Explanation and Social Scientific Modeling

Abstract

My dissertation examines how different conceptions of explanation and requirements of explainability influence modeling in the social sciences and computer science and the use of models within institutions. It does so using three case studies: revealed preference approaches in microeconomics and sociology, market design in welfare economics, and interpretable algorithms in computer science. Chapters 1 and 2 of my dissertation examine how different background explanatory commitments influence descriptive modeling in the social sciences, through the case study of revealed preference approaches. I defend a revealed preference interpretation of the concept of a preference, on which the concept of a preference is defined in entirely behavioral terms. Chapters 3 and 4 of my dissertation examine how requirements of explainability influence modeling in the social sciences and computer science and constrain the use of models within institutions, through the case studies of market design and interpretable algorithms.