Thammasat University students interested in sociology, ethnography, qualitative research, and related subjects may find a new book useful.

Analytic Induction for Social Research is an Open Access book available for free download at this link:

https://luminosoa.org/site/books/#temp-204

The Thammasat University Library collection includes several books about different aspects of analytic induction.

It is by Professor Charles C. Ragin who teaches sociology at the University of California, Irvine, The United States of America.

The publisher’s description notes:

This book explores analytic induction, an approach to the analysis of cross-case evidence on qualitative outcomes that has deep roots in sociology. A popular research technique in the early decades of empirical sociology, analytic induction differs fundamentally as a method of social research from conventional variation-based approaches. In Analytic Induction for Social Research, Charles C. Ragin demonstrates that much is gained from systematizing analytic induction. The approach he introduces here offers a new template for conducting cross-case analysis and provides a new set of tools for answering common research questions that existing methods cannot address.

The American Psychological Association (APA) Dictionary of Psychology defines analytic induction as

a qualitative research strategy for developing and testing a theory in which the researcher tentatively defines a phenomenon, creates a hypothesis to explain it, and examines a single specific occurrence of the phenomenon in order to confirm or refute the hypothesis. If the hypothesis is confirmed, additional cases are examined to confirm the results until a sufficient degree of certainty about the correctness of the hypothesis is obtained and the study may be concluded. If the hypothesis is not confirmed, the phenomenon is redefined or the hypothesis revised so as to accommodate the findings.

The International Encyclopedia of the Social & Behavioral Sciences definition of the term follows:

Analytic induction (AI) is a research logic used to guide data collection, develop analysis, and organize the presentation of research findings. The strategy of AI is exclusively qualitative, seeking encounters with new varieties of data in order to force revisions that will make the analysis valid when applied to an increasingly diverse range of cases. Originally understood as an alternative to statistical sampling methodologies, AI continues primarily as a way to develop explanations of the interactional processes through which people develop what, in their experiences, are homogenous forms of distinctive social action. AI transforms and produces a sociological appreciation of phenomena along recurrent lines. The target phenomenon is progressively redefined to address a process; explanatory conditions are redefined to specify the interactions through which people, by learning, recognizing, or becoming aware of features of their biographies and circumstances, in effect set up the motivational dynamics of their own conduct. The methodology of AI thus dovetails with the theoretical perspective of symbolic interaction and produces a phenomenologically grounded sociology. AI also dovetails with ethnography’s narrative style. One frequently cited weakness is that AI specifies only necessary but not sufficient conditions. Another is that it produces tautological explanations. The methodology does, however, support what might be called ‘retrodiction’: predictions, made at time 3, that if a given behavior is observed to have occurred at time 2, specific phenomena must have come into existence at time 1. Although rarely touted as ‘policy research,’ the upshot of an AI study is a documented portrayal of some segment of social life that is systematically misrepresented by the culture that supports power. AI’s quest for systematic knowledge is no less secure than is the understanding that social life takes shape as people crystallize long-evolved perspectives, familiar behavioral techniques, and cultured interpersonal sensibilities into patterns of situationally responsive conduct and distinctive qualities of lived experience.

The Introduction to Professor Ragin’s book begins:

Social scientists ask diverse kinds of research questions. Usually, each such question calls for application of a specific analytic strategy to empirical evidence. For example, questions about the distribution of wealth in a population call for the analysis of variation in levels of wealth across a sample of households, using sociodemographic and other variables to predict levels. Analytic methods for the study of distributions are especially well developed in the social sciences today. Variation in a dependent variable (e.g., household wealth) is explained using variation in independent variables (e.g., race, ethnicity, immigration status, education). Social scientists have developed a vast array of variation-based analytic techniques, perfect for addressing questions about distributions. But not all research questions are so lucky. Often, the research goal is to understand “how” a qualitative outcome happens by examining a set of cases that display the outcome. The distribution of that outcome in a sample drawn from a population will be relevant, but the empirical focus in determining the how of the outcome must rest on cases that display the outcome. Cases without the outcome— key evidence in the analysis of variation in the distribution of the outcome—can provide only very limited information regarding how the outcome happens. Restricting the analytic focus to cases that display the outcome, however, transforms the “dependent variable” into a constant—which precludes using the many variation-based analytic techniques that social scientists have developed. There is no readymade technique, comparable in sophistication to techniques that rely on a dependent variable, for the analysis of constants as outcomes. Questions regarding how outcomes happen are quite common, though— especially in everyday discourse. Unfortunately, they are often recast by social scientists as questions about distributions. Imagine, for example, that instead of learning about the process of becoming a marijuana user by observing and interviewing users, Howard Becker had instead examined the distribution of marijuana use in a random sample drawn from a given population. Suppose he found high levels of use among musicians and certain other, related groups. While indirectly relevant to the how question, the finding does not address it head-on. To find out how one becomes a marijuana user, it is necessary to study users, focusing especially on their shared experiences in learning to use marijuana and on other widely shared antecedent conditions. This book offers a straightforward methodology for the assessment of research questions regarding the antecedent conditions linked to qualitative outcomes. A typical qualitative study has a set of cases that display the outcome in question— the focal outcome—along with evidence on relevant antecedent conditions. The goal of the analysis is to identify antecedent conditions shared by cases with the focal outcome. Shared antecedent conditions, in turn, may be interpreted as “recipes” for an outcome, especially when they make sense as combinations of causally relevant conditions. In the end, the researcher explains a constant (the focal outcome) by way of other constants or near-constants (shared antecedent conditions). My approach to the analysis of systematic cross-case evidence on qualitative outcomes has deep roots in sociology in the form of a technique known as analytic induction (AI). AI was a popular research technique in the early decades of empirical sociology, beginning with the publication of Florian Znaniecki’s (1934) The Method of Sociology. Exemplary AI studies include Alfred Lindesmith’s  Addiction and Opiates, Donald Cressey’s Other People’s Money, and Howard Becker’s Becoming a Marihuana User. AI seeks to establish invariant (or “universal”) conditions for qualitative outcomes, focusing exclusively on instances of the outcome and how it came about in each case.

(All images courtesy of Wikimedia Commons)