Bayesian Reasoning for Case Studies and Comparative Research (QMMR B)
Tasha Fairfield
1:30 PM – 5:30 PM
Los Angeles Convention Center, 406A
This short course introduces participants to the Bayesian logic of qualitative case studies, with practical advice, examples, and small-group exercises to enable them to use this method in their work. It builds on Social Inquiry and Bayesian Inference: Rethinking Qualitative Research, by Tasha Fairfield and Andrew Charman (Cambridge University Press, 2022). The material presented here complements the morning short course on process tracing led by Andrew Bennett, Jeffrey T. Checkel, and Tasha Fairfield, but each course can also be usefully taken independently from the other.
The core idea motivating this course is that the way we intuitively approach qualitative case research is similar to how we read detective novels. We consider various different hypotheses to explain what occurred—whether the emergence of democracy in South Africa, or the death of Samuel Ratchett on the Orient Express—drawing on the literature we have read (e.g. theories of regime change, or other Agatha Christie mysteries) and any salient previous experiences we have had. As we gather evidence and discover new clues, we update our beliefs about which hypothesis provides the best explanation—or we may introduce a new alternative that occurs to us along the way.
Bayesianism provides a natural framework that is both logically rigorous and grounded in common sense, that governs how we should revise our degree of belief in the truth of a hypothesis—e.g., “mobilisation from below drove democratization in South Africa by altering economic elites’ regime preferences,” (Wood 2001), or “a lone gangster sneaked onboard the train and killed Ratchett as revenge for being swindled”—given our relevant prior knowledge and new information that we obtain during our investigation. Bayesianism is enjoying a revival across many fields, and it offers a powerful tool for improving inference and analytic transparency in qualitative research.
The first part of this course introduces basic principles of Bayesian reasoning with the goal of helping us leverage common-sense understandings of inference and improve intuition when conducting causal analysis with qualitative evidence. We begin with the general logic of Bayesian inference, that is, how we update our prior view about which explanation is more plausible when we learn new evidence. We explain the importance of working with rival hypotheses and discusses how to formulate well-constructed explanations to compare. We then elaborate practical procedures for evaluating the inferential import of the evidence by “mentally inhabiting” the world of each hypothesis and asking which one makes the evidence more expected, and then updating our prior views about which hypothesis provides the best explanation. We include examples and exercises to illustrate how this process works with real-world qualitative evidence.
The second part of the course turns to comparative case studies. Methodological literature often treats cross-case (e.g., comparative) analysis and within-case analysis (e.g., process tracing) as distinct analytical endeavors that draw on different logics of inference. Within a Bayesian framework, however, there are no fundamental distinctions; all evidence contributes to inference in the same manner, whether we are studying a single case or multiple cases. In essence, each piece of evidence we obtain weighs in favor of one explanation over a rival to some degree, which we assess by asking which explanation makes that evidence more expected. Evidentiary weight then aggregates both within any given case, and across different cases that fall within the scope of the theories we are testing. In addition to showing how this process works with examples drawn from published comparative case studies, we will introduce a Bayesian approach to case selection and discuss how to articulate scope conditions and tentatively generalize our hypotheses.
Note: This course does not require any prior training in process training, Bayesianism, probability theory, or logic. The only technical skills that will be assumed are basic arithmetic.