Theoretical Approaches to Qualitative Data Analysis

RadiasaTutorial Rti. (n.d.). Flat isometric concept illustration, email marketing analysis data folder. Vecteezy. https://www.vecteezy.com/vector-art/8143929-flat-isometric-concept-illustration-email-marketing-analysis-data-folder

Almost every action people take in life leaves behind a trail and produces some data. After we interview people about their thoughts, observe their actions or examine the words and media they produce, we could be analyzing large amounts of data about different topics in various formats. This is why we need theories to tell us how to categorize our data and where to start analysis.

Data Immersion

One crucial step in analyzing the data you have collected is the process of data immersion. Repeatedly immersing yourself in the data allows you to identify and interpret patterns that produce a greater understanding of the topic of study (Borken, 2021). It ensures that you grasp the overall themes present within the data and prevents your work from becoming too focused on a limited part. Therefore, you will find it helpful to use this process repeatedly to refine your research question and analyze your data.

Now, we will be going through two common approaches to qualitative data analysis: Grounded Theory and Phenomenology. Both methods offer ways to dive deeper into your data while also looking at the big picture of what it’s describing. Though not a comprehensive list, these brief overviews serve as an introduction to the diverse field of analyzing qualitative data.

Guiding Theories for Analysis

Grounded Theory

Grounded theory is an approach where you identify the patterns in the data collected and then generate a theory through interpreting the data (Glaser and Strauss, 1967).

When you use grounded theory to approach your research, you often start coding your data as you collect it and derive an initial set of codes. As you continue collecting and coding more data, you may find new patterns and label them with new codes. Then, it is important that you go back and apply these new codes to the whole dataset – including sections you already coded. This is because the new pattern you found may also be present in the previously coded sections, but you may not have been able to notice until you reached your current position in the dataset (Gibbs & Taylor, 2010).

Grounded theory requires you to refine your codes through cycles of data collection and coding. The codes that ultimately emerge from this process will become the basis of your theory.

Phenomenology

The goal of phenomenology is to understand how participants experienced the phenomenon you are studying. Therefore, you often capture the participants’ subjective “first-person views” of the phenomenon by using notable quotes and statements in the data you collected (Gibbs & Taylor, 2010). For example, a phenomenological study of youth unemployment may involve interviewing young people about their experiences of unemployment and their attitudes towards finding work.

To analyze data from a phenomenological approach, you often write descriptions regarding the environment’s influence on the participant. From the descriptions, you can derive themes that summarize the participants’ experiences. In the example above, you may code the interviews by identifying factors that influence young people’s view of work, such as the workplace environment, comparison with peers, and the de-stigmatization of unemployment. These factors capture “the ‘essence’ of the phenomenon” and guide you toward a theory of youth unemployment (Creswell, 2006).

Flexible Coding: An Alternative Approach

At some point, you might work with a group of peer researchers to conduct interviews, code them, and analyze a large amount of material. It can be laborious for each of you to generate codes from the ground up and derive a universal set of codes through constantly comparing coded data among yourselves. Even when you are working individually, you may find yourself facing such large amounts of data that it becomes difficult to refine and keep track of your codes based on every new section. 

This is where an alternative approach to grounded theory, called flexible coding (Deterding & Waters, 2018), can be helpful. Flexible coding suggests that you apply your theoretical assumptions to the data to identify the most relevant sections. You only code this subset of the data intensively. Consider the following steps as a guide:

Let’s see how flexible coding helps researchers extract the most helpful sections from a vast amount of data in a real-world example.

For example…

Hamm, Lisa, designer. Ambitious and Anxious: How Chinese College Students Succeed and Struggle in American Higher Education, Yingyi Ma, Columbia University Press, 2020. Front cover.

For her book Ambitious and Anxious: How Chinese College Students Succeed and Struggle in American Higher Education, sociologist Yingyi Ma and colleagues collected survey responses from over 500 Chinese students and educators and conducted semi-structured interviews with 67 of them. The team indexed the interviews according to the respondents’ type of involvement in higher education, which included Chinese college students in America, Chinese high school students, high school counselors in China, high school principals/heads in China, and foreign teachers in China (Ma, 2020). 

Indexing the transcripts enabled the researchers to find and analyze any particular type of involvement more easily. They chose to focus on themes that the surveys did not capture in detail, such as the social experiences of Chinese students in the U.S. As Ma put it, while the surveys help assess the proportion of Chinese students who have close American friends, the interviews “explore the meanings of what these students might regard as a close relationship.” (2020) 

One theme the interviews highlighted was students’ ambivalence towards party culture: “Many were aware that they ought to attend college parties to be fully socialized on campus, so they tried. When they did, however, they witnessed things they typically shunned as bad behaviors based on their prior socialization in China.” While the finding is summarized into a few sentences in the published work, it might be the result of comparing across dozens of interviews where students discussed party culture.

Refining your Research Question

Whatever approach you choose, the analysis process often involves revisiting your data as codes develop. Coding your data iteratively gives you an opportunity to refine your research question based on what you have found. This allows you to reflect on how your findings connect to the field of study and the impact your research will have (Angee, 2009). Nevertheless, if you are conducting the study to address a particular problem, it is important to keep that original intention in mind as you refine the question. Consider the following steps as a guide to refining your research question: 

1. Draft a preliminary research question knowing it will probably change as you gather and interpret your data
2. Frame your questions so that they generate data for different aspects of your core question
3. Use the data collection and initial analysis phases as spaces to practice reflective inquiry
4. Write and re-write your research question until you land on a final version (Angee, 2009).

The last step is particularly important because the re-drafting of your research question, over and over again, will help you construct a answerable and specific question. This interactive process produces research questions that focus on one issue and but also address multiple aspects of that issue.