selective coding qualitative research

Our websites may use cookies to personalize and enhance your experience. By continuing without changing your cookie settings, you agree to this collection. For more information, please see our University Websites Privacy Notice .

Neag School of Education

Educational Research Basics by Del Siegle

Open, in vivo, axial, and selective coding.

Open, axial, and selective coding are three distinct processes used in qualitative research, particularly in the field of grounded theory. They involve the systematic analysis and categorization of data to identify patterns, themes, and relationships.

Open Coding:

Open coding is the initial stage of qualitative data analysis. It is a method where data are initially broken down and analyzed to identify concepts, categories, or themes. It involves generating initial codes that capture the main ideas or concepts found in the data. Open coding allows for exploration and discovery, as the researcher remains open to emerging patterns and concepts without predetermined categories. The researcher reads and re-reads the data, line by line or segment by segment, to identify significant concepts, actions, and meanings. Open coding helps in developing a comprehensive understanding of the data by identifying a wide range of ideas and perspectives.

In Vivo Coding:

In vivo coding is a specific technique sometimes used during the open coding phase. It involves using participants’ exact words or phrases as codes to capture their lived experiences and perspectives. In vivo codes are verbatim representations of participants’ language, preserving the authenticity and richness of their expressions. Therefore, in vivo coding is a technique employed within the broader open coding process. It is a way to create codes based on participants’ exact words, contributing to the development of categories and themes during the open coding phase of qualitative analysis.

Axial Coding:

Axial coding is the next step in the qualitative data analysis process. It involves a more focused and systematic examination of the data to identify relationships between categories and subcategories identified during the open coding phase. Axial coding aims to establish connections and linkages between concepts, exploring how they relate to each other and contribute to the overall phenomenon under study. This process involves reorganizing and re-categorizing the codes based on their relationships, often using visual tools such as diagrams or matrices to visualize the connections. Axial coding helps to identify key themes, subthemes, and the underlying structure or framework that emerges from the data.

Selective Coding:

In grounded theory, the researcher is attempting to develop a theory or explanation that accounts for the observed phenomena. Selective coding is an important step in this process and follows open coding and axial coding. It involves further refining and organizing the data to identify a core category or central theme that captures the essence of the research. The goal is to develop a comprehensive understanding of the data and to create a theory or explanation that accounts for the observed phenomena.

Therefore , open coding is the initial phase where data are broken down to identify concepts and generate codes. Sometimes in vivo coding is employed within open coding to capture participants’ exact words and expressions to preserve authenticity. Axial coding follows open coding and focuses on finding relationships between categories and subcategories. Finally, selective coding aims to develop a theory that explains the topic of study by refining and organizing data into a core category or central theme. When conducting a grounded theory study, the goal is to achieve a comprehensive understanding of the data and create an explanatory theory, and open, axial, and selective coding are often used to achieve this.

The Ultimate Guide to Qualitative Research - Part 2: Handling Qualitative Data

Handling qualitative data
Transcripts
Field notes
Survey data and responses
Visual and audio data
Data organization
Data coding
Coding frame
Auto and smart coding
Organizing codes
Qualitative data analysis
Content analysis
Thematic analysis
Thematic analysis vs. content analysis
Narrative research
Phenomenological research
Discourse analysis

What is grounded theory in simple terms?

Core components of grounded theory, role of the researcher in grounded theory, constructivist grounded theory, what tools will help with grounded theory.

Deductive reasoning
Inductive reasoning
Inductive vs. deductive reasoning
Qualitative data interpretation
Qualitative analysis software

Grounded theory

Among the various approaches to qualitative data analysis , grounded theory is among those that stay very close to the data to develop a theory. Grounded theory analysis is essential, particularly in research inquiries where there is little or no existing theory, to guide the organization of knowledge from data collection .

Let's look at the topic of grounded theory methodology by exploring its rationale, its potential for theory development, the steps employed in grounded theory procedures, and how ATLAS.ti can facilitate grounded theory methods.

Barney G. Glaser and Anselm L. Strauss are credited with developing grounded theory as a widely-used research methodology in qualitative analysis in the social sciences. Over the decades, other researchers, such as Kathy Charmaz, have further developed grounded theory approaches. Glaser and Strauss broke new ground in qualitative inquiry through grounded theory by arguing against any notions that science had maximized its potential for developing new theory. As a result, the primary purpose of grounded theory is to construct theories grounded in systematically gathered and analyzed data rather than beginning with a preconceived theory. The ultimate goal is to generate a theory that offers an explanation of the research question that stems from what emerges from the data.

What is the main point of grounded theory?

At its core, grounded theory is about the discovery of new concepts and relationships. Rather than starting research with a theory and then testing it, grounded theory researchers begin with an area of study and allow what is relevant in that area to emerge. The methodology was developed as a response to the traditional approach of having a hypothesis before conducting research, which can lead to forcing the data to meet preconceived notions. The main point of grounded theory is to cultivate an understanding of social phenomena from the perspective of those experiencing them.

When should you use grounded theory research?

Grounded theory research is appropriate when there is little prior information or established theory about a phenomenon. It is most suitable for investigations of processes, actions, and interactions. Grounded theory can be particularly useful in exploratory studies where the aim is to identify key issues, explore them in detail, and construct a model or theory that can be used to understand the phenomenon from a new perspective.

You might choose to use grounded theory when you want to learn about people's experiences, their perceptions of these experiences, and the actions they take as a result. The inductive nature of grounded theory makes it suitable for studying social processes over time, understanding the changes and development of a phenomenon, and gaining in-depth insights from the perspective of those directly involved.

Benefits of using grounded theory

One of the key advantages of using grounded theory is that it promotes the emergence of new theories and deepens our understanding of the social world around us. Here are some of the key benefits:

Flexibility: Grounded theory is adaptive and can be adjusted during the research process, providing a degree of flexibility not often seen in other research methods.
Inductive nature: Because it is inductive , grounded theory is well suited for discovering how individuals interpret their experiences and the world around them.
Rich, detailed insights: Grounded theory's focus on the exploration of phenomena can lead to rich, detailed insights and in-depth understanding.
Practical outcomes: Grounded theory generates theory and provides practical implications that can inform policy, intervention, or program development.

Limitations of grounded theory

While grounded theory offers many advantages, it is essential to be aware of its limitations:

Time-consuming: The process of data collection and analysis can be time-consuming due to the method's iterative nature.
Complexity: The method can be complex to apply correctly due to its abstract concepts and the various stages of coding and analysis.
Requires skill and experience: Successful implementation of grounded theory requires strong analytical skills and experience in qualitative research.
Subjectivity: While subjectivity can be a strength in understanding the experiences of others, it can also be a limitation if biases are not properly acknowledged and managed.

Grounded theory, as a research methodology , consists of several core components that guide the research process, from data collection to the development of a final theoretical framework . These components are interrelated, each influencing and shaping the others in a dynamic, iterative process. The core components of grounded theory include theoretical sensitivity, theoretical sampling, coding and analysis , theoretical saturation, and theoretical integration.

Theoretical sensitivity

Theoretical sensitivity refers to a researcher's ability to understand and define phenomena in terms of their underlying patterns or structures. It's an acquired skill that grows with experience, through exposure to literature, professional experiences, and personal experiences. It's about being sensitive to the nuances and complexities of the data , understanding the subtle cues or messages, and being able to pull these together to form a coherent understanding. Theoretical sensitivity can be developed in many ways. Reading and engaging with relevant literature , attending workshops or seminars, conducting preliminary interviews or observations , or even through casual conversations related to the research topic, can help to increase a researcher's theoretical sensitivity. It is about having a sense of what is important in the data, what to pay attention to, and what can be given less importance.

Theoretical sampling

Theoretical sampling is the process of data collection driven by the emerging theory. Instead of having a predefined sample at the start of the research, grounded theorists allow their theoretical ideas to guide them in selecting new data sources to explore. This iterative process means that data collection and analysis occur simultaneously, and both are influenced by the emerging theory. Theoretical sampling can be quite challenging for new researchers as it requires a level of flexibility and openness that is not typically found in more structured research designs. The researcher needs to be comfortable with uncertainty and willing to follow the data wherever it may lead.

Coding and analysis

Coding and analysis are key processes in grounded theory, consisting of several stages. The first step is open coding , where the researcher examines the data in a detailed and line-by-line manner to identify initial concepts. The focus is on breaking down the data into discrete parts and closely examining them for their underlying meaning. The next stage is axial coding , where the researcher begins to assemble the data in new ways after the initial breakdown during open coding. The aim is to identify relationships between the initial codes and to group them into more abstract categories. The final stage is selective coding , where the researcher integrates and refines the categories to form a cohesive theoretical framework . The focus is on developing a unifying theory around which all other categories are related.

Theoretical saturation

Theoretical saturation is a critical concept in grounded theory. It refers to the point at which no new insights or concepts are being found in the data, indicating that the categories are well-developed and that further data collection is unnecessary. Saturation doesn't mean that every single aspect of the data has been explored but rather that the categories within the theory are robust and comprehensive. The concept of saturation is tied closely to the idea of theoretical sampling. As the theory begins to take shape, the researcher focuses their data collection on areas that will help to further develop or refine their emerging categories.

Theoretical integration

Theoretical integration is the final stage in grounded theory. It involves pulling together all the categories that have been developed, linking them together, and integrating them into a cohesive and coherent theory. Integration also involves a process of validation, where the researcher returns to their data and checks that their theory fits and explains the data. At this stage, it's important that the researcher is able to explain their theory clearly and convincingly, showing how it offers a new and insightful understanding of the phenomenon they have studied.

Ready to try out ATLAS.ti?

Download a free trial of ATLAS.ti and put your qualitative data to work.

Detailed steps in grounded theory research

The research process in grounded theory consists of a series of interconnected and iterative steps. Each step is part of a holistic process designed to allow a theory to emerge from the data inductively. The process of constant comparison, also known as the constant comparative method or constant comparative analysis, is central to this process. Here, we'll walk through these steps in detail.

Data collection

The first step in grounded theory research is data collection. Data can come from various sources such as interviews , observations , documents, or any other source relevant to the research question . The form of data collection can vary greatly, and the selection depends largely on the nature of the research question and the context of the study. It's important to note that in grounded theory, data collection is an iterative process and continues throughout the entire research process. Initial data collection informs the early stages of analysis and the emerging theory, which then guides further data collection. This back-and-forth between data collection and analysis is a distinguishing feature of grounded theory.

Open coding

After some data has been collected, the process of open coding begins. This is the first step in the constant comparative method. During open coding, the researcher carefully reads and re-reads the data, breaking it down into discrete incidents or ideas. Each of these incidents is then given a code - a word or short phrase that represents the essence of that piece of data. Open coding is a line-by-line analysis, which means that every line of the data is scrutinized and potentially given a code. It's during this process that the researcher starts to see categories and properties emerge from the data.

The open coding process is also where constant comparison begins. As each piece of data is coded, it's compared to other data coded in the same way. This comparison process allows the researcher to refine the definitions of codes and begin to see patterns and relationships.

Axial coding

The next step in the grounded theory process is axial coding . This stage of the constant comparative method involves taking the initial categories developed during open coding and beginning to see how they relate to each other.

