Analytic Induction

Definition

A research strategy of data collection and analysis which explicitly takes the deviant case as a starting point for testing models or theories developed in research. It can be characterized as a method of systematic interpretation [Page 5]of events, which includes the process of generating hypotheses as well as testing them. Its decisive instrument is to analyse the exception or the case that is deviant to the hypothesis.

Distinctive Features

This procedure, introduced by Znaniecki in 1934, of looking for and analysing deviant cases is applied after a preliminary theory (hypothesis, pattern or model) has been developed. Analytic induction, above all, is oriented to examining theories and knowledge by analysing or integrating negative cases.

The procedure includes the following steps: (1) a rough definition of the phenomenon to be explained is developed; (2) a hypothetical explanation of the phenomenon is formulated; (3) a case is studied in the light of this hypothesis to find out whether the hypothesis corresponds to the facts in this case; (4) if the hypothesis is not correct, either the hypothesis is reformulated or the phenomenon to be explained is redefined in a way that excludes this case. Thereafter, the researcher actively searches for negative cases to discredit the hypothesis, model or typology.

Practical certainty can be obtained after a small number of cases has been studied, but the discovery of each individual negative case by the researcher or another researcher refutes the explanation and calls for its reformulation. Each negative case calls for the redefinition of concepts and/or the reformulation of hypotheses. Further cases are studied, the phenomenon is redefined and the hypotheses are reformulated until a universal relation is established.

Evaluation

Analytic induction is not based on enumerative argumentation. As it focuses on the single deviant case to test a more or less generalized model, it is a genuinely qualitative way of assessing the stability and limitations of research findings. With its emphasis on testing theories it goes one step further than the development of grounded theory. Three types of results are obtained with analytic induction: forms of activities (how something is done normally), accounts of self-awareness and explanations, and motivational and other reasons for specific behaviours are all analysed and presented.

Analytic induction does not start from conventional definitions or models of what is studied. As it implicitly assumes that hypotheses, theories and models are not immediately perfect, it is a strategy to refine interpretative conclusions from data. By searching for negative cases and by testing models and theories against them, it is a strategy to define their limits and to make explicit under what time, local and social conditions they are not valid. Therefore, analytic induction is a way to generalize and to delimit qualitative, case-based findings.

Analytic induction has been criticized as it does not – as originally intended by Znaniecki (1934) – provide a means for establishing causal laws and universals. There are question marks against the generalization of case studies and external validity in general. Nevertheless, analytic induction has its own importance as a procedure for assessing and developing analyses by the use of negative cases.

...