A developmental approach to historical causal inference
- PDF / 713,729 Bytes
- 27 Pages / 439.37 x 666.142 pts Page_size
- 105 Downloads / 232 Views
A developmental approach to historical causal inference David A. Bateman1 · Dawn Langan Teele2 Received: 17 August 2019 / Accepted: 26 August 2019 © Springer Science+Business Media, LLC, part of Springer Nature 2019
Abstract Empirical historical research typically falls into one of three categories: the study of major historical events; the use of “history as data” to test general theories; and the study of the legacies of historical processes. We argue that because of data sparsity and dynamically unfolding processes, the study of major historical events is less well suited to design-based inference than other types of historical research. Drawing examples from our own work, we propose a set of research procedures for designing causally oriented work, and argue that the construction of a “timeline of relevant counterfactual nodes” can facilitate the organization of a research project investigating complex historical processes. The researcher can focus on relevant counterfactual moments as potential episodes of change using either statistical or qualitative techniques as appropriate, moving forward through the timeline and updating their beliefs about a hypothesized cause’s importance across the process. Keywords Causal inference · American political development · Historical institutionalism JEL Classification D72 · C18 · N41
1 Introduction While both description and interpretation are key elements of social science research, the goal to which many of us aspire is the ability to make and substantiate empirically causal explanations for complex social phenomena. We want to know what has happened, what it means, what its consequences were or will be; but, perhaps above all, we want to know why it’s happened. That is, to put it mildly, a tricky business, and debates about what causation is, and how best to go about doing causal inference, have been central to social science since its emergence as a distinctive field of study.
* David A. Bateman [email protected] Dawn Langan Teele [email protected] 1
Department of Government, Cornell University, 218 White Hall, Ithaca, NY 14853, USA
2
Political Science Department, University of Pennsylvania, 434 Ronald O. Perelman Center for Political Science and Economics, Philadelphia, PA 19104, USA
13
Vol.:(0123456789)
Public Choice
With the rise of a more formalized counterfactual approach to causal inference—the potential outcomes framework (Rubin 1974; Holland 1986; Neyman 1990)—there has been a renewed appreciation for, and use of, experimental and so-called quasi-experimental methods across the discipline (Dunning 2012; Diamond and Robinson 2010; Morgan and Winship 2007; Pearl 1995). The advent of the potential outcomes framework has helped drive the “historical turn” in comparative politics and fostered broader appreciation for history in the study of American politics, reaching well-beyond the confines of American political development (APD) as a distinct scholarly community.1 But design-based inference has yet to produce a single standard for conductin
Data Loading...
