A theme is an interpretation, not a tally
Aliases: themes are not word counts · misused code frequencies · interpretive theme
What it is
Thirty-eight of forty sessions mention the menu, so Navigation becomes Theme One—that is counting codes, not doing thematic analysis. A theme is an account of how a set of extracts hangs together: what pattern they jointly show, under what conditions, and what that means for the research question. Frequency can flag where to look. It cannot convert occurrence into importance or truth. Thematic analysis answers “what is going on here,” not “how often did this token appear.”
Why it happens
How units are sliced determines the count. Cut one long help-seeking episode into ten Difficulty codes and Difficulty leads the table; leave it as one unit and the count drops. A repeated sentence may be emphasis or a sign of not being heard. Topics named in the guide systematically inflate counts; consequential exceptions never asked about sit near zero. Interpretation has to deal with those generative mechanisms: why these extracts can be told as one story, and why counterexamples do not yet overturn it. A tally hides the generative mechanism inside an integer. It reads like evidence and is often an echo of the slicing rule.
Studying it
For each candidate theme, assemble supporting extracts and extracts that do not fit, and test whether the account still holds with the counterexamples in view. Compare a code list ranked by count with a theme list ranked by explanatory force for the question; note high-count codes that never become themes and low-count exceptions that carry the interpretation. A theme write-up should state the pattern, the conditions it depends on, and what observation would rewrite it. An analysis that only delivers frequencies should be labeled a content inventory, not thematic analysis.
Where it stops holding
Content analysis, checklist audits, and estimates of “how many people mentioned X” have frequency as the target quantity; they need not be promoted to themes. Saturation judgments sometimes watch whether new codes still appear—that is a stopping rule, not a definition of a theme. Word clouds and bar charts of codes will lure readers into treating height as importance; the caption must state the unit. In tiny samples, one sharp failure can explain more than ten polite mentions; weighting by count systematically suppresses rare, high-cost patterns.
Applying it
- Write a theme as one or two sentences of pattern, then attach supporting and contrary extracts; only then note roughly how many people or instances.
- Flag topics the guide named, so “we asked, therefore it ranked first” cannot hide.
- High-count codes with no mechanism stay in an appendix as inventory; they do not become theme titles.
- In review, hide the counts and leave the account plus excerpts. If the claim collapses, it was a tally.