Analysing your data

“I’ve coded everything. How do I get from codes to themes?”

Codes versus themes, and the move from describing to interpreting.

You coded every transcript. Then your supervisor said the themes are “just descriptions”, or that your headings “can’t be themes” — and you are not sure what is missing. The gap between a code and a theme is a specific move, and once you see it you can make it deliberately.

By Heidi Andersén, MD, PhD · Updated 18 July 2026 · 7 min read

“I am thinking the statement of the problem are my themes… yet my teacher told me it cannot be my themes.” — a student, stuck between codes and themes

The problemA pile of codes is not an analysis.

Coding feels like progress because it is visible: highlight, label, repeat, until every transcript is covered. But at the end you have an inventory, not a finding. The stuck moment arrives when someone asks “so what are your themes?” and your honest answer is a list of topics — the topics your questions were about. That is the tell: if a theme could have been written before you collected any data, it is a description of your interview guide, not a result of your analysis.

Why it mattersDescriptive themes are the classic revision.

“These themes are descriptive” is one of the most common things an examiner or reviewer writes on qualitative work, and it is hard to answer after the fact without redoing the analysis. Getting the codes-to-themes move right the first time is the difference between a findings section that reports what you asked and one that reports what you learned.

Common mistakesFour ways themes stay stuck as descriptions

Themes that mirror your questions. If your interview guide had five topics and you have five themes with the same names, you have summarised the guide, not the data.
Codes promoted straight to themes. A frequent code is not automatically a theme. A theme is a pattern of meaning that may cut across several codes.
Bucket themes. A heading like “Challenges” that holds anything vaguely negative is a filing cabinet, not an interpretation.
No trail back to the data. A theme no one can trace to specific coded extracts cannot be checked by a second coder — or defended in a viva.

The minimal explanationCode, category, theme

A code labels what a segment of data is about — close to the participant’s words. “Fear of asking for help” is a code.

A category groups related codes. “Barriers to seeking support” might gather several fear- and workload-related codes.

A theme makes an interpretive claim about a pattern: it says something happens, and suggests why it matters. “Support is available but asking for it feels like admitting failure” is a theme — it has tension, it explains, and it could be wrong, which is what makes it a finding.

The move from descriptive to analytic is the move from naming a topic to making a claim about it. Ask of each candidate theme: does this tell the reader something they could disagree with? If not, it is still a category.

Worked exampleFrom extracts to a theme

First-year nurses, moral distress

Coded extracts:

“I didn’t want to bother the senior nurse again” → code: reluctant to ask
“asking twice felt like I couldn’t cope” → code: asking = incompetence
“there was help, I just didn’t take it” → code: support unused

Category: help-seeking behaviour.

Descriptive theme (weak): “Participants faced challenges asking for help.” True, but it could have been written before the study.

Analytic theme (strong): “Support was available, but seeking it was experienced as an admission of not coping — so nurses rationed their own help.” This names a pattern, explains the mechanism, and is falsifiable against the extracts.

Notice the analytic theme still points back to the three coded extracts. That trail is what lets a second coder — or your supervisor — check the claim against the same data.

Checklist & templateMake the move deliberately

Downloadable · codes-to-themes worksheet

The candidate-theme worksheet

List the codes you think belong together.
Name the category (the topic they share).
Write a claim: what pattern do these codes show, and why does it matter?
Check it is falsifiable — could a reader disagree with it?
Check it is not just your interview guide restated.
Attach two or three anchor extracts that support the claim.
Note one extract that complicates it (themes with no tension are usually descriptions).

Candidate theme — worksheet Codes grouped: ____________________________________ Category (shared topic): __________________________ Candidate theme (a claim, not a topic): "____________________________________________________" Falsifiable? (could a reader disagree?) Y / N Just the interview guide restated? Y / N Anchor extracts (2-3): 1. ______________________________________________ 2. ______________________________________________ Extract that complicates it: ______________________ Decision + date: __________________________________ (who decided, and why — keeps the trail)

Download the worksheet (Markdown)

The honest limit. No worksheet decides whether a theme is right — that judgement is yours. What the worksheet does is keep the reasoning and the supporting extracts together, so the theme is one you can explain and a second coder can check.

Related guidesRead next

Recommended tool · the next step

QualiVahti Local

Keep your codes and candidate themes in one plain-text place you can show a second coder — the QualiVahti Local codebook does that, and records who decided what. The files stay on your machine.

Explore QualiVahti Local · €49 →