Thematic analysis: data wrangling in design

Wave of the Future | Source: VM Software

Thematic analysis is a way to understand qualitative data quantitatively, especially when there’s lots of it. It works by interpreting meaning from individual data points called fragments to create themes. According to Clarke and Braun, the method is unusual because it is ‘unbounded by theoretical commitment’. With proper application, thematic analysis reveals connections across many forms of data that are self-evident and useful.

Why is it useful in UX Research?

The style of UX studies depend on a few things:

1. Type of product.
2. Stage in the design process.
3. Resources available.

These can fluctuate wildly during the course of the project and as a result, studies can lose their value if they aren’t properly integrated with design direction. By connecting themes across studies, the value of data can compound. Themes provide a framework at the level of data fragments and allows the designer to build on existing insights, even if the studies vary in style.

This framework can provide design direction as well as understanding. If a design brief is nebulous or the problem is hard to define, thematic analysis can be used to categorise the problem space methodically by comparing discovered themes quantitatively. This comparison quantifies what participants collectively consider as the most important themes of an experience.

Why is it useful for Archetypes in Design?

Thematic analysis can be used as a catalyst for generating meaning in a framework of archetypal images. As discussed in the previous article of this series, archetypes are the best pattern to understand something. By mapping data fragments according to themes, the designer builds archetypal images from the bottom up. This process reveals the best pattern for understanding the customer experience. When applied to a product, this illuminates merits and flaws through a set of archetypal behaviours observed during the user-product relationship.

What’s the process?

Because thematic analysis has been used in many disciplines, there are lots of variations of the process. For the design context, A simplified version of Clark and Brauns 6 phase process strikes a nice balance between structure and freedom. I’ve come up with the Six Cs as mnemonic: Comb, Clean, Code, Combine, Classify and Communicate.

1. Comb

The comb phase is a quick skim of the data to become familiar with what sort of tags (words or short phrases that capture fragment meaning) are best suited across a set of data fragments. Bias can form quickly if you begin applying tags based on the first fragments you saw. Researcher bias is difficult to avoid in qualitative research, but we avoid as much as we can through combing and cleaning.

Some tips for the comb phase:

• Team members should Do this by themselves. It’s good to get a quick focused impression to begin. Discussion will come in phase 2
• Have a notebook handy to write down potential tags and keep track of what data fragments inspired tags ideas. It doesn’t have to be perfect, just grab a few to assist the conversation in phase 2.
• You don’t have to tag every fragment just yet. Some styles of thematic analysis use what’s called a ‘code bank’ to establish all codes before the official tagging process. This is a bit too structured for design research. In phase 3, you can make up tags on the fly!
• Keep an eye out for big fragments. each fragment should consist of one idea. if there are two things being said, it needs to be two separate fragments.

2. Clean

The clean phase is a peer review of your tags ideas with the other designers. Usually designers will find the similar ideas and tag them differently; a rose by any other name would smell as sweet. Comparing tags and referring to the fragments that inspired them reveals the best tags to name the ideas in fragments moving forward.

Some tips for the clean phase:

• Time box this phase. Discussions can veer into analytical philosophy very quickly when discussing the meaning of different tags. Give yourself 2 minutes to decide on the name for a tag and then move on.
• The meaning comes from the network not the node so there’s no need to strive for perfect tags.

3. Code

In the code phase, you apply tags to fragments. Its a grind but if you’re looking for answers, it’s worth it.

Some tips for the code phase:

• Use dropdown menus in excel or sheets to speed up the process. I prefer sheets
• Keep a notebook handy to jot down insights as you go. Some of the best understanding comes when the designer is right in the grind. these insights will be useful in phase 6 when it comes time to communicate your findings.
• Stick to no more than 10 seconds per tag. Anymore than that and you’re over thinking it or the fragment needs to be split as there’s are too many ideas in the fragment.
• Give yourself an estimated time by timing the 50 tags or so. It will help with motivation and planning out the project.

4. Combine

In the combine phase, tags are grouped into themes. Doing this by Card sorting in Miro is a treat, especially since you can copy cells straight into Miro as post-its.

Some tips for the combine phase:

• Its fine to nest tags in other tags. Sometimes a hierarchy of tags will emerge.
• You might want to get creative with how you visualise this in excel but don’t go overboard if you’re on a time crunch, the understanding will crystalise in phase 6 based on the most valuable points you wish to communicate.
• Don’t feel over-encumbered by the structure. The purpose of the process is to build a cohesive research narrative, not to create a perfect structure of quotes.

5. Classify

Classify is about defining the themes. Useful definitions are found through discussion. What tags are grouped under this theme? What are the ideas in the fragments tagged? Discussion of these questions will reveal shared meaning that creates a theme.

Some tips for the classify phase:

• Have a go at defining themes individually before discussing together. Everyone will have slightly or perhaps very different views and the best definitions will appear through discussion.

6. Communicate

The communicate phase is about explaining what’s been discovered from the process. This could be in the form of a speech, presentation, report or interpretive dance if you like. My team prefers reports during the week and interprative dance on Fridays. Make sure to record and store these artifacts as they are great for continual research of a product or experience. What’s important is you demonstrate the value of what has been discovered to others and to yourselves.

Some tips for the conclude phase:

• In explaining your research, you may find there are new avenues to pursue. A trend might clouded in uncertainty or a question begging to be answered. Include these problems in the artifacts and discuss with your team whether the client would find more research valuable or if it would benefit the customer experience.
• Explaining trends in through numbers are a great way to simplify and contextualize discoveries.
• Pivot tables can be handy in dissecting data further allowing a weighted understanding of specific participants or study types.
• If you’re unfamiliar with the data visualisation space, Cole Knaflics ‘Storytelling with Data’ is a great start. Knaflic focuses on the perspective of someone viewing data for the first time and how to make that as visually valuable as possible by eliminating clutter and emphasising what’s important.

Templates?

I was thinking of including a template for how thematic analysis can play out. If you’re keen for a template, I’d love to hear about it in the comments!

This was the second part in a three-part series:

1. Archetypes in Design: Understanding Behaviour as a Network
2. Thematic Analysis: Data Wrangling in Design
3. Archetypes in Agencies: Strategy through Behaviour (coming soon)