The constant comparative method was developed by sociologists Barney Glaser and Anselm Strauss in the 1960s as a core component of their grounded theory approach to qualitative research.
It involves systematically comparing new data with existing data to identify patterns, similarities, and differences. This comparison happens at three levels: incident with incident, incident with category, and category with category.
The discipline matters. It forces the researcher to justify why two segments count as the same, which pushes the analysis toward abstraction rather than description.
This process is iterative and continues throughout the research process, from initial coding to theoretical sampling and memo writing.
The constant comparative method continues until researchers reach theoretical saturation, meaning that no new information or insights emerge from the data.
Key Takeaways
- Constantly comparing data: As data is collected, it is constantly compared to existing data to identify similarities, differences, and patterns.
- Developing codes and categories: Based on the comparisons, codes are developed to label and categorize the data. These codes are then grouped into categories to further organize the data.
- Refining codes and categories: As new data is collected, the codes and categories are refined and modified to ensure they accurately reflect the data.
- Developing theory: Through this iterative process of comparing, coding, and refining, a grounded theory emerges from the data.
Here is a step-by-step approach to applying the constant comparative method:
Open coding
Open coding, also called initial coding, is the first stage of data analysis in grounded theory.
During this stage, researchers engage with the data in detail, going line by line to identify important words and label them. In vivo codes use words from the participants as labels.
In grounded theory, codes are labels given to data segments that have a similar meaning
Before coding begins, researchers typically read through the data once to get a feel for it as a whole. They then choose how finely to break it apart.
Line-by-line coding assigns a code to each line of a transcript, which keeps the analysis close to the participants’ own words. Incident-by-incident coding instead treats a whole event or episode as the unit, which suits field notes and observational data.
Open coding produces several kinds of code. In-vivo codes borrow the participant’s own words, such as “feeling lost.” Descriptive codes summarise a passage in the researcher’s own words.
Process codes, usually gerunds like “concealing the diagnosis” or “rationing energy,” capture what people are doing. Analytic codes name a more abstract idea in the data. This moves the analysis toward theory.
Begin by analyzing your data line by line, identifying key concepts, ideas, and events.
Assign initial codes to these elements, remaining close to the data and using gerunds (verbs ending in “ing”) to focus on processes and actions.
Example:
- Interview 1: The first participant might say, “I felt so lost when I got laid off. I didn’t know what to do with myself. I spent weeks just watching TV and feeling sorry for myself.”
- Initial codes: The researcher could assign initial codes such as “feeling lost,” “lack of direction,” “inactivity,” and “self-pity.”
Axial coding
Axial coding involves refining and connecting the initial codes generated during open coding to develop a more comprehensive understanding of the data.
Compare each new piece of data with previously coded data, looking for similarities and differences.
This helps to ensure the codes “fit” the data and capture the essence of what is happening.
- Compare codes with additional data: Researchers compare additional data with the open codes they have developed to ensure that they accurately reflect the meaning of the data. Researchers would examine whether new data support, contradict, or add nuance to these codes.
- Compare the codes with each other: Researchers compare codes to see if they are similar enough to be merged into a more comprehensive code, or if they need to be split into separate categories to represent distinct concepts.
- Revising codes: Codes can be revised to better fit the data.
- Combining codes: Codes that are very similar can be merged together into a new, more comprehensive code.
- Creating new codes: Sometimes codes need to be split to represent separate categories.
When multiple researchers are involved in coding, they independently code data and then collaborate to unify their coding.
By comparing their coding, researchers can identify disagreements, leading to discussions that illuminate nuances in the data.
These discussions can highlight nuanced interpretations and enrich the overall analysis.
Analyzing these disagreements helps refine the categories and ensure they reflect the participants’ experiences.
Example:
- Interview 2: The second participant might say, “I was angry when I was fired. I felt like I had been betrayed by the company. I started going to the gym every day to channel my frustration.”
- Comparing and refining codes: The researcher compares these data to the codes from the first interview. They realize “feeling lost” and “anger” could both be categorized under a broader concept of “experiencing negative emotions.”
- Creating new codes: The second interview introduces the idea of “physical activity” as a coping mechanism, so the researcher creates this new code.
Create Categories
The constant comparison method helps researchers group conceptually similar data under a conceptual label, facilitating the development of concepts and categories grounded in the data.
Categories are groups of related codes that represent patterns or themes in the data. Each category has properties.
A property is a general characteristic, and it varies along a dimension. A category such as “managing disclosure,” for example, might have a property like how much is told, ranging from total concealment to full disclosure.
As you compare codes, identify connections and group them together into broader categories.
Continue comparing new codes and categories with existing ones. As your understanding evolves, you may:
- Refine existing codes and categories.
- Develop new codes and categories.
- Merge or split existing codes and categories.
Through constant comparison, researchers identify the most significant and frequent category. This becomes the core category: the central category to which every other category is related.
This core category then serves as a focal point, integrating the other categories into a coherent theory.
Example:
- Interview 3: The third participant might say, “I knew I had to get back on my feet quickly. I started networking and updating my resume right away. I also reached out to a career counselor for advice.”
- Further refinement and categorization: The researcher sees that this data reflects proactive coping strategies. They might create categories like “job search activities” and “seeking support.”
