Once you’ve completed your field work you’ll have a large amount of qualitative data. Depending on your research methodology this might be in the form of:

• Field notes: usually taken by hand during a research activity and then collated into digital format.
• Interview transcripts: a verbatim account of the interview created from interview recordings that are transcribed. This is very time intensive and can be expensive.
• User generated content can take the form of annotated photos or transcribed videos. Photos can be further annotated by the researcher to describe what they’re showing. Videos can be transcribed verbatim and researcher observations added.

To move forward in the design process, you’ll need to find a way to pull useful insights out of this data and summarise them. It’s vital to do this in a way that doesn’t introduce your own biases back into the data.

Our approach to qualitative data analysis (QDA) adapts academic methods that provide structure and rigour. While it is never possible to completely eliminate researcher bias, a structured approach helps to minimise its effect. It also enables other researchers or clients to follow the way we’ve approached the data and the basis of our findings.

Note:

For a rigorous approach in design research, it is important to be aware that taking field notes during a research activity effectively represents the first level of analysis. These notes would include statements and observations that the researcher considers relevant to their research questions.

Coding data

The first step in analysing is the coding of field notes. The process of coding is a way to order and organise the data collected during the research activity in order to identify themes and patterns

A code is a essentially a description of the data it is applied to, all data that fits this description is coded to create a group that shares this description. In practice, you work through your notes and highlight excerpts in order to attach codes to them. For this process we use QDA tools such as Dedoose or NVivo which group text excerpts that share the same codes.

Broadly, there are two directions of codes to apply to design research data, top-down codes and bottom-up codes.

Top-down codes

Top-down codes are informed by the researcher’s knowledge of a topic or the scope of a project (a-priori). You’ll decide on these codes before even conducting the research and these will help you structure the data towards what you want to find out.

For example, when researching people’s lived experience with insomnia you might decide to apply the code “attitude towards medication” because it is within your research scope to explore this or you know from literature that attitudes can influence medication adherence. This would allow you to collate all the data you have gathered on the attitude towards medication in one place.

Bottom-up codes

Bottom-up codes emerge from the data as you’re exploring it. For this type of coding it might be necessary to refine codes after a first round and then go through the data again.

For instance, in relation to the above example of attitudes towards medication in insomnia, some participants might tell you that they prefer other methods to help them sleep, such as calming teas or relaxation methods. In this case perhaps you first code the mention of teas as “herbal remedies” but then combine it with relaxation methods as “alternative methods”. During this stage an iterative approach is often necessary to get more familiar with the data and bring clarity into your structure. The more you do this, the more familiar you’ll be with the data and the clearer your structure.

Note:

During the coding stage more than one code is often applied to any excerpt. This leads to code co-occurrence, which indicates a relationship between two codes. If a co-occurrence is common (prevalent) within the data, this indicates that there might be a theme to identify.

Identifying themes in the data

The next stage of data analysis is about identifying prevalent commonality (i.e. patterns) in your data.

As mentioned before, QDA tools allow you to see how many excerpts have been coded with a single code and display the frequency with which certain codes co-occur. In order to identify themes, we select the most frequently co-occurring codes and search for further commonality between them.

To stick to our example from above, let’s assume that all of your participants have spoken about their attitude towards medication at some point (either because you have asked them directly or they mentioned it without prompt); such prevalence would mean that it is a theme within the data and it would inform a finding.

Exporting excerpts

All QDA tools allow exporting all the excerpts coded with one or more given code into one text file. In a final step, we export these text files and explore for commonalities within the excerpts and rearrange them into groups. This process is repeated until each group of excerpts shares a common theme distinct from all other groups. For example, if half of the times that “relaxation techniques” occurs in the data is has been related to the code “long-term treatment”, it would be necessary to analyse the excerpts that both codes are applied to and identify why this might be the case.

It could be that, for example, people are keen to use relaxation methods as an alternative treatment but state that they’re difficult to learn and adopt, therefore, they see it as a long-term approach but it might not impact their attitude towards medication in the short term. This would be one of your insights. Continue repeating this process until you’ve analysed all the data, and have a list of all the insights.

In the next section we’ll look at ways you can turn these insights into a format for dissemination and presentation.