Framework Analysis vs. Thematic Analysis: What's the Difference?

Framework Analysis and thematic analysis both start with codes and end with themes, but they get there through different mechanics. If your team needs a visible, case-by-case audit trail that a funder or clinical co-investigator can actually scan, Framework Analysis is usually the better fit. If you're working alone or in a small qualitative team chasing something more interpretive and slow-cooked, reflexive thematic analysis will probably serve you better.

That's the short version. Here's what actually separates the two in practice, and how to tell which one your project needs before you're three transcripts deep and regretting the choice.

What Is Framework Analysis?

Framework Analysis (sometimes called the Framework Method) is a structured, matrix-based way of managing and comparing qualitative data by case and by theme at the same time. Jane Ritchie and Liz Spencer developed it in the late 1980s at the UK's National Centre for Social Research, built for large-scale policy studies where conclusions needed to be traceable back to the original transcript.

The method survived because it does something ordinary coding doesn't: it forces you to hold the whole case in view, a participant's row, while still comparing across cases in a theme's column. Ritchie and Spencer's original model runs five stages: familiarization, identifying a thematic framework, indexing, charting, and mapping and interpretation. Gale and colleagues expanded this into a seven-stage version in 2013 for team-based health research, splitting out transcription as its own step and adding an explicit meeting where the team agrees on a working framework after coding the first few transcripts.

How Is Framework Analysis Different from Thematic Analysis?

The mechanical difference is the matrix. Ordinary thematic analysis groups data by theme, and honestly, it's easy to lose track of who said what once you've pulled fifteen quotes under "distrust of providers." Framework Analysis keeps every participant's full picture intact in their own row while still letting you scan a column and compare across everyone.

That last row is the real selling point for applied research. A framework matrix lets a co-investigator who has never coded a transcript in their life open a spreadsheet and see, cell by cell, why you concluded what you concluded.

When Should You Actually Use Framework Analysis?

Use it when your question is at least partly deductive, your team is bigger than one or two qualitative specialists, or your funder wants something they can audit. It shows up constantly in health services research, nursing, public health, and UK-style policy evaluation for exactly that reason: it gives non-qualitative collaborators, like an epidemiologist or clinician on the same grant, a shared structure they can actually follow. Should You Use Inductive or Deductive Coding? is worth a read if you're deciding your approach.

It's a bad fit when your data is genuinely heterogeneous, meaning participants aren't really answering the same underlying questions, because the matrix needs shared columns to function. It's also the wrong tool if your real interest is narrative structure, conversational turn-taking, or discourse and power. Those need narrative analysis, conversation analysis, or discourse analysis instead, not a grid.

How Do You Build the Matrix?

You build it by charting, the step most first-timers underestimate. Once your team is trained and the framework has stabilized, plan on roughly half a day per hour-long transcript for coding and charting combined. A single hour of interview typically transcribes out to fifteen to thirty pages, and all of that has to get read, coded, and compressed down into cells. Early transcripts take longer because you're still negotiating the framework itself.

The charting step is where Codes Are Not Themes: How to Actually Make the Leap becomes directly relevant: a cell isn't a copy-pasted quote, and it isn't a one-word label either. Good practice is a compressed summary that keeps the participant's own language, tagged back to a page and line number so anyone can trace it to source.

Tool choice varies more than you'd expect. Swallow and colleagues' widely-cited 2003 study charted its matrix in Excel. Parkinson and colleagues, studying young people's depression in 2016, used NVivo's built-in Framework Matrix tool, which keeps live links between each cell and the source transcript. A study of traditional birth attendants in Ebonyi State, Nigeria coded three focus group transcripts in a CAQDAS package, then charted the summaries into a Word table by hand.

None of these choices is wrong, the logic works whether it lives in a spreadsheet or dedicated software. Some teams now lean on AI-assisted coding tools like Paideias to draft a first-pass matrix skeleton, though someone still has to read every cell and judge whether the summary is honest.

What Goes Wrong the First Time Teams Try This?

The most common mistake is treating the matrix like a spreadsheet of numbers instead of what it is: condensed qualitative text. Teams new to Framework Analysis often slip into writing things like "13 of 20 participants said X," which misreads purposive sampling as if it were a representative survey. It isn't, and the number doesn't mean what it looks like it means.

The second common failure is charting focus group data by individual participant. Not everyone in a focus group answers every question, so you end up with a matrix riddled with empty cells. Chart focus groups by group, not by person, unless you also ran an individual survey to fill those gaps.

Frequently Asked Questions

Do you need special software for Framework Analysis?

No. Swallow et al.'s classic study used plain Excel, and the Nigerian traditional birth attendant study charted its matrix into a Word table after coding in a CAQDAS package. Dedicated tools like NVivo's Framework Matrix or AI-assisted platforms just make the charting faster; they don't replace the judgment of deciding what belongs in each cell.

Can you combine Framework Analysis with reflexive thematic analysis?

Yes, and in practice many teams do exactly that. Build your initial columns deductively from research objectives, chart the data, then apply a more interpretive, reflexive read across the finished matrix to develop typologies or explanations rather than stopping at description.

How long does Framework Analysis actually take?

Once your team has trained and the framework has stabilized, budget about half a day per hour-long transcript for coding and charting together. Early transcripts take longer, and multidisciplinary teams should build in time for the consensus meetings where coders compare labels after the first few transcripts.

Is Framework Analysis tied to a particular epistemology?

No. It's a data-management method, not a theoretical commitment, so it works whether you lean realist, interpretivist, or constructivist. That flexibility is part of why policy and health research teams reach for it: it doesn't force a paradigm fight before the coding even starts.