During axial coding, the researcher is constantly comparing data within a category, as well as comparing categories to each other. This process allows for more abstract thinking about the data. It helps to identify central phenomena, contexts, conditions, strategies, and consequences - elements that help to give a structure to the emerging theory.

Selective coding

Selective coding is the final stage of constant comparative analysis. At this point, the researcher has a clear idea of the main categories and how they relate to each other. The goal of selective coding is to integrate these categories around a central, core category. This core category represents the main theme or process that the theory explains.

During selective coding, the researcher is still using constant comparison, but the focus is now on making sure that all categories are connected to the core category and that all categories are well-developed. This stage of the research process ends when theoretical saturation is reached - when no new data appears to add to the understanding of the core category.

Writing the theory

The final step in grounded theory research is to write up the theory. This is an important part of the process because it's where the researcher takes the abstract ideas that have been developed and turns them into a concrete, coherent theory.

Writing the theory involves clearly defining the core category, explaining how other categories relate to the core, and demonstrating how the theory explains the process or phenomena under study. The result is a well-integrated set of theoretical concepts that can offer new insights into the research question.

The researcher plays a critical role in grounded theory. They are not a passive observer but an active participant in the research process. From data collection to analysis to theory formation, the researcher's perspectives, experiences, and interpretive skills significantly shape the research process and outcomes. This section discusses the role of the researcher in grounded theory, including aspects of objectivity and subjectivity , as well as the importance of reflective practice.

Objectivity and subjectivity in research

In grounded theory research, the objectivity and subjectivity of the researcher are both significant considerations. Objectivity refers to the ability to conduct research in a neutral, unbiased manner. On the other hand, subjectivity acknowledges the researcher's personal experiences, backgrounds, and perspectives that they bring to the study.

In grounded theory, researchers aim for a balance between these two. While striving for objectivity helps foster the study's credibility, it's also important to recognize and consider the subjectivity of the researcher. It's this subjectivity that allows the researcher to interpret the data , relate to the participants, and understand the phenomenon in depth. Researchers should be transparent about their assumptions, biases, and preconceptions. Acknowledging these factors not only aids reflexivity but also contributes to the credibility and trustworthiness of the research.

Importance of reflective practice

Reflective practice is a cornerstone of grounded theory methodology. It involves the researcher critically reflecting on their own role in the research process and the impact they may have on the data collection , analysis , and theory formation. Through reflective practice, researchers become more aware of their own assumptions and perspectives and can better understand how these elements might influence their research.

Reflective practice takes place throughout the research process. During data collection, researchers might reflect on their interactions with participants, considering how their questions, demeanor, or reactions might influence the responses. During data analysis, reflective practice helps researchers understand how their preconceptions and interpretations shape the coding and emerging theory. In grounded theory, reflective practice is not a linear step but a continuous process that loops back and forth throughout the research. It's through this reflective practice that researchers can build a comprehensive and nuanced understanding of the phenomenon under study.

Role of the researcher in data collection and analysis

In grounded theory, the researcher is considered the primary tool of data collection and analysis. This is different from quantitative research , where data collection tools are often standardized questionnaires or tests.

As the primary tool of data collection, the researcher is involved in interviewing participants, observing behavior, and gathering documents or other artifacts. The researcher must be skilled in establishing rapport with participants, asking insightful questions, and carefully observing and noting details.

In terms of data analysis , the researcher's intellectual capacity, intuition, and creativity play a crucial role. The process of coding data , recognizing patterns, developing categories , and forming an overarching theory heavily relies on the researcher's analytical skills. Moreover, their ability to critically reflect on their own role and influence in the research process is vital to ensure the study's trustworthiness.

When carefully considering where they stand in any qualitative study, especially in a grounded theory study, the researcher should carefully reflect on their thinking and methods. Reflexivity is a process where researchers continuously evaluate and reflect upon their entire research process and their role within it. Researchers need to be conscious of their potential influence on the research and actively work to verify their conclusions. Maintaining a research diary, where thoughts, ideas, and reflections can be recorded throughout the study, is a common strategy used to promote reflexivity.

Grounded theory has evolved since its original inception by sociologists Barney Glaser and Anselm Strauss in the 1960s. One significant development is constructivist grounded theory, an approach that emphasizes the interpretive aspects of knowledge creation. This approach, most notably propagated by Kathy Charmaz, views research as a co-construction of knowledge between the researcher and the participants. Let's examine the foundations of constructivist grounded theory and the associated constructivist grounded theory methods.

Foundation of constructivist grounded theory

Constructivist grounded theory stems from the philosophical perspective of constructivism, which asserts that reality is socially constructed and subjective. Constructivists believe that people construct their own understanding of the world based on their experiences and interactions. Applying this viewpoint to grounded theory, constructivist grounded theorists argue that researchers and participants co-construct the data and the ensuing analysis. Hence, the researcher is not an objective observer but an active participant in the research process, contributing their interpretations and perspectives .

Key characteristics of constructivist grounded theory

There are several key characteristics that differentiate constructivist grounded theory from its traditional counterpart. These include the emphasis on researcher-participant interaction, the recognition of multiple realities, the focus on interpretive understanding, and the flexible use of grounded theory methods.

Emphasis on researcher-participant interaction: Constructivist grounded theory acknowledges that data doesn't exist in a vacuum. It's produced through interactions between the researcher and the participant. These interactions are dynamic, context-dependent, and mutually influential, contributing to the co-construction of knowledge.
Recognition of multiple realities: In line with constructivist philosophy, this approach recognizes the existence of multiple realities. Each participant and the researcher have their unique interpretation of reality, informed by their experiences, values, and social contexts.
Focus on interpretive understanding: Constructivist grounded theory prioritizes interpretive understanding. Rather than seeking an objective truth, this approach aims to understand how individuals interpret and make sense of their experiences.
Flexible use of grounded theory methods: While constructivist grounded theory maintains the core grounded theory methods, such as coding and theoretical sampling, it's more flexible in its use. The emphasis is on using these methods as tools to facilitate understanding rather than rigid steps to be followed.

Constructivist grounded theory methods

The process of conducting a constructivist approach to grounded theory study largely mirrors the steps of traditional grounded theory, albeit with a greater emphasis on reflexivity and the interpretive role of the researcher.

Data collection in constructivist grounded theory often involves in-depth interviews , observations , and document analysis , with the researcher actively engaging with the participants to co-construct the data. During the analysis, the researcher remains reflexive about their interpretations and assumptions, constantly checking them against the data .

Coding in constructivist grounded theory still involves open, axial, and selective coding, but the process is more flexible and intuitive. The researcher uses their insights and perspectives to guide the coding process, constantly comparing the data and remaining open to multiple interpretations.

The ultimate goal of constructivist grounded theory is to generate an interpretive theory that makes sense of the participants' experiences and actions. This theory is not seen as a concrete truth but a context-dependent, co-constructed interpretation of the phenomenon under study.

While this section has focused on the philosophical and methodological dimensions of grounded theory research, it's also important to think about what tools might be useful for researchers involved in conducting a grounded theory study. Previously in this guide, we have explored how ATLAS.ti can aid you in coding your data . That said, there are additional tools in qualitative data analysis and especially in ATLAS.ti that can facilitate the grounded theory coding and analysis process.

The axial coding stage of grounded theory, which deals with theoretical development, shifts the focus of data analysis from coding discrete instances of data to drawing connections between those codes. The researcher is responsible for identifying relationships between discrete phenomena that might have otherwise been thought of as unrelated to each other. Without such relationships, there would be no foundation for developing a novel theory relevant to the social world. Moreover, the sorting of knowledge and information cannot be done, nor can scientific knowledge be easily retrieved and understood, without visualizing these networks of social phenomena.

Whatever your analysis, ATLAS.ti can make it happen

An intuitive coding interface is just a few clicks away, starting with a free trial.

A guide to coding qualitative research data

Last updated

12 February 2023

Reviewed by

Each time you ask open-ended and free-text questions, you'll end up with numerous free-text responses. When your qualitative data piles up, how do you sift through it to determine what customers value? And how do you turn all the gathered texts into quantifiable and actionable information related to your user's expectations and needs?

Qualitative data can offer significant insights into respondents’ attitudes and behavior. But to distill large volumes of text / conversational data into clear and insightful results can be daunting. One way to resolve this is through qualitative research coding.

Streamline data coding

Use global data tagging systems in Dovetail so everyone analyzing research is speaking the same language

What is coding in qualitative research?

This is the system of classifying and arranging qualitative data . Coding in qualitative research involves separating a phrase or word and tagging it with a code. The code describes a data group and separates the information into defined categories or themes. Using this system, researchers can find and sort related content.

They can also combine categorized data with other coded data sets for analysis, or analyze it separately. The primary goal of coding qualitative data is to change data into a consistent format in support of research and reporting.

A code can be a phrase or a word that depicts an idea or recurring theme in the data. The code’s label must be intuitive and encapsulate the essence of the researcher's observations or participants' responses. You can generate these codes using two approaches to coding qualitative data: manual coding and automated coding.

Why is it important to code qualitative data?

By coding qualitative data, it's easier to identify consistency and scale within a set of individual responses. Assigning codes to phrases and words within feedback helps capture what the feedback entails. That way, you can better analyze and understand the outcome of the entire survey.

Researchers use coding and other qualitative data analysis procedures to make data-driven decisions according to customer responses. Coding in customer feedback will help you assess natural themes in the customers’ language. With this, it's easy to interpret and analyze customer satisfaction .

How do inductive and deductive approaches to qualitative coding work?

Before you start qualitative research coding, you must decide whether you're starting with some predefined code frames, within which the data will be sorted (deductive approach). Or, you may plan to develop and evolve the codes while reviewing the qualitative data generated by the research (inductive approach). A combination of both approaches is also possible.

In most instances, a combined approach will be best. For example, researchers will have some predefined codes/themes they expect to find in the data, but will allow for a degree of discovery in the data where new themes and codes come to light.

Inductive coding

This is an exploratory method in which new data codes and themes are generated by the review of qualitative data. It initiates and generates code according to the source of the data itself. It's ideal for investigative research, in which you devise a new idea, theory, or concept.

Inductive coding is otherwise called open coding. There's no predefined code-frame within inductive coding, as all codes are generated by reviewing the raw qualitative data.

If you're adding a new code, changing a code descriptor, or dividing an existing code in half, ensure you review the wider code frame to determine whether this alteration will impact other feedback codes. Failure to do this may lead to similar responses at various points in the qualitative data, generating different codes while containing similar themes or insights.

Inductive coding is more thorough and takes longer than deductive coding, but offers a more unbiased and comprehensive overview of the themes within your data.

Deductive coding

This is a hierarchical approach to coding. In this method, you develop a codebook using your initial code frames. These frames may depend on an ongoing research theory or questions. Go over the data once again and filter data to different codes.

After generating your qualitative data, your codes must be a match for the code frame you began with. Program evaluation research could use this coding approach.

Inductive and deductive approaches

Research studies usually blend both inductive and deductive coding approaches. For instance, you may use a deductive approach for your initial set of code sets, and later use an inductive approach to generate fresh codes and recalibrate them while you review and analyze your data.

What are the practical steps for coding qualitative data?

You can code qualitative data in the following ways:

1. Conduct your first-round pass at coding qualitative data

You need to review your data and assign codes to different pieces in this step. You don't have to generate the right codes since you will iterate and evolve them ahead of the second-round coding review.

