Dissertation Research: Methods and Causes in Social Science

点击翻译按钮获取中文摘要

This Science and Technology Studies Dissertation Improvement Grant explores the two most prominent methods of causal inference in social science, the inductive and deductive approaches. It follows John Stuart Mill in focusing on causal tendencies, as knowledge of such tendencies is necessary for social policy and planning. On the inductive side, a recently proposed qualitative comparative method used in sociology will be critically analyzed. This method borrows certain basic principles of Boolean algebra in order to identify the necessary and sufficient conditions for the occurrence of a social phenomenon. Since the problem of spurious correlation is the main obstacle to the successful application of this method, several solutions to this problem drawn from the recent philosophical literature on causality will be explored. On the deductive side, the project will study the use of rational choice modeling for inferring causes in economics and political science. This method is often claimed to provide social scientists with knowledge of the most general and important causal tendencies in the social world, namely those that result from rational strategic action. Yet this requires an explanation of how models with highly restrictive assumptions can actually describe real-world causes. Such an explanation, the project argues, is in practice often absent. One of the most successful applications of rational choice theory is generally agreed to be the recent use of game theory in the design of spectrum auctions. So the project will study in detail how robust causal tendencies are obtained from game theoretical models in this case, by visiting the experimental laboratories used to test the models and interviewing the auctions. designers. The authors will draw conclusions from this for the general prospects of rational choice modeling in social science. The issues addressed in this project carry both practical and theoretical importance. The practical significance of formulating a normative account of causal inference in social science is indubitable. Causal claims made by social scientists often enter into some of the most important policy decisions and figure widely in debates about their appropriateness. Making explicit the methodological presuppositions of different inductive and deductive approaches and critically analyzing them, will help to lift the curtain on the realities of predicting social trends and defending certain policy measures, and hence give us more rigorous tools with which to evaluate social science and its applications. For example, the so-called rational choice revolution is said to have made social sciences more scientific. Yet its characteristic formal rigor and mathematical expression cannot by themselves improve our ability to study causes. It is thus important socially and politically to understand both the potential and the limitations of this paradigm. The theoretical importance of studying how social science does, can and should infer causal tendencies from empirical data and theoretical models also cannot be overemphasized. In natural sciences such as physics the task of causal inference has been closely connected with the practice of experimentation. Since the latter is not nearly as pervasive and feasible in the context of social research, the study of causal laws and mechanisms in social sciences raises the question of the unity of method in science. This project will directly contribute to the question of whether the social sciences should strive to develop distinct methods to study the social world, or instead attempt to imitate the controlled experiments of the natural sciences. Despite their prominence in social science, neither the deductive nor inductive methods of causal inference explored in this project have received sufficient attention from philosophers of science. By correcting that oversight, this project will move the intellectual agenda of philosophy of science closer to the actual concerns of social scientists.