- Core category: The researcher might identify “taking control” as the core category that explains the main processes participants experience.
- Relationships between categories: The theory might propose that taking control through proactive coping strategies (like job search activities and seeking support) can help mitigate negative emotions and foster a sense of hope and agency.
Theoretical Sampling
Theoretical sampling is used in grounded theory to strategically select participants or data sources that will best inform the developing theory.
This iterative process starts after the initial coding of data from a purposive sample, which is chosen to maximize variation.
Constant comparison informs theoretical sampling. It reveals gaps and areas that require further exploration. By comparing emerging categories with existing data, researchers can identify gaps in what they already know. This tells them what to sample next.
Theoretical sampling ensures that the constant comparative method is not limited to the initial data set.
It allows the analysis to grow and evolve as the research progresses, leading to a more nuanced and well-developed grounded theory.
Example:
- Theoretical sampling: The researcher may start to notice certain patterns. For example, they may observe that people who engage in proactive coping strategies tend to report lower levels of negative emotions. To explore this further, the researcher might intentionally seek out more interviewees who actively engage in job search activities or seek support, a strategy known as theoretical sampling.
Memo Writing
Memo writing is an essential part of the constant comparative method.
Throughout the coding process, researchers write memos to document their thoughts, insights, questions, and connections they observe during their comparisons.
This continuous reflection and documentation align with the iterative nature of the constant comparative method, where new insights are constantly compared with existing ones to refine understanding.
These memos serve as:
- An audit trail: They track the evolution of the researcher’s thinking and decision-making process.
- A tool for reflection: Memos encourage researchers to critically examine their assumptions and interpretations.
- A catalyst for theoretical insights: The act of writing can often lead to new connections and ideas that might not have emerged otherwise.
Critical Evaluation
Like any qualitative method, the constant comparative method has clear strengths and recognised limitations.
Strengths
The method offers several advantages:
- Grounded in the data: Codes and categories are built up from what participants actually say, so they fit the material closely rather than being imposed from existing theory.
- Discovers the unexpected: Because the analyst asks “what is happening here?” of each segment, the method surfaces processes and categories a fixed coding frame would miss.
- Systematic and transparent: Constant comparison and memo-writing leave a documented audit trail that other researchers can inspect, countering the charge that qualitative analysis is merely impressionistic.
- Keeps meaning close to participants: Using participants’ own words as codes reduces the risk that the researcher’s categories quietly replace what participants actually meant.
- Theory-generating: Unlike purely descriptive coding, the method is explicitly a route to new theory, not just a summary of what was said.
Limitations and Debates
The method also has real limitations:
- Labour-intensive: Line-by-line coding across many transcripts, alongside constant comparison and memo-writing, is demanding of time and expertise, so it suits small, in-depth studies better than large ones.
- Researcher subjectivity: Because the analyst is the coding instrument, two researchers may fracture and label the same data differently, reflecting their own assumptions.
- The fragmentation critique: Breaking an account into many discrete codes can strip out its narrative coherence, cutting a person’s story into fragments that lose their original connections.
- The Glaser-Strauss schism: Glaser broke with co-founder Strauss, arguing that prescribed coding steps force data into pre-set frames instead of letting categories emerge (Glaser, 1992).
- The saturation debate: When to stop coding is contested, and “theoretical saturation” is often confused with simply running out of new topics rather than the categories themselves being fully developed.
Contemporary Research
Recent methodological work has sharpened exactly this stopping-point question.
- Aim: To examine how “saturation” is understood across qualitative traditions and clarify how it should be defined for a given study (Saunders et al., 2018).
- Method: A conceptual and methodological review that synthesised how saturation is defined and applied across the qualitative research literature, rather than a new empirical study.
- Results: The authors identified four distinct models. These include theoretical saturation (the grounded-theory sense, tied to categories and theoretical sampling) and data saturation, where new data merely repeat what is already known.
- Conclusion: Which model applies depends on the analytic approach. For the constant comparative method, coding should stop once the categories themselves are fully developed, not simply once interviews start to sound repetitive.
This reframes theoretical saturation as a property of the categories, not a count of interviews, which is the sense the constant comparative method has always used. The implication is simple.
Overall, the constant comparative method is a demanding but powerful technique. Its value lies in building theory from data, rather than testing a theory assumed true in advance. That quality depends on the researcher’s disciplined comparison and honest reflexivity throughout.
Sources
Glaser, B. G. (1992). Basics of grounded theory analysis: Emergence vs. forcing. Sociology Press.
Glaser, B., & Strauss, A. (2017). Discovery of grounded theory: Strategies for qualitative research. Routledge.
Guba, E. G., & Lincoln, Y. S. (1994). Competing paradigms in qualitative research. Handbook of qualitative research, 2(163-194), 105.
Maykut, P., & Morehouse, R. (2002). Beginning qualitative research: A philosophical and practical guide. Routledge.
Saunders, B., Sim, J., Kingstone, T., Baker, S., Waterfield, J., Bartlam, B., Burroughs, H., & Jinks, C. (2018). Saturation in qualitative research: Exploring its conceptualization and operationalization. Quality & Quantity, 52(4), 1893–1907. https://doi.org/10.1007/s11135-017-0574-8