Let's look at examples of the coding methods you may use in this step.

Open coding : This involves the distilling down of qualitative data into separate, distinct coded elements.

Descriptive coding : In this method, you create a description that encapsulates the data section’s content. Your code name must be a noun or a term that describes what the qualitative data relates to.

Values coding : This technique categorizes qualitative data that relates to the participant's attitudes, beliefs, and values.

Simultaneous coding : You can apply several codes to a single piece of qualitative data using this approach.

Structural coding : In this method, you can classify different parts of your qualitative data based on a predetermined design to perform additional analysis within the design.

In Vivo coding : Use this as the initial code to represent specific phrases or single words generated via a qualitative interview (i.e., specifically what the respondent said).

Process coding : A process of coding which captures action within data. Usually, this will be in the form of gerunds ending in “ing” (e.g., running, searching, reviewing).

2. Arrange your qualitative codes into groups and subcodes

You can start organizing codes into groups once you've completed your initial round of qualitative data coding. There are several ways to arrange these groups.

You can put together codes related to one another or address the same subjects or broad concepts, under each category. Continue working with these groups and rearranging the codes until you develop a framework that aligns with your analysis.

3. Conduct more rounds of qualitative coding

Conduct more iterations of qualitative data coding to review the codes and groups you've already established. You can change the names and codes, combine codes, and re-group the work you've already done during this phase.

In contrast, the initial attempt at data coding may have been hasty and haphazard. But these coding rounds focus on re-analyzing, identifying patterns, and drawing closer to creating concepts and ideas.

Below are a few techniques for qualitative data coding that are often applied in second-round coding.

Pattern coding : To describe a pattern, you join snippets of data, similarly classified under a single umbrella code.

Thematic analysis coding : When examining qualitative data, this method helps to identify patterns or themes.

Selective coding/focused coding : You can generate finished code sets and groups using your first pass of coding.

Theoretical coding : By classifying and arranging codes, theoretical coding allows you to create a theoretical framework's hypothesis. You develop a theory using the codes and groups that have been generated from the qualitative data.

Content analysis coding : This starts with an existing theory or framework and uses qualitative data to either support or expand upon it.

Axial coding : Axial coding allows you to link different codes or groups together. You're looking for connections and linkages between the information you discovered in earlier coding iterations.

Longitudinal coding : In this method, by organizing and systematizing your existing qualitative codes and categories, it is possible to monitor and measure them over time.

Elaborative coding : This involves applying a hypothesis from past research and examining how your present codes and groups relate to it.

4. Integrate codes and groups into your concluding narrative

When you finish going through several rounds of qualitative data coding and applying different forms of coding, use the generated codes and groups to build your final conclusions. The final result of your study could be a collection of findings, theory, or a description, depending on the goal of your study.

Start outlining your hypothesis , observations , and story while citing the codes and groups that served as its foundation. Create your final study results by structuring this data.

What are the two methods of coding qualitative data?

You can carry out data coding in two ways: automatic and manual. Manual coding involves reading over each comment and manually assigning labels. You'll need to decide if you're using inductive or deductive coding.

Automatic qualitative data analysis uses a branch of computer science known as Natural Language Processing to transform text-based data into a format that computers can comprehend and assess. It's a cutting-edge area of artificial intelligence and machine learning that has the potential to alter how research and insight is designed and delivered.

Although automatic coding is faster than human coding, manual coding still has an edge due to human oversight and limitations in terms of computer power and analysis.

What are the advantages of qualitative research coding?

Here are the benefits of qualitative research coding:

Boosts validity : gives your data structure and organization to be more certain the conclusions you are drawing from it are valid

Reduces bias : minimizes interpretation biases by forcing the researcher to undertake a systematic review and analysis of the data

Represents participants well : ensures your analysis reflects the views and beliefs of your participant pool and prevents you from overrepresenting the views of any individual or group

Fosters transparency : allows for a logical and systematic assessment of your study by other academics

What are the challenges of qualitative research coding?

It would be best to consider theoretical and practical limitations while analyzing and interpreting data. Here are the challenges of qualitative research coding:

Labor-intensive: While you can use software for large-scale text management and recording, data analysis is often verified or completed manually.

Lack of reliability: Qualitative research is often criticized due to a lack of transparency and standardization in the coding and analysis process, being subject to a collection of researcher bias.

Limited generalizability : Detailed information on specific contexts is often gathered using small samples. Drawing generalizable findings is challenging even with well-constructed analysis processes as data may need to be more widely gathered to be genuinely representative of attitudes and beliefs within larger populations.

Subjectivity : It is challenging to reproduce qualitative research due to researcher bias in data analysis and interpretation. When analyzing data, the researchers make personal value judgments about what is relevant and what is not. Thus, different people may interpret the same data differently.

What are the tips for coding qualitative data?

Here are some suggestions for optimizing the value of your qualitative research now that you are familiar with the fundamentals of coding qualitative data.

Keep track of your codes using a codebook or code frame

It can be challenging to recall all your codes offhand as you code more and more data. Keeping track of your codes in a codebook or code frame will keep you organized as you analyze the data. An Excel spreadsheet or word processing document might be your codebook's basic format.

Ensure you track:

The label applied to each code and the time it was first coded or modified

An explanation of the idea or subject matter that the code relates to

Who the original coder is

Any notes on the relationship between the code and other codes in your analysis

Add new codes to your codebook as you code new data, and rearrange categories and themes as necessary.

How do you create high-quality codes?

Here are four useful tips to help you create high-quality codes.

1. Cover as many survey responses as possible

The code should be generic enough to aid your analysis while remaining general enough to apply to various comments. For instance, "product" is a general code that can apply to many replies but is also ambiguous.

Also, the specific statement, "product stops working after using it for 3 hours" is unlikely to apply to many answers. A good compromise might be "poor product quality" or "short product lifespan."

2. Avoid similarities

Having similar codes is acceptable only if they serve different objectives. While "product" and "customer service" differ from each other, "customer support" and "customer service" can be unified into a single code.

3. Take note of the positive and the negative

Establish contrasting codes to track an issue's negative and positive aspects separately. For instance, two codes to identify distinct themes would be "excellent customer service" and "poor customer service."

4. Minimize data—to a point

Try to balance having too many and too few codes in your analysis to make it as useful as possible.

What is the best way to code qualitative data?

Depending on the goal of your research, the procedure of coding qualitative data can vary. But generally, it entails:

Reading through your data

Assigning codes to selected passages

Carrying out several rounds of coding

Grouping codes into themes

Developing interpretations that result in your final research conclusions

You can begin by first coding snippets of text or data to summarize or characterize them and then add your interpretative perspective in the second round of coding.

A few techniques are more or less acceptable depending on your study’s goal; there is no right or incorrect way to code a data set.

What is an example of a code in qualitative research?

A code is, at its most basic level, a label specifying how you should read a text. The phrase, "Pigeons assaulted me and took my meal," is an illustration. You can use pigeons as a code word.

Is there coding in qualitative research?

An essential component of qualitative data analysis is coding. Coding aims to give structure to free-form data so one can systematically study it.

Get started today

Go from raw data to valuable insights with a flexible research platform

Editor’s picks

Last updated: 21 December 2023

Last updated: 16 December 2023

Last updated: 6 October 2023

Last updated: 5 March 2024

Last updated: 25 November 2023

Last updated: 15 February 2024

Last updated: 11 March 2024

Last updated: 12 December 2023

Last updated: 6 March 2024

Last updated: 10 April 2023

Last updated: 20 December 2023

Latest articles

Codes and Coding

Ajay Gupta 2
First Online: 27 January 2024

208 Accesses

Part of the book series: Springer Texts in Social Sciences ((STSS))

Qualitative research is built on codes, and researchers must master the processes for creating codes and drawing insights from their analyses. In this chapter, the author discusses the various types of codes and approaches to coding. The quality of the output of qualitative data analysis is dependent on codes and the coding process. Codes may be relevant or irrelevant and the differences between them, and their significance, are explained in this chapter. The chapter discusses code categories and groups, and themes are derived from them. This chapter introduces coding cycle and Computer Assisted/Aided Qualitative Data Analysis (CAQDAS). After a brief overview, the author highlights the advantages and limitations of CAQDAS.

This is a preview of subscription content, log in via an institution .

Buying options

Available as PDF
Read on any device
Instant download
Own it forever
Available as EPUB and PDF
Compact, lightweight edition
Dispatched in 3 to 5 business days
Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Author information

Authors and affiliations.

Human Resource Management, VES Business School, Mumbai, Maharashtra, India

You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Ajay Gupta .

Rights and permissions

Reprints and permissions

Copyright information

About this chapter

Cite this chapter.

Gupta, A. (2023). Codes and Coding. In: Qualitative Methods and Data Analysis Using ATLAS.ti. Springer Texts in Social Sciences. Springer, Cham. https://doi.org/10.1007/978-3-031-49650-9_4

Download citation

DOI : https://doi.org/10.1007/978-3-031-49650-9_4

Published : 27 January 2024

Publisher Name : Springer, Cham

Print ISBN : 978-3-031-49649-3

Online ISBN : 978-3-031-49650-9

eBook Packages : Social Sciences Social Sciences (R0)

Share this chapter

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Publish with us

Policies and ethics

Find a journal
Track your research

AI & NLP
Churn & Loyalty
Customer Experience
Customer Journeys
Customer Metrics
Feedback Analysis
Product Experience
Product Updates
Sentiment Analysis
Surveys & Feedback Collection
Try Thematic

Welcome to the community

Coding Qualitative Data: How to Code Qualitative Research

How many hours have you spent sitting in front of Excel spreadsheets trying to find new insights from customer feedback?

You know that asking open-ended survey questions gives you more actionable insights than asking your customers for just a numerical Net Promoter Score (NPS) . But when you ask open-ended, free-text questions, you end up with hundreds (or even thousands) of free-text responses.

How can you turn all of that text into quantifiable, applicable information about your customers’ needs and expectations? By coding qualitative data.

Keep reading to learn:

What coding qualitative data means (and why it’s important)
Different methods of coding qualitative data
How to manually code qualitative data to find significant themes in your data

What is coding in qualitative research?

Coding is the process of labeling and organizing your qualitative data to identify different themes and the relationships between them.

When coding customer feedback , you assign labels to words or phrases that represent important (and recurring) themes in each response. These labels can be words, phrases, or numbers; we recommend using words or short phrases, since they’re easier to remember, skim, and organize.

Coding qualitative research to find common themes and concepts is part of thematic analysis . Thematic analysis extracts themes from text by analyzing the word and sentence structure.

Within the context of customer feedback, it's important to understand the many different types of qualitative feedback a business can collect, such as open-ended surveys, social media comments, reviews & more.

What is qualitative data analysis?

Qualitative data analysis is the process of examining and interpreting qualitative data to understand what it represents.

Qualitative data is defined as any non-numerical and unstructured data; when looking at customer feedback, qualitative data usually refers to any verbatim or text-based feedback such as reviews, open-ended responses in surveys , complaints, chat messages, customer interviews, case notes or social media posts

For example, NPS metric can be strictly quantitative, but when you ask customers why they gave you a rating a score, you will need qualitative data analysis methods in place to understand the comments that customers leave alongside numerical responses.

Methods of qualitative data analysis

Thematic analysis.

