Book chapter summary - Abduction: The Logic of Discovery of Grounded Theory, by Jo Reichertz

What I read

In this text, Reichertz considers the idea of abductive reasoning in the context of grounded theory, and proposes that, while not explicitly described as part of grounded theory in foundational texts, this form of logic plays a considerable role in analyzing and making sense of research data.

First, Reichertz describes that grounded theory is “one of the most successful methods ever developed and has added a more qualitative note to social research.” This is in comparison to what was a standing view, that theories emerge from data (as is in the case of a traditional scientific experimental approach.) However, researchers who have embraced grounded theory have also become divided on the nature of where knowledge comes from—if it comes from pre-knowledge in the researcher being applied to the data, or if it comes from the data itself, inductively. Reichertz introduces abductive reasoning as a way of bridging these two perspectives.

Next, Reichertz describes abduction. As an approach, it is known and referenced by many social researchers, but Reichertz feels that those approaches are largely wrong because they view it as a way towards “rule governed and replicable production of new and valid knowledge.” It is, instead, as “sensible and scientific as a form of inference, however it reaches to the sphere of deep insight and new knowledge.” Peirce is referenced as one of the first to position abductive reasoning in the context of social science knowledge production, and Reichertz feels that his work has been misinterpreted.

Reichertz then focuses on the various qualities of logic present in data analysis, in order to identify the differences in an abductive way of reasoning. There are three procedures that are present in abduction that differ from a more traditional approach to coding or sorting.

The first of these procedures is subsumption. This is the process of deductive reasoning: information is identified in data based on an existing set of rules. The next procedure is generalization, which is extracting a rule out of an organized analysis of data; this is the process of inductive reasoning. In qualitative research, this often draws from and across different participants.

The next form of data processing consists of “assembling or discovering, on the basis of interpretation of collected data, such combination of features for which there is no appropriate explanation or rule in the store of knowledge that already exists.” The activity leads to a surprise (often an actual emotional surprise, not simply unexpected knowledge). In order to see where the new information can exist, a new rule is invented for organizing the content. Reichertz concludes that it is a “cerebral process, an intellectual act, a mental leap, that brings together things which one had never associated with one another: A cognitive logic of discovery.”

Peirce has described two different ways to provoke or encourage this form of abductively reasoned discovery. The first is to use an external pressure, like fear, to force a conclusion. The second is to use a blank canvas of a mind to allow a new idea to form. In both approaches, the “consciously working mind, relying on logical rules, is outmaneuvered.” In this way, Reichertz concludes that abductive inferencing is an attitude as compared to a method or methodology. It develops a set of “mental constructs with which one can live comfortably or less comfortably.” These constructs are then tested through other means of meaning making, and if they survive, they become more solidified. The “testing” occurs through a set of steps.

First, a set of predictions is derived from the abductive hypothesis (which is a deductive activity.) Then, a search for facts is performed (which is an inductive activity.) Then, this is repeated. It is a loop of all forms of logic. The output cannot be verified, according to Peirce, because “the rule would have to be checked by all members of society, including those that don’t yet exist.

Reichertz then returns to the relationship between abductive reasoning and grounded theory. Grounded theory is a way for scientists to code data intellectually, and to develop concepts and theories, intellectually, while “moving to and fro between the collection of data, coding, and memoing.” For grounded theory to leverage abductive reasoning, it would have to “systematically count on the appearance of new codes of hypotheses.” Reichertz concludes that, based on quotes from the foundational texts related to grounded theory, the theory does provide room for this form of reasoning, and also for qualitative induction; “the logic of later GT thus permits abductive reasoning, counts on it, enables it, grants it a place.”

What I learned and what I think

I find myself back to abduction, but this time in the context of a more generalized theory of data interpretation rather than data interpretation in design-focused qualitative research. I am still struggling to really understand the “grounded” part of grounded theory, maybe because I haven’t read the source material of Strauss and Corbin, or maybe because I don’t know about the arguments present internally in the social science research community, or maybe because designers already do this, so it doesn’t seem new to me. But my evolving view of it, in light of how this article embraces abductive reasoning, is that it’s a back-and-forth between themes emerging from the data (as if the data “holds” the answers), an application of theming and organization to the data (as if our life experiences “hold” the answers), and honoring abductive leaps when they occur (as if the lack of a categorical organization rule leads to a new rule, and the new rule “holds” the answers.) And, if I had to put a fine point on what I think is the part rejected by grounded theory that is more familiar to me in design interpretation, it would be the first—that somehow the data itself “holds” answers.

I think all of these are true.

I know the data “holds” the answers, at least as much as there are answers in qualitative design research. Each utterance stands for something that a participant meant, or does, or thinks, and if I’m using any form of user-centeredness, that meaning, doing, and thinking is a first-class citizen on the way towards meaning. In fact, in the context of one single participant, it is the guiding principle, because it’s their life, and I’m “just” the researcher. There’s an unresolved role of user-centered design here in grounded theory, I think. I want to come back to this.

I also know that I “hold” the answers, again as much as there are answers to hold, because my interpretation across participants is on the way towards a goal of making, not just understanding. This is a big chasm that I need to think through, between research in the context of a design problem and research in the context of an experiment or in a pursuit of knowing. If I know that I’m going to redirect my research data in order to bring something new to life, my pattern language on the design problem itself is always acting as a very clear spotlight that I aim here or there based on what I almost already know will emerge.

And I know that abductive reasoning is at play, because neither the participants, nor I, have the whole picture if there is going to be something new happening, and there always is, because circumstances are always different. My output may end up being the same at a broad level, if I’m directing this towards a design problem. If I do a project for Bank of America and then for Chase and then for Whatever Bank, and conduct research around the same focus for all three companies, I’m going to find my way to a similar sort of broad conclusions or directions in each case. But the details of the supporting argument will be based on very different stories and interpretations, and those will cover the details of the design in very different ways, because I’ll be “solving” for different participants wants and needs and desires. There’s a clear and unresolved role that patterns and pattern identification play in the success and strategies of abductive reasoning. I want to come back to this, too.

So, what does this look like in a method, or in a study?

Would this “count”?

I conducted research, and then performed a discourse analysis based on grounded theory to understand why people say what they say in the context of an interview. [My interpretations, applied on top of data—inductive, within a participant.]

After that interpretation, I extracted meaningful utterances by comparing my analysis to a set of initial hypotheses.

Then, I conducted a reflexive theming of the various utterances, now across participants.

And then, I described insight statements for the themes, based on abductive leaps towards meaning, focusing on behavior, sentiment, and things that are particularly surprising or anomalous.

I like that I’m back at abductive reasoning, because it’s familiar to me—I feel like I’m on solid ground here. So, I probably shouldn’t keep reading about it. I’ll find my way back to raw creativity next.