Call for Applications
Causality lies at the heart of any scientific explanation: Understanding how an intervention or treatment did or would change the course of the world or how a particular state of the world came about, requires an understanding of cause and effect. The currently dominant perspective of counterfactual causality was subject of two workshops funded by the Akademie für Soziologie in 2019 and 2020. The workshop “Causality in the Social Sciences III – Heterogeneous Causal Effects” builds on both by addressing a specific topic in causal inference that is nevertheless fundamental to all social sciences, namely effect heterogeneity.
Effect heterogeneity is present whenever individual-level causal effects deviate from the average effect, for example, when individuals—or any other unit of observation—differ in their response to an intervention or treatment. Across different social science disciplines, researchers have concluded that ignoring existing effect heterogeneity can lead to incorrect conclusions when testing hypotheses and to inefficient or even harmful policy recommendations (Bolger et al. 2019; Imai and Strauss 2011; Morgan 2001; Angrist 2004; Heckman, Smith, and Clements 1997). On the other hand, incorrectly specified models—e.g., models that do not fully account for selection bias—may detect heterogeneous effects when the true effect is homogeneous (Breen, Choi, and Holm 2015).
Traditionally, social scientists have been tackling effect heterogeneity mainly through interaction terms. While this simple approach can be effective, it is not without limitations and pitfalls (e.g., Mize 2019). Recently, technically more advanced and substantively different approaches—e.g., parametric and nonparametric techniques based on propensity scores, quantile regression models—have gained popularity and are subject of ongoing debates about their potential and limitations (Breen, Choi, and Holm 2015; Killewald and Bearak 2014; Xie, Brand, and Jann 2012).
The workshop “Causality in the Social Sciences III” picks up on these recent approaches and debates from three different angles:
-
It’s complicated: Interpretation of heterogeneous effects
-
Estimating heterogeneous effects with observational and experimental data
-
Machine learning techniques for specification search
The two-day workshop aims to provide opportunities for exchange and discussions about ongoing substantial and methodological research tackling one or several of these three specific topics. Confirmed keynotes will be delivered by Jennie E. Brand, Professor of Sociology and Statistics at UCLA, and Richard Breen, Professor of Sociology at the University of Oxford and Fellow of Nuffield College.
Besides these general talks, early career researchers (PhD students and PostDocs) have the opportunity to present, discuss, and reflect on how far aspects of heterogeneity in causal effects are relevant for their research (10 minutes for presentation and 20 minutes for discussion). Each presentation will be discussed by the invited keynote speakers as well as the workshop participants. In addition, each participant is expected to thoroughly discuss two previously assigned papers that he or she will have to read before the workshop.
A maximum of 15 presentations will be accepted. The workshop is supported by the German Akademie für Soziologie and GESIS – Leibniz Institute for the Social Sciences. Workshop participation is free of charge. However, travel, accommodation, as well as lunch and dinner expenses, cannot be covered. Priority in admission will be given to members of the Akademie für Soziologie.
For more information click "LINK TO ORIGINAL" below.