These three get used almost interchangeably in write-ups, which is how projects end up describing one method and performing another. They answer different questions and produce different outputs, and choosing between them is a design decision, not a labelling one.
Thematic analysis
Best when you want to understand meaning and you are not committed to a framework in advance. Braun and Clarke's reflexive version is inductive by default: codes come from the data, themes are constructed from the codes, and the researcher's interpretation is explicitly part of the process.
It produces themes with names, definitions and illustrative extracts — a story about what the data means. It does not produce counts, and adding them undermines the logic of the method.
- Use it for: exploratory work, experience-focused questions, most doctoral qualitative chapters
- Avoid it when: you need to compare cases systematically, or a funder wants numbers
Framework analysis
Developed for applied policy research, and still the most under-used method in academic work. You build a matrix: cases down the rows, themes across the columns, summarised data in the cells. Everything stays visible at once.
That structure is what makes it powerful for comparison. You can read across a row to understand one participant in full, or down a column to compare everyone on one theme — which is exactly what you need when the question is whether outcomes differ by site, group or condition.
- Use it for: evaluations, multi-site studies, anything with a pre-set set of questions to answer
- Avoid it when: the point is to discover what matters rather than to compare on what you already know matters
Content analysis
The quantitative cousin. Codes are applied to bounded units and counted, and the counts are analysed statistically. It answers 'how often' and 'does this differ between groups', which the other two cannot.
It requires a coding frame fixed before you start, mutually exclusive categories, and reliability testing. In exchange you get numbers you can put in a table and test.
- Use it for: media and document analysis, open-ended survey responses at scale, mixed-methods quantitative strands
- Avoid it when: the interesting thing is what was meant rather than how often it was said
The test is simple. If the finding you want to report is a story, use thematic. If it is a comparison, use framework. If it is a number, use content analysis.
Mixing them
You can, and plenty of good studies do — a content-analytic count of an open-ended survey item alongside a thematic reading of the follow-up interviews, for instance. What you cannot do is start thematically, find the themes disappointing, and retrofit counts to make them look more substantial. That shows, and reviewers catch it.
