Having collected your data such as audio or video files, pictures, written statements or notes you took during testing sessions, the next step is making sense of all the data you collected. When you’re evaluating data gathered from qualitative user research, your goal is usually to find surprising or interesting pieces of knowledge or patterns and themes you can later use to make informed decisions, figure out more about your users and build comprehensive personas. If you’re in the early stages of your research process and don’t have a fleshed out idea what patterns you dataset might look like using thematic analysis can be extremely helpful.
This method is a very good fit for preliminary exploratory research as it helps you make sense of data that seem unrelated at first glance. It can also help you to discover and frame specific problems, research questions, or emerging themes that can be used to conduct furthermore granular research.
Why should you do a thematic analysis?
When dealing with unstructured data, a thematic analysis will help you get the most out of your data: you can look for specific patterns you already suspect in your data while at the same time working in a more explorative manner where you don’t exactly know what you’re looking for from the beginning. No matter what kind of research you conducted, always try to sum up and evaluate what has been said as closely as possible.
If you gathered the data you’re evaluating yourself, you may feel like you already have a sufficient understanding of your data. It often feels like conducting surveys, interviews or user tests yourself gives you a pretty good idea of what your users want to tell you so you’re eager to start implementing your insights immediately. If you want to make sure you don’t overlook important facts or are influenced by certain biases you might have, it’s a good idea to do a planned analysis and go over your data in a systematic way.
Documenting your analysis process as well as preliminary insights will also help your colleagues and stakeholders retrace the steps you took to arrive at your conclusions which will help them understand your reasoning and decision-making process. While documentation takes time-especially if you work in a more explorative manner-it will also make your results more trustworthy.
How to run a thematic analysis
As is often the case in user research, thematic analysis is an iterative process. It will lead you from raw, unstructured data to having a deep understanding of themes that occur in your dataset. There’re six steps to conducting a thematic analysis:
- Get a general idea of your data: Depending on the kind of data you have, you might need to transcribe it before you start your analysis process. Having done that you can start going over all the available data to try to get a general feeling for your data and generate first ideas about structures that might exist within your data. If you already have certain themes you’re looking for in mind, be open for other patterns that may emerge from your data.
- Gather first ideas for codes: You will use your codes, to sum
upthings your users said. For example, if you note that users talk a lot about looking for the sign-up button, you could assign a code to statements concerning this problem. Codes should always be descriptions of what has been said and not include interpretations. You’ll later use your codes to organize your data into groups. Usually, two or three lines of your transcript will be summed up into one code. Having coded all your data, collect all pieces of data you assigned the same code and look for common statements or emerging patterns.
- Find themes: Continue looking for similarities within the statements you assigned certain codes. This is where you start interpreting your findings and try to sum up multiple codes into one theme or multiple themes into a bigger theme. Some codes may not fit into one theme with other codes and thus become a theme themselves. If codes don’t fit and aren’t interesting enough to justify their own theme, keep them in a miscellaneous theme for later interpretation.
- Go over your themes: make sure your themes make sense and that statements within a theme share certain characteristics or are similar to each other. Different categories should be distinct or at least overlap as little as possible. Eliminate contradictions within themes by splitting them or moving certain codes from one theme to another. Once you defined a set of relevant themes go over your uncoded data again to make sure you didn’t miss anything that fits one of your themes. You may also catch interesting statements you overlooked and discover new themes that way. This process is iterative in itself so you should stop once you feel like you’re not adding anything relevant anymore. If you want to make absolutely sure your interpretations aren’t influenced by some kind of bias you might hold, have other people review them, too.
- Name and describe themes: give your themes names that already hint at their content. When describing themes don’t only describe their content but also try to describe why this content is relevant or interesting. You may also want to describe how themes relate to each other and your research question. If you have trouble with this step you may have to take one step back and go over your themes again.
- Report your results: Report what you did and why you did what you did. Use the descriptions of your themes as the basis for your report. You can also user pictures, quotes, audio or video clips of your participants to support your findings. If you do this you should, of course, get their consent first. If you need more ideas about how to communicate your UX research findings take a look at our blog post concerning this topic.
One additional step you can think about after you’re done with your thematic analysis is scoring the themes you discovered depending on the severity of the problem. There are people who like scoring their themes to prioritize problems and quantify their seriousness as an aid for decision-making. On the other hand, there are also those who dislike it since every piece of data adds to the bigger picture.
When you’re taking a deep dive into your research data, it can be easy to get distracted by interesting but off-topic pieces of data. Make sure you answer the questions that were the reason for your research in the first place. Also, don’t lose sight of the bigger picture.
Do you need a tool to help you structure and tag qualitative research data? Try consider.ly–Get an invite!