
University of Groningen, The Netherlands, 2011-2016

INTRODUCTION

The concept of probability is of utmost importance to science: many scientific claims are expressed in terms of probability and uncertainty, and almost all methods in the sciences rely on probability and statistics. In many of these applications, the probabilities express chances: they do not express a personal opinion or a degree of belief about events in the world, but rather an objective characteristic of the events themselves. But what are these chances in the world? How do they hang together with the, often deterministic, mechanisms underlying the events? And can we indeed determine these chances by applying statistical methods?

This research aims to explicate chance as a characteristic of events. The focal point for the project is the so-called reference class problem. For instance, if smokers have a chance of 10% to live over 70 years, and vegetarians have a chance of 80%, then what are the chances for a smoking vegetarian? The key idea of the proposal is that only the chances associated with irreducible, or in other words unrefinable, descriptions are correct for the events. The resulting concept of chance provides a basis for calling particular ascriptions of chance correct, even if other descriptions lead to different chances for the event, or to complete determination.

The project throws new light on a long-standing philosophical debate. Moreover, the new concept of chance will motivate us to reconsider and reappraise various statistical methods. It will deepen our understanding of statistical model selection and the nature of simplicity, trading on the idea that model selection guides us to the correct chances. Furthermore it will clarify the role of chance in causal modelling, trading on the idea that changes to the chances effected by experiment are in some sense real. This will eventually improve our insight into, and thus the application of, the statistical methods at hand.

PEOPLE

The project employs three people: Jan-Willem Romeijn, Patryk Dziurosz-Serafinowicz, and Ronnie Hermens.

		The leader of the research project is Jan-Willem Romeijn. He is professor on a tenure-track at the Faculty of Philosophy of the University of Groningen, where he teaches philosophy of science. His research interests include scientific method, the foundations of statistics, inductive logic, and causal modelling.
		Patryk Dziurosz-Serafinowicz is a PhD student at the Faculty of Philosophy of the University of Groningen. He holds a PhD in law, MPhil in Philosophy and MPhil in Law, all obtained from the Jagiellonian University in Krakow. His research interests include philosophy of probability, Bayesian epistemology, abductive reasoning, and philosophy of law.
		Ronnie Hermens is a PhD student at the Faculty of Philosophy of the University of Groningen. His research focuses on the nature of probability in theories of physics. Further interests include the philsophy of mathematics and physics.

EVENTS

Events are listed if they are organised by project members or partly funded by the project.

• Workshop on radical uncertainty at the Munich Centre for Mathematical Philosophy.
• Guest lecture on model selection by Kevin Kelly. Slides will be available soon.
• Conference on foundations and applications of model selection entitled All Models are Wrong, in collaboration with Ernst Wit and Edwin van den Heuvel.

RESEARCH PAPERS

For full lists of manuscripts and publications we refer to the websites of the project members. The following is a list of publications directly relevant to the project's topic.

     Jan-Willem Romeijn

• Philosophy of Statistics, lemma in the Stanford Encyclopedia of Philosophy, to appear
• One size does not fit all: derivation of a prior-adapted BIC, in Probabilities, Laws, and Structures, ed. by D. Dieks, W. Gonzales, S. Hartmann, F. Stadler, T. Uebel, and M. Weber. Berlin: Springer, to appear

PRESENTATIONS

For full lists of talks, see the websites of the project members. The following concerns talks with material directly relevant to the project's topic.

     Jan-Willem Romeijn

• "Frequencies, Chances and Undefinable Sets" at the Combined research seminar for the Munich Centre for Mathematical Philosophy and the Statistics Department of the Ludwig Maximilians University, July 2011
• "Specificity, Accommodation and the Sub-family Problem" at the Conference on Novel Predictions, University of Duesseldorf, February 2011