Tasha Fairfield

Social Inquiry and Bayesian Inference: Rethinking Qualitative Research

with A.E. Charman

Cambridge Univeristy Press 2022  Strategies for Sociel Inquiry Series

​

Now in Print!  click: Discount for Purchase extended through end of 2023

​

We reexamine the logic of inference in qualitative social science by drawing on the Bayesian school of “probability as extended logic” from the natural sciences. Bayesianism is enjoying a revival in many fields, and it can provide a rigorous but as yet underappreciated foundation for inference in qualitative research.  

 

Logical Bayesianism conceptualizes probability as the rational degree of belief we should hold in a proposition given the information we possess.  Bayesian probability serves as    the uniquely consistent extension of deductive logic, where we know whether any given proposition is true or false, to situations where information is incomplete and hypotheses can rarely be definitively confirmed or disproven.  In principle, Bayesian probability provides a unified framework for all scientific inquiry.  From this perspective, we reexamine central debates regarding inferential methods and best practices for research design and analytic transparency in qualitative scholarship.  

 

The book provides an extended treatment of how to apply Bayesian analysis to evaluate complex, real-world, qualitative case-study evidence, with fully-worked example applications.  We also elaborate Bayesian insights for avoiding cognitive biases and improving analytical judgments in traditional case study narratives. Beyond case study applications, we argue that Bayesianism guides inference in cross-case comparative studies, facilitates combining quantitative and qualitative information, and lessons distinctions between large-N vs.  small-N research, probabilistic vs. determinist causation, and deductive vs. inductive stages of analysis.  Moreover, we illustrate that Bayesianism grounds many common practices in qualitative research that are not justified within a frequentist statistical approach to inference.  The intended audience includes social science scholars at the graduate level and above who conduct qualitative and/or multi-method research.  

We also envision that our methodological perspective could prove highly valuable for policy analysts and practitioners, who must often make consequential decisions based on incomplete information. 

 

Table of Contents

​

Read Chapter 1: Introduction

​