TLDR: Coding is the bridge between raw qualitative data and meaningful analysis. Start by immersing yourself in your data through pre-coding, then work systematically through one first-cycle coding method at a time (e.g. in vivo, emotion, process, or values coding), while writing memos to capture your ideas, questions, and emerging insights. In second-cycle coding, group related codes into broader, well-defined themes that answer your research question and tell a coherent analytical story. Avoid rushing to themes, confusing codes with themes, or simply paraphrasing participants. Good coding is iterative, reflective, and interpretive, and it’s where the real analytical work begins.
If youâve ever stared at a stack of interview transcripts wondering where on earth to begin, you are not alone. Coding is the first step for turning raw data into analysis with insight.
What Are You Actually Trying To Do?Â
What Exactly is a Code?Â
Memos: Your Running Conversation With the Data
- How you are responding to the data.
- Emerging categories, patterns, or themes.
- Possible links between codes.
- Connections you are starting to make with theory.
- Or anything else that pops into your head.
Memo writing is part of your audit trail, helping to demonstrate the transparency and trustworthiness of your interpretations and arguments. Memos are thinking tools that provide a space for you to be in dialogue with your data.
Getting Your Hands Dirty: Pre-Coding
Before jumping into complicated software, there is real value in printing out your data and working with a notebook, highlighters, sticky notes, and coloured pens. Touch the data.
Circle, highlight, or underline anything that catches your attention. Â As you read, ask: What is interesting here? Why is it interesting? How does it relate to my research concern? Write memos as you go, capturing your assumptions and first impressions.
First Cycle Coding Methods
- In Vivo coding uses the participantsâ own words as codes instead of using your own words to capture the meaning of a data segment.
- Attribute coding means assigning attributes to data such as age, gender, location etc. This ca be useful for later comparisons across subgroups.
- Values, attitudes, and beliefs (VAB) coding means labelling what participants value, how they react to things, and what they believe to be true. These three are coded together since they tend to occur together.
- Emotion coding is labelling emotions clearly present in the data (fear, anxiety, loneliness) without psychoanalyzing participants.
- Versus coding is identifying conflicts tensions or oppositions, whether between people, institutions, or concepts such as them versus us, then versus now, that versus this.
- Evaluation coding is capturing participantsâ judgments about something.
- Process (or action) coding involves assigning codes in the form of gerunds (âing” words like resisting, negotiating, or initiating). Coding in this way helps us to trace processes or sequences of action.
The golden rule across all of these: run one cycle of the washing machine at a time. Don’t mix coding types within the same pass, and keep writing hunches and memos throughout.
Second Cycle Coding
First cycle codes are like the parts needed to build something. They are not the finished product. Second cycle coding is about how everything fits together. Here you group similar codes, eliminate redundant ones, and choose umbrella headings that pull several codes together into emergent themes. This might mean devising an entirely new label to capture several codes, or elevating one existing code into an umbrella category.
In second cycle coding you:
- Promote an important, recurring code into a theme.
- Cluster similar codes together.
- Use thematic maps or tables to visualise relationships.
- Ask whether each candidate theme has a genuine central organising concept, clear boundaries, and enough supporting data.
- Name and define each theme, and describe how it fits into the larger story your analysis tells.
- Produce a final table: theme, definition, and illustrative examples.
Common Pitfalls
- Losing sight of the research question as the analysis develops.
- Too many codes masquerading as themes, or too few themes overall.
- Themes that overlap heavily or feel unrelated to one another.
- Themes so vague you would not be able to write a coherent paragraph explaining them.
- Mismatches between the data extracts chosen and the analytic claims being made.
- Paraphrasing data instead of analysing it.
- “Psychologizing” (speculating about why a participant said what they said, rather than engaging with what is actually in the data).
The Bottom Line
Coding is slow, iterative, and often unglamorous work. But it is also where the real thinking in qualitative data analysis happens. Resist the urge to rush toward themes; premature theme-generation almost always produces superficial results. Instead, immerse yourself in the data, run it through multiple coding cycles, write memos relentlessly, and construct themes that are analytical, interpretive and conceptual rather than descriptive labels.

