What Is Thematic Analysis?
When you need to analyze qualitative data from transcripts or interviews, use a thematic analysis. You will see which meaning patterns, ideas, or topics are repeated to make valid conclusions.
Thematic analysis can be approached in different ways. Though, the sequence of six steps is the most widely used, including familiarization and coding. Then, you will proceed with the generation of themes and reviewing them. After that, you will have to define and name these themes and write up the results. In this way, you will be able to avoid minimizing confirmation bias in the outcomes.
Victoria Clarke and Virginia Braun initially used this sequence in psychology research. Thematic analysis there revealed its flexibility, and since then, it has been adapted to many fields of research.
When Do You Need Thematic Analysis?
When you need to conclude about the audience’s experiences, opinions, points of view, knowledge, or values, it is preferable to use a thematic analysis.
You apply it to different sets of qualitative data, like survey responses, profiles on social media, or interview transcripts.
Research questions you can answer with the help of thematic analysis may be the following:
- What do patients think about doctors’ qualifications in medical institutions?
- What are the customers’ experiences with online shopping?
- What non-expert ideas do your respondents have about water pollution and ways of its prevention?
- How is self-improvement treated in high school environments?
If you need to answer these questions, you have to make a survey or questionnaire for a group of participants you are interested in, collect data from them, and then analyze it. You will get as much flexibility as you need in interpreting this data, even if the data set is large. You will achieve that by sorting our broader themes and subjects into subdivisions.
Nevertheless, you may need more details and nuances here because the thematic analysis can be subjective. A researcher’s opinions and judgments can affect a lot of data interpretations, so you must be careful when choosing. Avoid paying more attention to the things that you want to find in the answers though they are actually not there or expressed rather vaguely.
Use Thematic Analysis with Different Approaches
You can choose a specific approach to your thematic analysis between the two options - inductive and deductive.
- An inductive approach is considered when the data can define the themes.
- A deductive approach means analyzing the data already having the assumed expectations from it that were obtained from existing knowledge or theories, so you collect the data based on these assumptions.
To choose the correct approach, you should be sure that you have enough theoretical basis for making assumptions about what themes you will find in the data. This consideration should be applied to the deductive approach. Or you may be sure that you are going to create your unique framework based on your specific findings and apply this idea to your inductive studies.
You can also choose between a semantic and latent approach. If you analyze the explicit content of the data, you apply a semantic approach. Though, if you want to read into the underlying assumptions and subtext, a latent approach is more appropriate.
To make the right choice, decide whether you want to know the stated opinions of your respondents (with the help of a semantic approach) or research the social context and their ideas about it (dealing with a latent approach).
Do Thematic Analysis in 6 Steps
After you choose thematic analysis as the core method of your data analysis and decide on the approach you will follow, do your analysis in the steps suggested by Braun and Clarke and applied by many researchers already.
1. Get Familiarized
When you start the thematic analysis of the collected data, overview everything that you have managed to collect. Look through the texts, transcribe audio, and take notes if you need to understand what you are planning to see while analyzing individual points of the data.
2. Code the Data
Create shorthand labels (codes) for all the sentences or phrases from different parts of the texts to be able to thoroughly describe the content.
Suppose you need to analyze the responses of community members aged 45-55 given to the questions and water pollution in their area and possible measures to eliminate it. You have conducted interviews, and you can see that one interview response is like the following:
“As for me, yes, I know that the river in our place is rather polluted, but I am not sure about the causes. I don’t know why this has happened. Probably some professionals know better, though I can never be sure that they will tell us what we want to hear but not the actual results. I cannot say that it always happens, but I cannot 100% trust them, either. It is a global problem, not only in our area, because everything is changing rapidly with all these technologies.”
So, you can code it as follows:
- acknowledgment of water pollution in the area
- lack of trust in expert opinions
- uncertainty
- making changes in terminology
You can even highlight the phrases from the interview that correspond to different codes in various colors to be sure that they describe exactly what code you can find in every specific part of the text. Try to be consistent. Look exactly via every interview transcript and highlight all the sentences that correspond to the codes defined by you. You can also add new codes if you see new ideas in every other interview you are analyzing.
After completing such coding, divide the data into groups corresponding to every specific code. In this way, you will obtain a generalized view of the common meanings and main ideas repeated via all the data sets.
3. Generate Main Themes
Continue looking over all your formulated codes and denote the specific patterns. You will generate the themes on their basis. Themes are more extended than codes, so several codes can make up one theme. Let’s look at how we can combine codes into themes in our example.
Such codes as “uncertainty,” “experts know better,” and “explanations can be alternative” equal the theme of “uncertainty.” You can connect the codes of “changes in terminology,” “lack of trust in expert opinions,” “resentment toward scientists,” and “uneasiness about possible authority’s control measures” into the theme of “distrust of professionals.” The theme of “misinformation” is also present here under the codes of “lack of scientific understanding,” “biased perception of media sources,” and “distorted facts.”
Here, you can also see that some codes need to be more specific and relevant because they appear only once throughout all the interviews, so you will decide to discard them. Some codes, on the contrary, can become themes because they are broad, such as “uncertainty.” Then, other codes can be included in it.
Your choices of codes and themes and their variations may also depend on what exactly we want to find out. Potential themes need to add something to meet your aims and research purposes.
4. Review Created Themes
You must ensure that the themes you have created are relevant and accurately represent all the useful data. Look at your data set again. Check whether you have not missed anything and whether all the themes you have composed are represented in the data. If the formulation is not clear, you can change the theme at this stage to make it work better.
If there is a problem with the theme's clarity, you can split it up or combine two themes. You can also discard something and create new themes in their place. Every change should serve usefulness and accuracy, anyway.
You may decide that “changes in terminology” do not fit “distrust,” but it is better to go to “uncertainty” because if a person is not sure about the terms, they are confused and not sure about the correctness of their assumptions. That does not mean they distrust anyone.
5. Define the Themes and Name Them
You have arrived at the final list of themes, so now you need to define and name them. To provide clear definitions, you have to formulate exactly what you meant by each theme and how it can help to understand the research outcomes. When you name the theme, try to pick a clear, short, and easily understandable name to ensure other researchers understand.
When you use the word “experts” in your codes and themes, define who you mean by this word related to this set of data and themes. Then you can rename “distrust of experts” into something like “distrust toward the authorities” or “distrust towards science.”
6. Write Up the Process and Results
Your thematic data analysis is complete now. Though, you need to write up the process and results. It has to be a common academic text with a clear introduction, a methodology section, a results section, and a conclusion. Indicate your research question, aims, and approaches you have used in the introduction.
The methodology section should include a description of how you collected the data (e.g., via open-ended questionnaires or semi-structured interviews) and conducted the thematic analysis. The section describing the results or findings has to describe every theme, how often it comes up, and what it means (with examples of evidence from the collected data).
The conclusion has to be about the main takeaways and about how well your thematic analysis has answered the research question.
Your example may lead to the conclusion that “distrust toward authorities” about water pollution in the region is common among people aged 45-55 with rather conservative views. The analysis has pointed out their uncertainty based on poor scientific knowledge and misinformation from the media.
Final Thoughts
You can see now that thematic analysis is a very effective method for analyzing qualitative data obtained from interviews and surveys.
However, you need to ensure that this method does not involve any bias or misinterpretation. That is why you need to follow the 6-step procedure of thematic analysis described in this article.
You can re-read the article every time you need to decide what analysis method you have to choose for your data processing. We hope that our tips and recommendations will be helpful.