This refers to the uncovering of themes, by analyzing the patterns and relationships in a set of qualitative data. A theme emerges or is built when related findings appear to be meaningful and there are multiple occurences. Thematic analysis can be used by anyone to transform and organize open-ended responses, online reviews and other qualitative data into significant themes.

Content analysis:

This refers to the categorization, tagging and thematic analysis of qualitative data. Essentially content analysis is a quantification of themes, by counting the occurrence of concepts, topics or themes. Content analysis can involve combining the categories in qualitative data with quantitative data, such as behavioral data or demographic data, for deeper insights.

Narrative analysis:

Some qualitative data, such as interviews or field notes may contain a story on how someone experienced something. For example, the process of choosing a product, using it, evaluating its quality and decision to buy or not buy this product next time. The goal of narrative analysis is to turn the individual narratives into data that can be coded. This is then analyzed to understand how events or experiences had an impact on the people involved.

Discourse analysis:

This refers to analysis of what people say in social and cultural context. The goal of discourse analysis is to understand user or customer behavior by uncovering their beliefs, interests and agendas. These are reflected in the way they express their opinions, preferences and experiences. It’s particularly useful when your focus is on building or strengthening a brand , by examining how they use metaphors and rhetorical devices.

Framework analysis:

When performing qualitative data analysis, it is useful to have a framework to organize the buckets of meaning. A taxonomy or code frame (a hierarchical set of themes used in coding qualitative data) is an example of the result. Don't fall into the trap of starting with a framework to make it faster to organize your data. You should look at how themes relate to each other by analyzing the data and consistently check that you can validate that themes are related to each other .

Grounded theory:

This method of analysis starts by formulating a theory around a single data case. Therefore the theory is “grounded’ in actual data. Then additional cases can be examined to see if they are relevant and can add to the original theory.

Why is it important to code qualitative data?

Coding qualitative data makes it easier to interpret customer feedback. Assigning codes to words and phrases in each response helps capture what the response is about which, in turn, helps you better analyze and summarize the results of the entire survey.

Researchers use coding and other qualitative data analysis processes to help them make data-driven decisions based on customer feedback. When you use coding to analyze your customer feedback, you can quantify the common themes in customer language. This makes it easier to accurately interpret and analyze customer satisfaction.

What is thematic coding?

Thematic coding, also called thematic analysis, is a type of qualitative data analysis that finds themes in text by analyzing the meaning of words and sentence structure.

When you use thematic coding to analyze customer feedback for example, you can learn which themes are most frequent in feedback. This helps you understand what drives customer satisfaction in an accurate, actionable way.

To learn more about how Thematic analysis software helps you automate the data coding process, check out this article .

Automated vs. Manual coding of qualitative data

Methods of coding qualitative data fall into three categories: automated coding and manual coding, and a blend of the two.

You can automate the coding of your qualitative data with thematic analysis software . Thematic analysis and qualitative data analysis software use machine learning, artificial intelligence (AI) , and natural language processing (NLP) to code your qualitative data and break text up into themes.

Thematic analysis software is autonomous , which means…

You don’t need to set up themes or categories in advance.
You don’t need to train the algorithm — it learns on its own.
You can easily capture the “unknown unknowns” to identify themes you may not have spotted on your own.

…all of which will save you time (and lots of unnecessary headaches) when analyzing your customer feedback.

Businesses are also seeing the benefit of using thematic analysis software. The capacity to aggregate data sources into a single source of analysis helps to break down data silos, unifying the analysis and insights across departments . This is now being referred to as Omni channel analysis or Unified Data Analytics .

Use Thematic Analysis Software

Try Thematic today to discover why leading companies rely on the platform to automate the coding of qualitative customer feedback at scale. Whether you have tons of customer reviews, support chat or open-ended survey responses, Thematic brings every valuable insight to the surface, while saving you thousands of hours.

Advances in natural language processing & machine learning have made it possible to automate the analysis of qualitative data, in particular content and framework analysis. The most commonly used software for automated coding of qualitative data is text analytics software such as Thematic .

While manual human analysis is still popular due to its perceived high accuracy, automating most of the analysis is quickly becoming the preferred choice. Unlike manual analysis, which is prone to bias and doesn’t scale to the amount of qualitative data that is generated today, automating analysis is not only more consistent and therefore can be more accurate, but can also save a ton of time, and therefore money.

Our Theme Editor tool ensures you take a reflexive approach, an important step in thematic analysis. The drag-and-drop tool makes it easy to refine, validate, and rename themes as you get more data. By guiding the AI, you can ensure your results are always precise, easy to understand and perfectly aligned with your objectives.

Thematic is the best software to automate code qualitative feedback at scale.

Don't just take it from us. Here's what some of our customers have to say:

I'm a fan of Thematic's ability to save time and create heroes. It does an excellent job using a single view to break down the verbatims into themes displayed by volume, sentiment and impact on our beacon metric, often but not exclusively NPS.

It does a superlative job using GenAI in summarizing a theme or sub-theme down to a single paragraph making it clear what folks are trying to say. Peter K, Snr Research Manager.

Thematic is a very intuitive tool to use. It boasts a robust level of granularity, allowing the user to see the general breadth of verbatim themes, dig into the sub-themes, and further into the sentiment of the open text itself. Artem C, Sr Manager of Research. LinkedIn.

AI-powered software to transform qualitative data at scale through a thematic and content analysis.

How to manually code qualitative data

For the rest of this post, we’ll focus on manual coding. Different researchers have different processes, but manual coding usually looks something like this:

Choose whether you’ll use deductive or inductive coding.
Read through your data to get a sense of what it looks like. Assign your first set of codes.
Go through your data line-by-line to code as much as possible. Your codes should become more detailed at this step.
Categorize your codes and figure out how they fit into your coding frame.
Identify which themes come up the most — and act on them.

Let’s break it down a little further…

Deductive coding vs. inductive coding

Before you start qualitative data coding, you need to decide which codes you’ll use.

What is Deductive Coding?

Deductive coding means you start with a predefined set of codes, then assign those codes to the new qualitative data. These codes might come from previous research, or you might already know what themes you’re interested in analyzing. Deductive coding is also called concept-driven coding.

For example, let’s say you’re conducting a survey on customer experience . You want to understand the problems that arise from long call wait times, so you choose to make “wait time” one of your codes before you start looking at the data.

The deductive approach can save time and help guarantee that your areas of interest are coded. But you also need to be careful of bias; when you start with predefined codes, you have a bias as to what the answers will be. Make sure you don’t miss other important themes by focusing too hard on proving your own hypothesis.

What is Inductive Coding?

Inductive coding , also called open coding, starts from scratch and creates codes based on the qualitative data itself. You don’t have a set codebook; all codes arise directly from the survey responses.

Here’s how inductive coding works:

Break your qualitative dataset into smaller samples.
Read a sample of the data.
Create codes that will cover the sample.
Reread the sample and apply the codes.
Read a new sample of data, applying the codes you created for the first sample.
Note where codes don’t match or where you need additional codes.
Create new codes based on the second sample.
Go back and recode all responses again.
Repeat from step 5 until you’ve coded all of your data.

If you add a new code, split an existing code into two, or change the description of a code, make sure to review how this change will affect the coding of all responses. Otherwise, the same responses at different points in the survey could end up with different codes.

Sounds like a lot of work, right? Inductive coding is an iterative process, which means it takes longer and is more thorough than deductive coding. A major advantage is that it gives you a more complete, unbiased look at the themes throughout your data.

Combining inductive and deductive coding

In practice, most researchers use a blend of inductive and deductive approaches to coding.

For example, with Thematic, the AI inductively comes up with themes, while also framing the analysis so that it reflects how business decisions are made . At the end of the analysis, researchers use the Theme Editor to iterate or refine themes. Then, in the next wave of analysis, as new data comes in, the AI starts deductively with the theme taxonomy.

Categorize your codes with coding frames

Once you create your codes, you need to put them into a coding frame. A coding frame represents the organizational structure of the themes in your research. There are two types of coding frames: flat and hierarchical.

Flat Coding Frame

A flat coding frame assigns the same level of specificity and importance to each code. While this might feel like an easier and faster method for manual coding, it can be difficult to organize and navigate the themes and concepts as you create more and more codes. It also makes it hard to figure out which themes are most important, which can slow down decision making.

Hierarchical Coding Frame

Hierarchical frames help you organize codes based on how they relate to one another. For example, you can organize the codes based on your customers’ feelings on a certain topic:

In this example:

The top-level code describes the topic (customer service)
The mid-level code specifies whether the sentiment is positive or negative
The third level details the attribute or specific theme associated with the topic

Hierarchical framing supports a larger code frame and lets you organize codes based on organizational structure. It also allows for different levels of granularity in your coding.

Whether your code frames are hierarchical or flat, your code frames should be flexible. Manually analyzing survey data takes a lot of time and effort; make sure you can use your results in different contexts.

For example, if your survey asks customers about customer service, you might only use codes that capture answers about customer service. Then you realize that the same survey responses have a lot of comments about your company’s products. To learn more about what people say about your products, you may have to code all of the responses from scratch! A flexible coding frame covers different topics and insights, which lets you reuse the results later on.

Tips for manually coding qualitative data

Now that you know the basics of coding your qualitative data, here are some tips on making the most of your qualitative research.

Use a codebook to keep track of your codes

As you code more and more data, it can be hard to remember all of your codes off the top of your head. Tracking your codes in a codebook helps keep you organized throughout the data analysis process. Your codebook can be as simple as an Excel spreadsheet or word processor document. As you code new data, add new codes to your codebook and reorganize categories and themes as needed.

Make sure to track:

The label used for each code
A description of the concept or theme the code refers to
Who originally coded it
The date that it was originally coded or updated
Any notes on how the code relates to other codes in your analysis

How to create high-quality codes - 4 tips

1. cover as many survey responses as possible..

The code should be generic enough to apply to multiple comments, but specific enough to be useful in your analysis. For example, “Product” is a broad code that will cover a variety of responses — but it’s also pretty vague. What about the product? On the other hand, “Product stops working after using it for 3 hours” is very specific and probably won’t apply to many responses. “Poor product quality” or “short product lifespan” might be a happy medium.

2. Avoid commonalities.

Having similar codes is okay as long as they serve different purposes. “Customer service” and “Product” are different enough from one another, while “Customer service” and “Customer support” may have subtle differences but should likely be combined into one code.

3. Capture the positive and the negative.

Try to create codes that contrast with each other to track both the positive and negative elements of a topic separately. For example, “Useful product features” and “Unnecessary product features” would be two different codes to capture two different themes.

4. Reduce data — to a point.

Let’s look at the two extremes: There are as many codes as there are responses, or each code applies to every single response. In both cases, the coding exercise is pointless; you don’t learn anything new about your data or your customers. To make your analysis as useful as possible, try to find a balance between having too many and too few codes.

Group responses based on themes, not words

Make sure to group responses with the same themes under the same code, even if they don’t use the same exact wording. For example, a code such as “cleanliness” could cover responses including words and phrases like:

Looked like a dump
Could eat off the floor

Having only a few codes and hierarchical framing makes it easier to group different words and phrases under one code. If you have too many codes, especially in a flat frame, your results can become ambiguous and themes can overlap. Manual coding also requires the coder to remember or be able to find all of the relevant codes; the more codes you have, the harder it is to find the ones you need, no matter how organized your codebook is.

