Welcome to another tips edition.
Last outing, I covered crafting insights from your qualitative data. Step 5 in our consulting take on Braun & Clarke's Thematic Analysis method.
This week jumps back to steps 2 & 3 aka 'Tag' & 'Theme'.
What?
It’s week 1 of your project. Helping a retailer understand the challenges around their customer experience. A decent understanding of the situation and context has formed. From the 7 interviews conducted across frontline and headoffice.
You have some high level themes already in mind. But, you know you need further detail to validate your thoughts… or catch what you might have missed!
Attention turns to carving meaning out of what was or wasn’t heard or seen.
To 'code' your data sources, you analyse the interview first (horizontal), then scan across the whole dataset (vertical).
Most consulting projects will do the reverse. Horizontal analysis to carve out themes across the entire interview dataset (e.g., recurring pain points raised by staff). Vertical analysis to dig deeper into patterns across roles (floor manager vs call centre rep). This can lead to anchoring or confirmation bias if not carefully managed.
Instead, go through each interview (or any other data source) one by one. Tag what you find interesting in the text. I.e., relevant to the 'exam question'. Think about how what you’ve coded relates to each other.
What are the key themes or conclusions from that conversation?
You’re now in a better place to see what’s repeated / infrequent / peculiar?
What are the key themes or conclusions from these conversations?
This is the critical starting point for your analysis. The soil in which your assumptions, biases and insights will either wilt or grow.
So What?
Theming is subjective.
It’s easy to become anchored to one theme. Especially If you haven’t created the right system to ensure you anchor your views in data instead.
Coding can be deductive (you work off a framework) or inductive (you decide as you go). It can be surface level (your interpretation of what the actual words mean). Or latent (your perspective on the underlying ideas and significance behind the words). Either way, it should always be within the context of the problem statement.
High-level to discover overarching themes. Granular to uncover nuanced insights within those themes.
Deductive using a pre-existing coding frame is what we try to do in consulting. It's more efficient. Using previous material/ reading to steer the coding (e.g., a CX quality assurance framework).
Inductive coding (without a pre-existing frame) is what we often end up doing. It takes time to find prior material to create a codebook!
Truth is, we need a balance. Project goals determine the balance. E.g., A codebook of risks for a deductive CX solution assurance project. An inductive approach for a CX strategy. Allowing fresh insights to emerge that team biases could miss.
"The definitions of codes arise from a reciprocal process of induction and deduction, involving steps taken back and forth between the data and the conceptual level."- Michael Huberman, Data Management and Analysis Methods.
What next?
Next time you're gearing up for qualitative analysis.
Don't skip the coding. It's tempting to stay at surface level themes. Intentionally develop codes tied to the context and goals. These can be consistently applied, and refined.
Go beyond the interview guidebook in your preparation. Timebox an activity with your team to create a flexible codebook. One that serves the project and leaves room for discovery.
If you're in a team, have several people review a source. If you can't, divide and conquer. But then review. Help learning and reduce bias.
A codebook isn't etched in stone. It's more of an orienteering map to keep you and your team aligned on the big signs and how you approach them in turn.
Also, if you're rocking Discy's smart tagger on your project. It now takes minimal time to create, maintain and refine.
"Coding is not just labeling, it is linking. It leads you from the data to the idea, and from the idea to all the data pertaining to that idea." - Johnny Saldaña, The Coding Manual for Qualitative Researchers