21. Causal Inference in Social Policy Analysis  

Thomas Biegert (London School of Economics and Political Science)

Elias Naumann (GESIS Leibiz Institute for the Social Sciences & University of Mannheim) 

Social policy analysis at heart is interested in identifying causal relationships. We are interested in the reasons for policy changes, we are interested in policy effects, we are interested in why some policies are more popular than others. Yet, social policy analysis has been struggling with the establishment of causal relationships because it has to deal with a special set of problems. We often encounter issues such as collinearity, multiple alternative explanations, and limited variation in our explanatory variables as a consequence of the country-comparative setup of our research. Over the past decades, the social sciences have witnessed a surge in studies aiming to isolate causal effects. By applying the so-called potential outcomes framework of causality, this wave of research does not overly emphasize advanced econometric models but puts the focus on research design. Based on the "gold standard" of randomized experiments it brings along a distinct way of thinking about how to set up studies and how we can identify causal relationships. Moreover, with the emergence of further practical tools and methods, such as directed acyclic graphs, and better data availability, more and more analyses of social policies have come to tackle these issues and are able to identify causal relationships in our research field. This stream aims to foster exchange between researchers in comparative social policy analysis who put a special focus on causal analysis. In particular, we invite contributions relying on natural or quasi-experiments, survey experiments, matching, instrumental variables, fixed effects panel designs, difference-in-differences-approaches, and regression discontinuity designs. 

The stream invites paper proposals from all fields of social policy research. Paper proposals should include the research question, theoretical background, and (first) results, but also provide specific detail on the analytical approach taken to establish causality. The stream is also open to contributions that aim to present research proposals and discuss research designs.