Make accuracy a priority

Manually coding qualitative data means that the coder’s cognitive biases can influence the coding process. For each study, make sure you have coding guidelines and training in place to keep coding reliable, consistent, and accurate .

One thing to watch out for is definitional drift, which occurs when the data at the beginning of the data set is coded differently than the material coded later. Check for definitional drift across the entire dataset and keep notes with descriptions of how the codes vary across the results.

If you have multiple coders working on one team, have them check one another’s coding to help eliminate cognitive biases.

Conclusion: 6 main takeaways for coding qualitative data

Here are 6 final takeaways for manually coding your qualitative data:

Coding is the process of labeling and organizing your qualitative data to identify themes. After you code your qualitative data, you can analyze it just like numerical data.
Inductive coding (without a predefined code frame) is more difficult, but less prone to bias, than deductive coding.
Code frames can be flat (easier and faster to use) or hierarchical (more powerful and organized).
Your code frames need to be flexible enough that you can make the most of your results and use them in different contexts.
When creating codes, make sure they cover several responses, contrast one another, and strike a balance between too much and too little information.
Consistent coding = accuracy. Establish coding procedures and guidelines and keep an eye out for definitional drift in your qualitative data analysis.

Some more detail in our downloadable guide

If you’ve made it this far, you’ll likely be interested in our free guide: Best practices for analyzing open-ended questions.

The guide includes some of the topics covered in this article, and goes into some more niche details.

If your company is looking to automate your qualitative coding process, try Thematic !

If you're looking to trial multiple solutions, check out our free buyer's guide . It covers what to look for when trialing different feedback analytics solutions to ensure you get the depth of insights you need.

Happy coding!

Authored by Alyona Medelyan, PhD – Natural Language Processing & Machine Learning

CEO and Co-Founder

Alyona has a PhD in NLP and Machine Learning. Her peer-reviewed articles have been cited by over 2600 academics. Her love of writing comes from years of PhD research.

We make it easy to discover the customer and product issues that matter.

Unlock the value of feedback at scale, in one platform. Try it for free now!

Questions to ask your Feedback Analytics vendor
How to end customer churn for good
Scalable analysis of NPS verbatims
5 Text analytics approaches
How to calculate the ROI of CX

Our experts will show you how Thematic works, how to discover pain points and track the ROI of decisions. To access your free trial, book a personal demo today.

AI is Transforming Qualitative Research Coding

Discover the significance of coding in qualitative research, explore traditional methods, and witness the transformative impact of AI.

by Jessica Dubin

Chief Product Officer at Remesh

Coding transforms raw data into meaningful insights, the cornerstone of any research endeavor. As the research field progresses, we find ourselves at the cusp of a transformative era, where the integration of Generative AI into research practices is changing the game. How exactly does this technology work? Does it change current research practices? And what does it mean for researchers and insights professionals? Let’s dive in.

The importance of coding in qualitative research

When it comes to valuable data, we know that qualitative data has a wealth of insights. But when you ask open-ended questions, you’re left with hundreds or even thousands of responses that are all over the place. How do you make sense of it all? With coding.

Coding is the essential link connecting researchers and the insights buried within mountains of qualitative data. It's the process of categorizing, organizing, and extracting meaning from open-ended responses. Through coding, researchers can identify patterns, themes, and trends, ultimately providing a coherent narrative for the data.

Traditionally, coding has been a manual task, demanding researchers to meticulously sift through large volumes of text to identify and categorize relevant information. While effective, this manual approach has drawbacks—it's time-consuming, susceptible to human bias , and lacks scalability. This article explores how machine learning technology can be fitted to traditional coding methodologies in order to drive the efficiency, objectivity, and scalability of qualitative research.

Two principal manual coding methodologies

Before we look into how AI enters the picture, it’s important to understand traditional methodologies. AI doesn’t create new ideas; it works based on how we train it.

There are two principal methodologies for coding: inductive and deductive coding.

Inductive Coding

In this approach, researchers derive codes and themes directly from the data itself. Grounded theory, a popular framework in qualitative research , embodies this approach. It starts with open coding, applying initial codes, then axial coding to identify the connections between codes, and then selective coding, where researchers create clusters of similarly themed codes. This method offers a comprehensive framework to code large datasets and allows for the emergence of unexpected insights.

Deductive Coding

In contrast, deductive coding starts with a predefined codebook. Researchers apply pre-established codes to the data based on a priori theories or hypotheses. While it offers a more structured approach, it may limit the discovery of novel insights.

Emergence of Generative AI in coding

There’s been some automation in coding for a long time, but it’s mostly relied on term frequency. This process involves pulling specific words that appear often and then coding based on those words. While this approach can help code a large amount of data much quicker than manual methods, it relies on keywords alone, neglecting nuance and meaning.

In contrast, Generative AI can evaluate the meaning of the entire text to derive comprehensive and nuanced codes. It is trained on vast amounts of text data - hundreds of billions of words - enabling it to predict relevant words and phrases related to a text input. In an application like ChatGPT, a user submits a prompt, and the model can generate a human-like response that feels relevant. In the context of coding, the predictive nature of Generative AI makes it well-suited to identifying potential themes in a body of text.

Looking under the hood of Remesh AI

As we start embracing AI in our research practices , it’s essential to have complete transparency about how AI is used and what’s happening under the hood. This is how we can maintain the integrity of our research while utilizing innovation to improve our processes and insights.

At Remesh, we’re proud to be a pioneer in integrating Generative AI into qualitative research coding through our new feature, Auto Code . By blending the strengths of AI algorithms with proprietary models, our platform is shaving hours off of the researcher’s plate. Our process is loosely inspired by grounded theory for inductive coding, which allows for the emergence of unexpected insights. It also requires several handoffs between Remesh’s proprietary algorithm and the Generative AI models in order to achieve the right level of nuance and accuracy in the codes.

In the first step, Remesh prompts the Generative AI models to consider both the question and the response and asks it to output many descriptive codes for each response. Meta parameters like temperature are set to ensure nuanced responses.

The Generative AI models are powerful, but they are not always consistent. They often provide slightly different codes with essentially the same meaning to two different responses. To handle this, Remesh runs all of the codes that it gets back from Generative AI through its own proprietary algorithm that can map all of the codes by semantic similarity, which outputs groupings of the codes.

Some of those groupings contain just one code, and some are very large. Remesh has designed a process to assign all the singletons to an existing cluster and break large, multi-theme clusters into unique groups. Once the system has the right set and size of clusters, it asks Generative AI to create a unique name for each. Those names become the codes that are shown to the researcher on Remesh.

All of the codes are then assigned to a larger category grouping, using a similar process to the one used to assign codes to responses. The insights professional can review all of the codes and categories - editing, adding, and organizing as needed to ensure accuracy. This keeps the researcher in the driver’s seat while still utilizing AI’s powerful capabilities to make their existing processes faster and better.

While Remesh's platform assists in coding, human researchers remain integral to the process. The AI serves as a powerful tool, accelerating coding tasks and reducing the risk of errors, but the researcher's expertise and context are invaluable in refining and interpreting the results.

The future potential of AI-assisted coding

As we look ahead, the potential of AI-assisted coding is promising. The technology can drastically expedite coding tasks, making them accessible to a broader audience. With the assistance of AI, researchers can focus more on analyzing nuanced data cuts, deepening insights, and providing real-time business context. However, AI is not a cure-all; the symbiotic relationship between human researchers and AI tools is paramount for accurate and meaningful results.

The evolving role of the researcher

In this AI-driven era, the role of the researchers must evolve. Researchers are the stewards of data, leveraging AI tools to extract more profound insights efficiently. They are responsible for interpreting AI-generated codes and ensuring accuracy. They may need to validate and refine the codes to better tell the story of the data, especially given their familiarity with the questions and nuances of the dataset. Additionally, researchers have the invaluable context that AI lacks. The researcher understands the stakeholder needs and historical context that is needed to make recommendations for next steps. AI is a powerful ally, but the human touch turns data into actionable knowledge.

Jessica Dubin

The views, opinions, data, and methodologies expressed above are those of the contributor(s) and do not necessarily reflect or represent the official policies, positions, or beliefs of Greenbook.

Comments are moderated to ensure respect towards the author and to prevent spam or self-promotion. Your comment may be edited, rejected, or approved based on these criteria. By commenting, you accept these terms and take responsibility for your contributions.

About partner

Remesh is revolutionizing the insights industry with its platform. Engage live with up to 1,000 participants or asynchronously with up to 5,000, using AI to organize and analyze open-ended responses in real-time. More than 1,000 companies trust it with their insights, such as Deloitte, Barclays, Mercer, and Nestlé. To date, millions of insights have been enabled by Remesh. Learn more at www.remesh.ai.

Top in Quantitative Research

Research Methodologies

Moving Away from a Narcissistic Market Research Model

Why are we still measuring brand loyalty? It isn’t something that naturally comes up with consumers, who rarely think about brand first, if at all. Ma...

Devora Rogers

Chief Strategy Officer at Alter Agents

May 31, 2023

Read article

The Stepping Stones of Innovation: Navigating Failure and Empathy with Carol Fitzgerald

Natalie Pusch

Senior Content Producer at Greenbook

March 16, 2023

Play Episode

Get content that matters, written by top insights industry experts, delivered right to your inbox.

67k+ subscribers

Weekly Newsletter

Greenbook Podcast

Event Updates

I agree to receive emails with insights-related content from Greenbook. I understand that I can manage my email preferences or unsubscribe at any time and that Greenbook protects my privacy under the General Data Protection Regulation.*

We will send you a greatest letters one per week for your happy

Your guide for all things market research and consumer insights

Create a New Listing

Manage My Listing

Find Companies

Find Focus Group Facilities

Tech Showcases

GRIT Report

Expert Channels

Get in touch

Marketing Services

Future List

Publish With Us

Cookie policy

Want to create or adapt books like this? Learn more about how Pressbooks supports open publishing practices.

18.6 Grounded theory analysis

Learning objectives.

Learners will be able to…

Explain defining features of grounded theory analysis as a strategy for qualitative data analysis and identify when it is most effectively used
Formulate an initial grounded theory analysis plan (if appropriate for your research proposal)

What are you trying to accomplish with grounded theory analysis

Just to be clear, grounded theory doubles as both qualitative research design (we will talk about some other qualitative designs in Chapter 22) and a type of qualitative data analysis. Here we are specifically interested in discussing grounded theory as an approach to analysis in this chapter. With a grounded theory analysis , we are attempting to come up with a common understanding of how some event or series of events occurs based on our examination of participants’ knowledge and experience of that event. Let’s consider the potential this approach has for us as social workers in the fight for social justice. Using grounded theory analysis we might try to answer research questions like:

How do communities identity, organize, and challenge structural issues of racial inequality?
How do immigrant families respond to threat of family member deportation?
How has the war on drugs campaign shaped social welfare practices?

In each of these instances, we are attempting to uncover a process that is taking place. To do so, we will be analyzing data that describes the participants’ experiences with these processes and attempt to draw out and describe the components that seem quintessential to understanding this process.

