Customer feedback, interviews, work diaries, observation records, and open-ended survey answers may contain important facts even when they are not numerical. Yet when we encounter unfamiliar information, we often explain it through concepts and theories we already know. If we merely sort observations into predefined categories, phenomena for which we do not yet have names can disappear from view.
Qualitative data analysis is not simply organizing text or counting words. It is also an attempt to understand what is happening in the world while examining our assumptions. The original motivation behind grounded theory offers a particularly valuable perspective on this task.
The Starting Point of Grounded Theory: Do Not Force the Data into a Theory
In 1967, Barney Glaser and Anselm Strauss published The Discovery of Grounded Theory. They challenged a research environment in which testing hypotheses derived from established theories often received more recognition than developing theory through empirical investigation. Their work articulated strategies for generating theories grounded in the study of actual phenomena.
The point was not that researchers should approach the world as blank slates, without prior knowledge. Rather, a familiar theory should not be treated as the answer before the phenomenon has been investigated. Existing concepts can help us ask questions and make comparisons, but they must remain open to revision when the evidence requires it.
Grounded theory subsequently developed into different traditions, including Glaserian, Strauss–Corbin, and constructivist approaches, with different understandings of coding and theory construction. ThinkNavi treats their original intellectual challenge as more important for general readers than any one prescribed set of procedural steps.
Example: What We May Miss by Calling Something “Resistance to Change”
Imagine that employees continue using an old paper form after a new business system is introduced. If management immediately labels this “resistance to change,” its response may focus on training and changing employee attitudes.
Closer observation might reveal that the form records exceptions for particular customers or safety checks for which the new software has no fields. The issue may not be resistance at all. The official system may be unable to represent knowledge essential to the work.
Suspending the initial label and comparing actual behaviors and circumstances creates room for different explanations and interventions. This is a hypothetical example, but it illustrates a central value of qualitative inquiry.
Different Approaches to Qualitative Analysis
| Approach | Main purpose | Common mistake |
|---|---|---|
| Grounded theory | Develop concepts and explanations through comparison across cases | Assuming that any coding exercise constitutes grounded theory |
| Thematic analysis | Interpret patterns of meaning across a dataset | Treating themes as mere topic headings |
| Content analysis | Categorize content, sometimes using frequencies and counts | Equating frequency with importance |
| Narrative analysis | Examine how experiences are organized and understood through stories | Removing statements from their narrative context |
| Discourse analysis | Examine language, social assumptions, and sometimes power relations | Reducing discourse to word counts |
| Affinity diagramming / KJ method | Group fragments to reveal patterns and relationships | Mistaking grouping for a tested explanatory theory |
These approaches have different objectives and sometimes different epistemological commitments. Similar-looking procedures do not necessarily represent the same method, and no approach is universally best.
Let Concepts Grow Through Comparison
One influential element of grounded theory is the constant comparative method. Cases are compared not only to identify similarities but also to investigate differences, unexpected examples, and exceptions. Categories and their conditions of application are reconsidered as the analysis develops.
Different customer comments categorized as “dissatisfaction” may reflect pricing issues, unmet expectations, or changes in the customer’s operations. Conversely, comments expressed in very different words may point to the same underlying process. As comparisons continue, relationships that were initially invisible may emerge.
Grounded theory research can also employ theoretical sampling: collecting additional cases because an emerging explanation raises new questions. Instead of classifying a fixed dataset once, the researcher changes what is investigated next in response to provisional findings.
How LLMs Can Help — and Mislead
LLMs can locate relevant passages in large collections, group similar statements, and suggest potential explanations. They could substantially reduce the mechanical burden of qualitative analysis. But because LLMs generate plausible interpretations from learned patterns, they can also assimilate unfamiliar phenomena into familiar categories too quickly. This closely resembles the tendency that grounded theory originally challenged.
AI-assisted qualitative analysis should preserve original passages, sources, and context while distinguishing observations from AI-generated interpretations. Competing explanations, unclassified cases, and contradictory evidence should remain visible. When necessary, analysis should lead back to additional investigation.
Clustering can help group similar documents, but the meaning of a cluster still requires interpretation. A rare comment may signal a far more important new problem than the most frequent category.
Questions to Ask Before Accepting a Classification
- Was this category developed through investigation, or was it assumed from the beginning?
- Where are the cases that do not fit the favored explanation?
- Do different speakers really mean the same thing when they use the same word?
- Which observations become invisible when we adopt this interpretation?
- What should we investigate next to challenge or refine our explanation?
The value of qualitative analysis lies not merely in following a sophisticated procedure. It lies in moving repeatedly between observations and concepts so that the subject can teach us something our existing theories did not already contain.
References and Further Reading
- Barney G. Glaser and Anselm L. Strauss, The Discovery of Grounded Theory: Strategies for Qualitative Research (1967)
- Kathy Charmaz, Constructing Grounded Theory (2006; 2nd ed. 2014)
- Virginia Braun and Victoria Clarke, “Using Thematic Analysis in Psychology” (2006)
- Grounded theory methodology: has it become a movement? (2012)
- Braun & Clarke: Comparing thematic analysis and other qualitative approaches (2021)