Analytic Induction

Qualitative Approach

When employed within qualitative methods, the technique is largely for the use of negative case analysis, or examples that do not fit within the theoretical or analytic framework provided. The negative cases are examined to identify the nature of the departure and modify or add structure to the existing theory to account for the basis of the departure. Suppose you are examining compliments, and someone says, “Nice dress.” One issue with that kind of comment is that sarcasm can fit the form of a compliment but the meaning or intent of the utterance is not to provide a positive commentary. What this means is that every example has to be considered and a rule or element introduced that separates the “nice dress” comments into those capable of being identified as an affirmation or positive statement versus those that should be considered negative. Establishing that standard or method of evaluation would then improve the classification and identification of the relevant examples.

Each time a negative or unusual case that does not fit within the system is identified, a modification of the system is required. The assumption is that the process of examination and modification is ongoing and dynamic. The focus becomes on identifying those instances of discourse or statements that do not fit within the existing framework. The strength of this analysis is that the ability of a system to account for the overwhelming majority of examples provides the basis for a system that could simply handle a large part of communication situations and utterances. The process of adapting the system to counterexamples provides much of the time and effort when using this system of generating theoretical explanations.

Quantitative Approach

The quantitative version seeks to attain as complete a prediction of an outcome variable as possible, often using some combination of predictor variables. When a prediction is less than complete (not 100% accurate), the goal is to analyze residuals (errors from prefect prediction) and to find a means of adding an additional variable to increase significantly the accuracy of the prediction. The result should be a model that continues to increase or improve in prediction of the outcome feature as more and more data are collected.

Accounting for Counterfactuals and Outliers

Both approaches receive only limited use by members of the scholarly community. For qualitative researchers, the use of counterfactuals (negative cases) is extremely important for philosophic reasons, but negative cases, particularly those related to use of language or cultural issues, can result from a number of causes. For example, a person with mental illness may tell someone that he was “experimented on by both the CIA and gray aliens on a secret and invisible space station hovering over Manhattan.” The problem is that in a conversational sense, trying to make sense of cases that are inconsistent with an analytic system may require establishing rules external to the interaction regarding competency. The example, by claiming that the person is hallucinating creates an explanation that can be invoked when observing gibberish or nonsensical statements. Thus, the creation of an “other” category permits the ability to simply take disconfirming cases and place them in a position that requires no adaptation to the underlying structure used for explanation. Permission of this option provides a means to explain all counterexamples without addressing or requiring analysis of the underlying system or structure.

...