Variations in the approach

Differences in approaches to grounded theory analysis largely lie in the amount (and types) of structure that are applied to the analysis process. Strauss and Corbin (2014) [1] suggest a highly structured approach to grounded theory analysis, one that moves back and forth between the data and the evolving theory that is being developed, making sure to anchor the theory very explicitly in concrete data points. With this approach, the researcher role is more detective-like; the facts are there, and you are uncovering and assembling them, more reflective of deductive reasoning . While Charmaz (2014) [2] suggests a more interpretive approach to grounded theory analysis, where findings emerge as an exchange between the unique and subjective (yet still accountable) position of the researcher(s) and their understanding of the data, acknowledging that another researcher might emerge with a different theory or understanding. So in this case, the researcher functions more as a liaison, where they bridge understanding between the participant group and the scientific community, using their own unique perspective to help facilitate this process. This approach reflects inductive reasoning .

Coding in grounded theory

Coding in grounded theory is generally a sequential activity. First, the researcher engages in open coding of the data. This involves reviewing the data to determine the preliminary ideas that seem important and potential labels that reflect their significance for the event or process you are studying. Within this open coding process, the researcher will also likely develop subcategories that help to expand and provide a richer understanding of what each of the categories can mean. Next, axial coding will revisit the open codes and identify connections between codes, thereby beginning to group codes that share a relationship. Finally, selective or theoretical coding explores how the relationships between these concepts come together, providing a theory that describes how this event or series of events takes place, often ending in an overarching or unifying idea tying these concepts together. Dr. Tiffany Gallicano [3] has a helpful blog post that walks the reader through examples of each stage of coding. Table 18.13 o ffers an example of each stage of coding in a study examining experiences of students who are new to online learning and how they make sense of it. Keep in mind that this is an evolving process and your document should capture this changing process. You may notice that in the example “Feels isolated from professor and classmates” is listed under both axial codes “Challenges presented by technology” and “Course design”. This isn’t an error; it just represents that it isn’t yet clear if this code is most reflective of one of these two axial codes or both. Eventually, the placement of this code may change, but we will make sure to capture why this change is made.

Constant comparison

While ground theory is not the only approach to qualitative analysis that utilizes constant comparison, it is certainly widely associated with this approach. Constant comparison reflects the motion that takes place throughout the analytic process (across the levels of coding described above), whereby as researchers we move back and forth between the data and the emerging categories and our evolving theoretical understanding. We are continually checking what we believe to be the results against the raw data. It is an ongoing cycle to help ensure that we are doing right by our data and helps ensure the trustworthiness of our research. Ground theory often relies on a relatively large number of interviews and usually will begin analysis while the interviews are ongoing. As a result, the researcher(s) work to continuously compare their understanding of findings against new and existing data that they have collected.

Developing your theory

Remember, the aim of using a grounded theory approach to your analysis is to develop a theory, or an explanation of how a certain event/phenomenon/process occurs. As you bring your coding process to a close, you will emerge not just with a list of ideas or themes, but an explanation of how these ideas are interrelated and work together to produce the event you are studying. Thus, you are building a theory that explains the event you are studying that is grounded in the data you have gathered.

Thinking about power and control as we build theories

I want to bring the discussion back to issues of power and control in research. As discussed early in this chapter, regardless of what approach we are using to analyze our data we need to be concerned with the potential for abuse of power in the research process and how this can further contribute to oppression and systemic inequality. I think this point can be demonstrated well here in our discussion of grounded theory analysis. Since grounded theory is often concerned with describing some aspect of human behavior: how people respond to events, how people arrive at decisions, how human processes work. Even though we aren’t necessarily seeking generalizable results in a qualitative study, research consumers may still be influenced by how we present our findings. This can influence how they perceive the population that is represented in our study. For example, for many years science did a great disservice to families impacted by schizophrenia, advancing the theory of the schizophrenogenic mother [4] . Using pseudoscience , the scientific community misrepresented the influence of parenting (a process), and specifically the mother’s role in the development of the disorder of schizophrenia. You can imagine the harm caused by this theory to family dynamics, stigma, institutional mistrust, etc. To learn more about this you can read this brief but informative editorial article by Anne Harrington in the Lancet . [5] Instances like these should haunt and challenge the scientific community to do better. Engaging community members in active and more meaningful ways in research is one important way we can respond. Shouldn’t theories be built by the people they are meant to represent?

Key Takeaways

Ground theory analysis aims to develop a common understanding of how some event or series of events occurs based on our examination of participants’ knowledge and experience of that event.
Using grounded theory often involves a series of coding activities (e.g. open, axial, selective or theoretical) to help determine both the main concepts that seem essential to understanding an event, but also how they relate or come together in a dynamic process.
Constant comparison is a tool often used by qualitative researchers using a grounded theory analysis approach in which they move back and forth between the data and the emerging categories and the evolving theoretical understanding they are developing.

Resources for learning more about Grounded Theory

Chun Tie, Y., Birks, M., & Francis, K. (2019). Grounded theory research: A design framework for novice researchers .

Gibbs, G.R. (2015, February 4). A discussion with Kathy Charmaz on Grounded Theory .

Glaser, B.G., & Holton, J. (2004, May). Remodeling grounded theory .

Mills, J., Bonner, A., & Francis, K. (2006). The development of Constructivist Grounded Theory .

A few exemplars of studies employing Grounded Theory

Burkhart, L., & Hogan, N. (2015). Being a female veteran: A grounded theory of coping with transitions .

Donaldson, W. V., & Vacha-Haase, T. (2016). Exploring staff clinical knowledge and practice with LGBT residents in long-term care: A grounded theory of cultural competency and training needs .

Vanidestine, T., & Aparicio, E. M. (2019). How social welfare and health professionals understand “Race,” Racism, and Whiteness: A social justice approach to grounded theory .

Corbin, J., & Strauss, A. (2014). Basics of qualitative research: Techniques and procedures for developing grounded theory . Sage publications. ↵
Charmaz, K. (2014). Constructing grounded theory . Sage Publications ↵
Gallicano, T. (2013, July 22). An example of how to perform open coding, axial coding and selective coding. [Blog post]. https://prpost.wordpress.com/2013/07/22/an-example-of-how-to-perform-open-coding-axial-coding-and-selective-coding/ ↵
Harrington, A. (2012). The fall of the schizophrenogenic mother. The Lancet, 379 (9823), 1292-1293. ↵

A form of qualitative analysis that aims to develop a theory or understanding of how some event or series of events occurs by closely examining participant knowledge and experience of that event(s).

starts by reading existing theories, then testing hypotheses and revising or confirming the theory

a paradigm based on the idea that social context and interaction frame our realities

when a researcher starts with a set of observations and then moves from particular experiences to a more general set of propositions about those experiences

Part of the qualitative data analysis process where we begin to interpret and assign meaning to the data.

An initial phase of coding that involves reviewing the data to determine the preliminary ideas that seem important and potential labels that reflect their significance.

Axial coding is phase of qualitative analysis in which the research will revisit the open codes and identify connections between codes, thereby beginning to group codes that share a relationship.

Selective or theoretical coding is part of a qualitative analysis process that seeks to determine how important concepts and their relationships to each other come together, providing a theory that describes the focus of the study. It often results in an overarching or unifying idea tying these concepts together.

Constant comparison reflects the motion that takes place in some qualitative analysis approaches whereby the researcher moves back and forth between the data and the emerging categories and evolving understanding they have in their results. They are continually checking what they believed to be the results against the raw data they are working with.

Trustworthiness is a quality reflected by qualitative research that is conducted in a credible way; a way that should produce confidence in its findings.

claims about the world that appear scientific but are incompatible with the values and practices of science

Share This Book

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Publications
Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

Advanced Search
Journal List
Am J Epidemiol
PMC10160766

Novel Methods for Leveraging Large Cohort Studies for Qualitative and Mixed-Methods Research

Associated data.

Qualitative research methods, while rising in popularity, are still a relatively underutilized tool in public health research. Usually reserved for small samples, qualitative research techniques have the potential to enhance insights gained from large questionnaires and cohort studies, both deepening the interpretation of quantitative data and generating novel hypotheses that might otherwise be missed by standard approaches; this is especially true where exposures and outcomes are new, understudied, or rapidly changing, as in a pandemic. However, methods for the conduct of qualitative research within large samples are underdeveloped. Here, we describe a novel method of applying qualitative research methods to free-text comments collected in a large epidemiologic questionnaire. Specifically, this method includes: 1) a hierarchical system of coding through content analysis; 2) a qualitative data management application; and 3) an adaptation of Cohen’s κ and percent agreement statistics for use by a team of coders, applying multiple codes per record from a large codebook. The methods outlined in this paper may help direct future applications of qualitative and mixed methods within large cohort studies.

Abbreviations

There are more things in heaven and earth, Horatio, than are dreamt of in your [questionnaire]. Adapted from Hamlet ( 1 )

Epidemiologists are experts at measuring quantitative traits across populations using assessment instruments that require study participants to choose from a set of categorical response options ( 2 ). Questionnaires enable researchers to efficiently accumulate vast amounts of quantitative data from large samples. However, such “checkbox epidemiology” can frustrate study participants and investigators alike. To minimize participant burden and maximize participation, questionnaires are usually kept as brief as possible ( 3 ). Investigators often “leave on the table” questions of interest they were unable to squeeze into a questionnaire. When questionnaires require participants to distill complex life events into checkbox responses that fail to capture the totality of their experiences, misclassification and missing data can ensue. Another limitation of relying exclusively on quantitative measures is the lost opportunity for unexpected insights from participants: Investigators tend to query about only the domains for which there already exist questionnaire instruments ( 4 ). In unusual and fast-changing times, such as during a pandemic, it may be difficult to anticipate all the exposures and outcomes experienced by study populations that will inform public health and medicine. Further, when circumstances are evolving rapidly, an instrument designed at one moment in time may prove outdated by the time it is fielded.

While the superiority of quantitative or qualitative research is widely debated, both approaches have strengths and weaknesses ( 4 , 11 ). Recent studies have leveraged the strengths of both methods through mixed-methods research ( 12 ). However, sample size can be a challenge for mixed-methods studies because, traditionally, quantitative research demands large samples to detect statistically significant differences and generalize findings, while qualitative research samples focus on depth and rarely exceed 200 participants ( 9 , 13 ).

Our objective was to develop a methodological approach for analyzing open-ended qualitative data collected from a large participant sample. In this paper, we describe a novel process which was developed and used to apply qualitative research methods to a series of coronavirus disease 2019 (COVID-19) questionnaires returned by 58,614 participants in 3 US longitudinal studies (Nurses’ Health Study II, Nurses’ Health Study 3, and the Growing Up Today Study), of whom 32,947 (56%) contributed free-text comments. (See Web Appendix 1, available at https://doi.org/10.1093/aje/kwad030 , for more details on application of this method to our specific study, including participant flow and survey response (Web Appendix 1, Web Figure 1) and characteristics of respondents and nonrespondents (Web Appendix 1, Web Table 1).)

Design and positioning of open text boxes

Electronic questionnaires can include any number of free-text comment boxes with prompts of varying specificity. For example, the baseline questionnaire of our COVID-19 survey included 4 comment text boxes: 2 unprompted boxes labeled “Comments” following questions about COVID-19 symptoms and diet-related questions; a specific prompted box about personal protective equipment (PPE) that was tied to a study aim (“Please include any information about your use of improvised, non-standard PPE”); and a more general prompted box at the end of the questionnaire (“We are interested in learning more about your experiences during this pandemic. Please add anything else you would like to tell us here.”). The length of the free-text comments should not be restricted; in reviewing tens of thousands of comments, very few were longer than a short paragraph. The placement of the comment boxes after specific items is likely to invite comments related to those questions; for example, we received many comments about immune-boosting supplements in the unprompted box that followed the dietary assessment.

