MeMa - Methodology Matters
Crash Course
Bayesian Inference in Qualitative Research
Tasha Fairfield
2017-18 Mellon Foundation Fellow, Center for Advanced Study in the Behavioral Sciences, Stanford University
London School of Economics
5 November 2018, h. 10.00-12.30 and 14.00-16.30
6 November 2018, h. 10.00-13.00 and 14.00-16.00
Room A - NASP Graduate School in Social and Political Sciences
Via Pace, 10 - Milan
Abstract
The way we intuitively approach qualitative research is similar to how we read detective novels. We consider 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. As we gather evidence and discover new clues, we continually 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, to govern how we should revise our degree of belief in the truth of a hypothesis—e.g., "mobilization 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 in qualitative research.
This interactive workshop will introduce basic principles of Bayesian reasoning with the goal of helping to leverage common-sense understandings of inference and hone intuition when conducting causal analysis with qualitative evidence. We will examine the foundations of Bayesian probability as well as concrete applications to single case studies and comparative case studies. Participants will learn how to construct rival hypotheses, assess the inferential weight of qualitative evidence, and evaluate which hypothesis provides the best explanation through Bayesian updating. The workshop will also cover key aspects of research design, including case selection and iteration between theory development and data analysis. Bayesian probability not only fits naturally with how we intuitively move back and forth between theory and data, but also provides a framework for rational reasoning that mitigates confirmation bias and ad-hoc hypothesizing—two common problems associated with iterative research. At the end of the course, the participants will be able to read qualitative case studies more critically and apply Bayesian principles to their own research.
Thanks to the contribution of the Compagnia di San Paolo, the course is free, and open to PhD students, MA students, and Post-doc researchers regardless of their background or previous knowledge. Due to organizational constraints, however, admittance is closed to 25 participants.
Candidate participants are kindly requested to send a motivated expression of interest to
This email address is being protected from spambots. You need JavaScript enabled to view it.
by October 26, 2018.
Faculty members are all welcome as observers.
Program
Monday, 5 November
10:00-12:30 Session I
Introduction to Bayesian Reasoning
14:00-16:30 Session II
Explicit Bayesian Analysis
Tuesday, 6 November
10:00-13:00 Session III
Iterative Research & Case Study Scrutiny
14:00-16:00 Session IV
Case Selection
Background Readings
1) Fairfield (2013), "Going Where the Money Is: Strategies for Taxing Economic Elites in Unequal Democracies", World Development, 47:42-57. Please skip pp. 42-45 and READ ONLY the Chilean cases, pp. 47-49.
2) Fairfield & Charman (2017), "Explicit Bayesian Analysis for process Tracing", Political Analysis 25 (3): 363-380.
Further readings will be provided at the end of the workshop.
r causal diagrams. However, it is very technically written.
Morgan, Stephen L. and Christopher Winship. (2007). Counterfactuals and Causal Inference: Methods and Principles of Social Research. Cambridge: Cambridge University Press.
Chapter 3 (p. 77-95) is a more applied discussion of directed graphs embedded in examples from sociology.
Robins, James M. (2001). “Data, Design, and Background Knowledge in Etiologic Inference,” Epidemiology 11 (3): 313-320.
Another less technical description of directed graphs.
Part II – Necessary and Unnecessary Controls or Closing the Backdoor
Time: 10:45-12:15
Recommended Literature
Morgan, Stephen L. and Christopher Winship. (2007). Counterfactuals and Causal Inference: Methods and Principles of Social Research. Cambridge: Cambridge University Press.
Chapter 3 (p. 95-130) is an applied discussion of the back-door criteria.
Pearl, J., Glymour, M., & Jewell, N. P. (2016). Causal inference in statistics: a primer. John Wiley & Sons.
Chapter 3 (p. 53-75).
Lunch Break 12:15-13:30
Part III – Effect Heterogeneity and the Average Treatment Effect
Time 13:30-15:00
Recommended Literature
Elwert, F., & Winship, C. (2010). Effect heterogeneity and bias in main-effects-only regression models. Heuristics, probability and causality: A tribute to Judea Pearl, 327-36.
Falleti, T. G., & Lynch, J. F. (2009). Context and causal mechanisms in political analysis. Comparative political studies, 9(42), 1143-1166.
Part IV – Interactions versus Structural Equations with Mediation Effects
Time 15:15-17:15
Recommended Literature
Pearl, J., Glymour, M., & Jewell, N. P. (2016). Causal inference in statistics: a primer. John Wiley & Sons.
Chapter 3 (p. 75-87).
Brambor, T., Clark, W. R., & Golder, M. (2006). Understanding interaction models: Improving empirical analyses. Political analysis, 14(1), 63-82.
Imai, K., Keele, L., & Yamamoto, T. (2010). Identification, inference and sensitivity analysis for causal mediation effects. Statistical Science, 1(25), 51-71.