Research & PolicyFree
Thematic Analysis Write-Up With Evidence
Every theme arrives with the number of participants behind it, the extracts that contradict it, and the reason it survived the review.
River reads the coded extracts, then counts each candidate theme two ways before writing anything: how many of your participants contributed to it, and what share of its extracts came from the single largest contributor. It searches the rest of the data for extracts that contradict the theme's organising idea. Themes that hold get written up with their verbatims, their participant count and their negative cases. Candidates that do not hold get named for what they actually are instead.
Search this and you get finished write-ups: dissertation excerpts, textbook specimens, six-phase walkthroughs. Each is a results section built from somebody else's data, and none of them can tell you whether your candidate themes are themes. The method's own fourth phase is checking the candidate themes against the coded data and the entire dataset, and it is the one phase that leaves no artifact behind. You do it in your head, and the write-up cannot show whether it happened.
Built for the doctoral candidate whose viva will ask how many people said this, the applied researcher writing up interviews for a commissioner, and anyone whose reviewer has asked for the disconfirming evidence. Reach for it once coding is finished and the codebook has stopped moving. Every count here depends on one speaker label per participant, the review synthesis writer covers the literature that framed the question, and the quantitative results writer handles the survey half. More research space packs sit alongside this one.
Nine of its twelve extracts came from one interview
Two of six candidate themes in an illustrative study of fourteen community pharmacists were not themes. One was called Benefits of the new service, a shape Braun and Clarke name directly: Experiences of Y or Benefits of X are classic examples of topic-summary type theme names. A topic summary collects what people said about a topic. A theme collects a shared meaning. The first one is an interview question with the answers stacked underneath it.
The other was called Pharmacist as gatekeeper and held twelve coded extracts, which looked respectable next to the smallest surviving theme. Three of fourteen participants contributed to it, and nine of those twelve extracts came from one locum. She was describing three different pharmacies, which is why she had so much to say. A finished write-up quotes two extracts per theme, so two quotes from that single interview would have carried it, and no reader could have seen that the other eleven interviews say nothing.
So the audit runs first and the prose runs second. The two rejected candidates held thirty-four of the study's one hundred and sixty-eight extracts, a fifth of the coded material, and dropping them left three themes and one subtheme rather than six. Reporting how each theme was evidenced is what the 32-item checklist for interviews and focus groups asks a submission to account for. The negative cases go inside each theme's own section, not into a limitations paragraph at the end.
How it works
Send the coded extracts
The coded data with participant identifiers intact, your question, and which thematic analysis you are doing.
Count before writing
Each candidate gets a participant count, a largest-contributor share, and a search for contradicting extracts.
Classify each candidate
A theme, a subtheme, a topic summary to dissolve, or one participant's argument reported as a divergent case.
Argue it in chat
Defend a rejected theme, split one carrying two ideas, or ask what the negative cases change.
What you get
- A Doc with each surviving theme, its definition, its organising idea and its verbatims
- The participant count and the largest contributor's share printed beside every theme heading
- Negative cases inside each theme's section, quoted, rather than gathered into a limitations paragraph
- A Sheet mapping every theme to its extracts, with the participant behind each one
- The candidates that failed the audit, named as topic summaries or single-participant arguments
- Theme names that state the pattern instead of labelling the interview question they answered
- A methods paragraph describing the theme development actually performed, ready for the manuscript
Common questions
Is counting participants not just turning qualitative work into a tally?
No, and the count is never the finding. It is the check that a pattern is shared before you claim it is. Braun and Clarke argue against agreement statistics for related reasons, and this is not one. Three of fourteen participants is not a weak theme. It is a divergent case, and reporting it as one is stronger than burying it.
My themes came out of a codebook, not from the codes. Does this still apply?
Yes, and the audit shifts slightly. A codebook or framework analysis sets themes in advance, so the question is not whether a theme emerged but whether your data filled it. An empty or thinly filled framework cell is a finding worth a paragraph. The participant count is how you tell thin from empty.
What if it finds no contradicting extracts for a theme?
It says so, and it says where it looked. Zero contradicting extracts across fourteen interviews is possible, and it is also the signature of a search nobody ran. The theme section records which codes and transcripts were searched, so a reader can tell a robust pattern from an unexamined one.
Can it code the transcripts for me, rather than starting from coded data?
It expects coding to be finished, because theme development from somebody else's codes is not your analysis. Send raw transcripts and it will say so, then offer to code them as a separate pass you review before any theme work starts. The write-up is only as good as the codes underneath it.
Will it rewrite my theme names?
It proposes new ones wherever the current name labels a topic rather than stating a pattern, and it shows both so you can refuse. Experiences of X and Benefits of Y are the two shapes it always flags. A name that could sit unchanged on your interview schedule is not a theme name.
How many themes should I end up with?
Fewer than you started with, usually. Six candidates becoming three themes and a subtheme is a normal outcome and not a loss. Each surviving theme has to carry a distinct organising idea, and two themes sharing one idea are a single theme whose name has not been found yet.
Does it work on focus groups rather than interviews?
Yes, with one change to the count. A focus group's unit is the group and the speaker, so a theme gets a group spread and a speaker spread, and dominance by one voice inside a group is recorded separately. Six groups of five is not thirty independent participants, and the sheet will not pretend otherwise.
Thematic Analysis Write-Up With Evidence
Fill in the form and your workspace opens with the work already underway.