Overview: codebook development and qualitative data analysis

Qualitative content analysis ( 14 , 15 ) can be used to interpret the meaning of participants’ free-text comments, using codes (topic labels) derived and assigned by trained individuals reviewing the text (coders) based on their consensus interpretation of the text. This involves an iterative process performed by the coding team consisting of coding, memoing, calculating interrater reliability (IRR), and testing and revising a codebook dictionary. This process continues until data saturation is reached (i.e., when no new information or insights emerge from reading additional records) ( 16 ). Coders purposively sample participants across groups (e.g., age, sex, race/ethnicity, geography, or exposures of interest (e.g., occupational or pregnancy status)) within the cohort to ensure saturation is based on a representative sample of the data ( 17 ).

Coding: open, axial, and selective

After open coding, axial coding occurs as coders convene to compare their individual preliminary codes, explore relational patterns qualitatively detected among the codes while reviewing the text responses, and organize them under broader categories, called “parent codes.” Our initial codebook sought to broadly capture all emergent themes relevant to the pandemic, resulting in a codebook dictionary with over 150 codes. However, coders in later projects with more focused aims (e.g., experiences of pregnancy among health-care workers during the pandemic) created their own codebooks with a limited number of focused codes. These “parent codes” and the “codes” nested under them formulate the first version of the codebook dictionary. The coders establish definitions for each code, articulate inclusion and exclusion criteria (when to use and not use each code), and identify an example of correct application. Table 1 depicts an example of several codes that were organized under the parent code of “Child Care and Concern” in our COVID-19 study.

Abbreviations: COVID, coronavirus disease 2019; NYC, New York City.

a Individual codes nested under the parent code “Child Care and Concern” and the corresponding definitions for each code, inclusion and exclusion criteria, and an example of correct application.

Once the first version of the codebook dictionary is created, coders proceed to selective coding. In this third stage, coders use the codebook dictionary to assign codes to both previously reviewed and newly sampled records. While the primary focus of selective coding is not to create new codes, coders make note of new themes that emerge to discuss with the coding team after each round of coding, which may result in revisions to the codebook dictionary. After each round of revision, more records are sampled and coded, and previously reviewed records are reviewed again to apply any changes in each iteration of the codebook dictionary. For efficiency, if the round of revision only modified existing codes (e.g., redefined the definitions of existing codes and inclusion/exclusion criteria, renamed codes, merged codes) without adding new codes, then coders rereviewed only records that had been assigned modified codes. In cases where distinct new codes were added, all previously reviewed records were rereviewed. IRR statistics are calculated in the selective coding phase to evaluate each iteration of the codebook dictionary. Specifics of the IRR calculation and codebook revision process are detailed below.

Calculating interrater reliability: the IRR application

The quality of the codebook dictionary is monitored by frequent checks of IRR during the selective coding phase. Low IRR results prompt revision of the codebook dictionary. Two measures of IRR are commonly used:

For both IRR statistics, we scored agreement for codes at the participant level—that is, across all free-text comments provided by each participant. (We use “record” to refer to all available qualitative responses provided by a given participant across all free-text comment boxes.) An alternative would be to score agreement across each free-text comment box separately. The IRR statistics were calculated as the arithmetic means between pairs of coders, across all coders for each code, and as a grand average across all coders and codes. These IRR statistics have varying functions in developing the codebook, training new coders, and reporting final codebook reliability.

Testing and revising the codebook dictionary

The iterative process of developing the codebook dictionary in the selective coding phase follows the cycle depicted in Figure 1 .

An external file that holds a picture, illustration, etc.
Object name is kwad030f1.jpg

First, for each iteration of the dictionary, coders apply any new codes to both the previously coded records and a new set of records. IRR is tested through “double-coding” of new records by multiple coders. For example, each of the 5 coders in our study was assigned 200 records to code, plus 20 records from each of the other 4 coders’ sets of 200 assigned records (for a total of 280 records reviewed by each coder). Thus, a total of 1,000 records were reviewed by the coding team, and 100 of these 1,000 records were coded by all 5 coders to test the IRR.

This iterative process results in multiple versions of the codebook dictionary. By design, data saturation cannot be attained until the codebook dictionary is finalized. Once the coders reach data saturation, all themes have been identified and captured in the codebook dictionary and no further records are required to be sampled and coded. However, investigators may continue to code records to increase representation of their sample; we chose to code at least 5% of the total records with free-text comments, sometimes oversampling particular groups to improve our contextual knowledge of those groups.

Quantitative analysis usually compares participants who were in/eligible or who did/did not return a questionnaire or respond to a particular item. The sampling frame of an epidemiologic study affords the opportunity to compare the characteristics of participants who did and did not contribute comments, or whose comments were or were not selected for coding; this is not typical of qualitative studies but can be leveraged to further understand the data and the context of comments analyzed (e.g., see Web Table 1 in Web Appendix 1).

Developing the codebook application

To store, organize, and assign codes from the codebook dictionary to records and to quantify data at different levels of codes, we developed a new qualitative data management application using Microsoft Excel (Microsoft Corporation, Redmond, Washington), which we named the Codebook Application. A spreadsheet template for the Codebook Application and detailed information about its use are included in Web Appendices 5–7; a video guide to the codebook is included in Web Video 1. Commercial software that could be used for this process includes NVivo ( 24 ). However, reported drawbacks of NVivo include cost and time-consuming learning processes due to the overwhelming amount of functions ( 25 ). Unlike NVivo, our application is free to use and does not require coders to have proficient knowledge of any software (including Excel) to load thousands of participant records (including both quantitative data and free-text comment boxes) into the workbook, assign codes, and calculate IRR. The main skills required for our Excel-based Codebook Application are the ability to organize codes and parent codes in a table, select labels from a drop-down box, and copy and paste cells from one worksheet to another. In addition, while NVivo and similar software allows concurrent multiuser access so that multiple coders can code the same records at the same time, this collaboration depends on Internet connectivity ( 26 ). The Codebook Application integrates with our IRR Application, which, as previously discussed, innovates upon existing IRR methods (i.e., Cohen’s κ, Fleiss’ κ, Krippendorff’s α) to accommodate various numbers of codes assigned per unit of text, multiple coders, and large data sets.

The Codebook Application allows coders to import participant records (identification and comment fields) into a spreadsheet and search for codes to assign to those records from drop-down menus that enable easy location of codes (in Web Appendix 6, Web Figure 4 shows a sample codebook entry; Web Figure 5 gives an example of parent and child codes; and Web Figures 6 and 7 depict the linkage of worksheets within the Codebook Application). Linked spreadsheets calculate IRR statistics, enabling coders to measure the quality of the codebook and identify inconsistently applied codes (Web Appendix 2).

Memoing: a practice of self-reflexivity

While analyzing comments, the coding team members record their thoughts and interpretations of the data through the process of memos. Memos, like a journal entry, include reflections on the analysis process, questions about ambiguous data, ideas about the codebook, and interpretations of and connections formed between larger themes revealed in the data. Memos aid in the process of developing consensus in the application of codes and code definitions between coders, as they enable each coder to articulate individual thoughts and prompt the group to examine multiple perspectives and proposed theories.

Qualitative analysis and manuscript preparation

Once coding is complete, investigators can perform purely qualitative or mixed-methods analyses. The selection of manuscript topics may be driven by high-frequency codes (reflecting their salience) or by novel insights gleaned by the coders during the analysis process.

With a large codebook, investigators need to select the most relevant codes to include in their analysis. For example, although we created a broad, general codebook dictionary covering all pandemic-related topics, for an article regarding health-care workers’ use of PPE we chose to work with 5 codes under the parent code “PPE,” 2 codes under the parent code “Virus Spread Concern,” and 1 code under the parent code “Work Stressors.” Typically, relevant codes are presented in a table in the publication with their definitions.

Codes can be used in many ways, including: to count the frequency of codes; to identify patterns across codes; and to gain contextualized information through illustrative quotations for inclusion in the manuscript. For example, the “PPE policy” code was applied to 6% of active health-care workers’ records. We then examined the distribution of the code according to the quantitative domains captured in the questionnaire, including whether the participant had treated patients with COVID-19, whether the participant had adequate PPE access, and the COVID-19 mortality rate of the census region in which the participant lived. Finally, the following representative quotation was identified in a record assigned the “PPE policy” code:

Our hospital administrators told us ‘per the CDC’ we didn’t need N95s unless doing an aerosolizing procedure. They didn’t routinely provide them at first. Five nurses got sick from this one patient. Our charge nurse was fired for speaking up about PPE.

The baseline questionnaire included a prompted comment box about PPE, so the many responses about PPE were expected. Other responses, particularly those from the prompt “Tell us about your experiences during the pandemic,” were more surprising. Several insights detected through coding of early questionnaires prompted us to incorporate new quantitative questions in subsequent questionnaires regarding gratitude, furloughs, parenting/work conflicts, and discrimination against health-care workers.

The pluripotent nature of cohorts and open coding can yield many topics and treatments, ranging from purely qualitative approaches to mixed methods and largely quantitative analyses illustrated by quotations.

The application of qualitative research methods to participants’ free-text comments allows participant perspectives to expand the breadth and depth of inquiry. The themes that emerge are probably broader than the restricted topics covered by most questionnaires; they inform the interpretation of quantitative data, permitting a triangulation across different types of data. Qualitative themes may suggest new hypotheses and prompt future data collection to test these hypotheses. Advantages of applying this approach in large studies over small, focused samples typical of qualitative research are the pluripotent comparisons and diverse perspectives enabled by the sheer number of respondents; for example, contrasts can be made according to factors such as demographic characteristics, occupational status, parental status, geographic region, and local rates of disease. Depending on the data set, it may be possible to explore the experiences of several intersectional marginalized identities. In this way, applying qualitative methods to a large study sample, as described here, can be analogous to conducting multiple qualitative studies in one, increasing efficiency and breadth of insights gained from a single participant population and data source (e.g., questionnaire). One limitation of the large sample-based qualitative research approach we propose here is its inability to probe participants’ responses in real time. Traditional qualitative research projects with small-to-modest sample sizes can achieve greater depth of insight by responding to participant cues during interviews and focus groups to probe topics/themes that arise—something that is impossible in the context of a large survey.

There is an inherent tension between breadth and depth; the application of qualitative methods to large surveys carries both the best and the worst of its parent disciplines. In a 1993 editorial, Britten and Fisher noted, “There is some truth in the quip that quantitative methods are reliable but not valid and that qualitative methods are valid but not reliable” ( 27 , p. 271). While qualitative research employs IRR to improve reproducibility, there are ongoing philosophical and practical debates about the purpose, choice, limitations, and interpretation of IRR statistics; these considerations are discussed in Web Appendix 8. The philosophical objection to IRR statistics is that they preference consensus above discovery; this debate also occurs regarding the handling and value of outlier data in quantitative sciences. Too much agreement among coders can reflect a lack of diversity of viewpoints and limit the generalizability of the codebook. Disagreements between coders may be as valuable as their consistency ( 28 ). There are also practical exceptions to IRR statistics. For example, critics of IRR statistics note their vulnerability to factors such as the length of the text segment, the number of codes within a codebook, the frequency with which codes are applied, and asymmetrical application of codes between coders ( 29 ). Thus, the importance of IRR may depend on the aims or context of a study. Ultimately, the best use of IRR in qualitative research may lie in the internal process of improving the codebook and training new coders.

Consensus and discovery are both involved in the research approach proposed here, but their contributions and value vary by stage. During the process of developing the codebook dictionary, discovery and consensus run in parallel, driving the iterative nature of the codebook dictionary revision process. Through discovery, new codes and identification of patterns are proposed by individual coders. By consensus, these discoveries are refined and incorporated into the codebook dictionary. Consensus ensures that all coders agree on the definition and proper application of the codes. Whereas quantitative researchers often identify and remove outliers for analysis, qualitative researchers commonly seek out and give voice to the apparent outliers. Information gained by probing discordant responses can provide valuable insights and ensure that the findings reported from qualitative or mixed-methods research accurately reflect the nuances of the lived experiences of participants.

The work detailed here adapted and extended traditional qualitative research techniques, typically used on modest-sized samples ( 9 ), for application to large questionnaires with prompted or unprompted free-text comment boxes. To our knowledge, this approach is novel in large epidemiologic studies. Adding a qualitative research component to large surveys may be especially useful in research endeavors at the “edge” of our knowledge: Novel situations (e.g., a pandemic), emerging new diseases (e.g., “long COVID” or the new Diagnostic and Statistical Manual of Mental Disorders diagnosis of “prolonged grief disorder” ( 30 )), cases where investigators realize that existing instruments map imperfectly onto complex phenomena, and the “lived experience” of people may be hard to capture in a 10-item survey. Qualitative research allows for unexpected findings and adaptation. From data collection to codebook development to IRR tests, these methods can facilitate the application of qualitative methods within large-scale population questionnaires to stimulate new breadth and depth of discovery, beyond what can be achieved with quantitative methods alone.

Supplementary Material

Web_material_kwad030, acknowledgments.

Author affiliations: Department of Chemistry and Chemical Biology, Harvard College, Harvard University, Cambridge, Massachusetts, United States (Katie Truc Nhat H. Nguyen); Division of Women’s Health, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts, United States (Katie Truc Nhat H. Nguyen, Jennifer J. Stuart, Jane Berrill, Janet W. Rich-Edwards); Department of Epidemiology, Harvard T.H. Chan School of Public Health, Harvard University, Boston, Massachusetts, United States (Jennifer J. Stuart, Bizu Gelaye, Janet W. Rich-Edwards); Department of Sociomedical Sciences, Mailman School of Public Health, Columbia University, New York, New York, United States (Aarushi H. Shah); NYU School of Global Public Health, New York University, New York, New York, United States (Madeline G. West); Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts, United States (Jane Berrill); Chester M. Pierce Division of Global Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, United States (Bizu Gelaye); and Department of Psychiatry, School of Medicine, Boston University, Boston, Massachusetts, United States (Christina P. C. Borba).

C.P.C.B. and J.W.R.-E. are co–senior authors.

This work was supported by the National Institute for Occupational Safety and Health, Centers for Disease Control and Prevention (award 75D30120P08670); the National Institutes of Health (award T23HD049339-15); and the Radcliffe Research Partnership.

Our public websites ( https://nurseshealthstudy.org/ , https://www.nhs3.org/ , and https://gutsweb.org/ ) include brief descriptions of the Nurses’ Health Studies and Growing Up Today Study cohorts, all questionnaires, and a description of resource-sharing procedures. An automated online form requests that investigators applying for data briefly describe their study’s hypothesis and aims, the variables needed, etc. Requests are presented to the cohort investigator meetings every other week, and replies are provided within 2 weeks. In general, if the cohorts have the data/resources required for a proposal, data-sharing is approved.

The views expressed in this article are those of the authors and do not reflect those of the National Institute for Occupational Safety and Health, the National Institutes of Health, or the Radcliffe Research Partnership.

Conflict of interest: none declared.

IMAGES

Grounded theory analysis using axial and selective coding
Coding matrix for qualitative data
How To Do Open, Axial, & Selective Coding in Grounded Theory
Coding in Qualitative Research by academiasolutionaus
Axial and selective coding of data
Coding Qualitative Data: A Beginner’s How-To + Examples

VIDEO

Fundamentals of Qualitative Research Methods: Data Analysis (Module 5)
Fundamentals of Qualitative Research Methods: Developing a Qualitative Research Question (Module 2)
Fundamentals of Qualitative Research Methods: Interviews (Module 3)
Qualitative Research for Public Health and Clinical Investigation
Telling a Complete Story with Qualitative and Mixed Methods Research
Theories of selective attention

COMMENTS

How To Do Open, Axial and Selective Coding in Grounded Theory
Selective coding is the last step in grounded theory, where you connect all your categories together around one core category.In doing so you define one unified theory around your research. Selective coding occurs later on in your research and connects categories you have developed from your qualitative data in previous coding cycles, such as axial coding.
Open, In Vivo, Axial, and Selective Coding
Open, axial, and selective coding are three distinct processes used in qualitative research, particularly in the field of grounded theory. They involve the systematic analysis and categorization of data to identify patterns, themes, and relationships. Open Coding: Open coding is the initial stage of qualitative data analysis.
What is Selective Coding?
Purpose of selective coding in qualitative research. When a study employs a grounded theory analysis, the researcher applies open codes (i.e., discrete units of meaning) and axial codes (i.e., relevant categories that are key to the research question) to the data.. For example, if you have a study that looks at smartphone usage, then the process of open coding might have produced initial codes ...
Beyond the Paradigm Conflicts: A Four-Step Coding Instrument for
Grounded theory (GT) is a method for generating theory through a systematic, iterative, and rigorous data collection and analysis process (Glaser, 1978).Glaser and Strauss introduced GT by publishing the book The Discovery of Grounded Theory: Strategies for Qualitative Research. Glaser and Strauss (1967) developed GT to strengthen qualitative research by offering analysis strategies to develop ...
A Design Framework for Novice Using Grounded Theory Methodology and
The process of selective coding in the current research primarily involved storytelling of the main ideas extracted from undertaking open and axial coding. ... In some qualitative research, the purpose of coding data is to arrange and systemise the overall ideas, concepts, and generate categories like the ones generated in the current research. ...
PDF The SAGE Encyclopedia of Qualitative Research Methods
Page 7 of 9 The SAGE Encyclopedia of Qualitative Research Methods: Codes and Coding Focused, Integrative, and Selective Coding The move from open coding to a more focused coding is not a clearly defined step. Many caution that although there are general guidelines that indicate a progression in the coding process from identifying new concepts ...
The Living Codebook: Documenting the Process of Qualitative Data
Transparency is once again a central issue of debate across types of qualitative research. Ethnographers focus on whether to name people, places, or to share data (Contreras 2019; Guenther 2009; Jerolmack and Murphy 2017; Reyes 2018b) and whether our data actually match the claims we make (e.g., Jerolmack and Khan 2014).Work on how to conduct qualitative data analysis, on the other hand, walks ...
What is Grounded Theory?
Core components of grounded theory. Grounded theory, as a research methodology, consists of several core components that guide the research process, from data collection to the development of a final theoretical framework. These components are interrelated, each influencing and shaping the others in a dynamic, iterative process.
A Guide to Coding Qualitative Research Data
The primary goal of coding qualitative data is to change data into a consistent format in support of research and reporting. A code can be a phrase or a word that depicts an idea or recurring theme in the data. The code's label must be intuitive and encapsulate the essence of the researcher's observations or participants' responses.
Codes and Coding
Abstract. Qualitative research is built on codes, and researchers must master the processes for creating codes and drawing insights from their analyses. In this chapter, the author discusses the various types of codes and approaches to coding. The quality of the output of qualitative data analysis is dependent on codes and the coding process.
PDF The Art of Coding and Thematic Exploration in Qualitative Research
Open, axial and selective coding of collected data results in the creation of theory, leading the researcher to construct deeper theoretical meaning. This method of coding provides researchers with ... Coding Qualitative research generally and Grounded Theory Methods specifically is an inductive, not a deductive, approach to qualitative ...
PDF Coding in Qualitative Research
Coding by Qualitative Design. Several types of qualitative approaches/designs. Ethnography, Phenomenology, Grounded Theory, Narrative Research, Case Study, Qualitative Descriptive. We will go through an example "chunk" of data, and code it together as if it were a Grounded Theory study.
Grounded theory research: A design framework for novice researchers
Figure 1. Research design framework: summary of the interplay between the essential grounded theory methods and processes. Grounded theory research involves the meticulous application of specific methods and processes. Methods are 'systematic modes, procedures or tools used for collection and analysis of data'. 25 While GT studies can ...
Systematic Coding in Qualitative Research
Methodology. This paper was presented at Uta Gerhardt's session "Issues in Qualitative Data Interpretation" (Research Committee "Logic and Methodology" - RC33) of the International Sociological Association's World Congress in New Delhi, India, in 1986. It is a much shortened version of a chapter, "Codes and Coding," in my book.
(PDF) Exploration of Coding in Qualitative Data Analysis: Grounded
Abstract. This study tries to identify, define, an d analyze the coding techniques that the grounded theory researchers. follow when they develop the qualitative research. Grounded theory is a ...
Selecting a Grounded Theory Approach for Nursing Research
Introduction. Grounded theory is a research approach that appeals to nurses for several reasons. Grounded theory helps nurses to understand, develop, and utilize real-world knowledge about health concerns (Nathaniel & Andrews, 2007).In practice, grounded theories enable nurses to see patterns of health in groups, communities, and populations and predict health and practice concerns in nursing ...
Coding Qualitative Data: How to Code Qualitative Research
Coding is the process of labeling and organizing your qualitative data to identify different themes and the relationships between them. When coding customer feedback, you assign labels to words or phrases that represent important (and recurring) themes in each response. These labels can be words, phrases, or numbers; we recommend using words or ...
AI is Transforming Qualitative Research Coding
With coding. Coding is the essential link connecting researchers and the insights buried within mountains of qualitative data. It's the process of categorizing, organizing, and extracting meaning from open-ended responses. Through coding, researchers can identify patterns, themes, and trends, ultimately providing a coherent narrative for the data.
18.6 Grounded theory analysis
Constant comparison is a tool often used by qualitative researchers using a grounded theory analysis approach in which they move back and forth between the data and the emerging categories and the evolving theoretical understanding they are developing. Chun Tie, Y., Birks, M., & Francis, K. (2019).
Novel Methods for Leveraging Large Cohort Studies for Qualitative and
Qualitative research methods, while rising in popularity, are still a relatively underutilized tool in public health research. ... Stage 1: selective coding of records with codes outlined in the codebook dictionary. "Double-coding" of records for IRR checks takes place. For example (B), each coder codes 280 records, the coding team reviews ...
The Art of Coding and Thematic Exploration in Qualitative Research
Coding in qualitative research enables researchers to identify, organize, and build theory. The roles of open, axial, and selective coding are critical to achieving the research goals of a study ...
What is Open, Axial, and Selective Coding?
Open coding, axial coding, and selective coding are steps in the grounded theory method of analyzing qualitative data. Learn more about open, axial and selec...