Framework analysis: A step-by-step guide for qualitative researchers

What is framework analysis?

Framework analysis is a systematic, matrix-based approach to qualitative data analysis that organizes cases against themes so researchers can compare findings across participants while keeping a clear, documented trail back to the original data.

The method was developed in the 1980s by Jane Ritchie and Liz Spencer from the Qualitative Research Unit at the UK's National Centre for Social Research (NatCen), originally to support applied social policy research. It later became especially prominent in multi-disciplinary health research, where teams needed a way to analyze large qualitative datasets—often collected by different researchers—while still producing findings that policymakers and stakeholders could trust.

What sets framework analysis apart from other qualitative methods is the framework matrix: a table where each case (a participant, site, or unit of analysis) occupies a row, each theme occupies a column, and each cell holds a summary of that case's data on that theme.

This structure is what makes the method's biggest strength possible: transparency. Every step is documented, every summary traces back to coded source data, and multiple researchers can review or replicate rigorous qualitative analysis without guessing how a conclusion was reached.

That transparency is exactly why applied researchers, especially in qualitative health research, reach for framework analysis. When findings need to hold up in front of funders, government bodies, or clinical review boards, an audit trail isn't optional—it's the difference between a defensible conclusion and an anecdote.

In summary, framework analysis:

• Is a qualitative method for analyzing data using a case-by-theme matrix
• Was developed by Ritchie and Spencer at NatCen for applied policy research
• Combines predefined (deductive) and emergent (inductive) themes
• Is designed to be transparent, comparable, and auditable

When should you use framework analysis?

Framework analysis works best when you have defined research questions, multiple cases to compare, and a need to show your analytical process clearly to others. It's less suited to research that's purely exploratory or aimed at generating new theory from the ground up.

Because the method starts with a structure—even if that structure evolves—it's a strong fit for applied and policy research, team-based projects where consistency across coders matters, and studies where stakeholders will expect to see how a conclusion was reached. It's a weaker fit when the goal is deep, single-case interpretive analysis, or when researchers want maximum flexibility to let entirely new theoretical categories emerge without a starting framework.

As with other qualitative analysis methods, framework analysis isn't without tradeoffs. Because the framework is built early and applied systematically, there's a real risk of locking themes in place before the data has fully revealed itself—flattening nuance in service of structure.

Researchers who use the method well treat the framework as a living document, refining it as familiarization and indexing surface new patterns, rather than treating the first draft as final.

Framework analysis vs. thematic analysis

Framework analysis is best understood as a structured, comparative form of thematic analysis. Both methods identify themes in qualitative data, but framework analysis adds a matrix output and a more explicit, auditable process.

Thematic analysis as a qualitative research practice gives researchers latitude to let themes emerge organically and to interpret data with fewer procedural constraints.

Framework analysis narrows that flexibility in exchange for consistency: every case is examined against the same themes, and the resulting matrix makes cross-case comparison and reporting more direct.

If your research calls for open-ended theme development without a matrix structure, our guide to thematic analysis covers that method in depth. Framework analysis is the better choice when you need the comparative structure a matrix provides.

The five steps of framework analysis

The framework method follows five sequential but iterative stages: familiarization, identifying a thematic framework, indexing, charting, and mapping and interpretation. In a framework analysis process, each stage builds on the last, though researchers often revisit earlier stages as understanding deepens.

1. Familiarization

Familiarization means immersing yourself in the raw data before attempting to analyze it. Read and re-read transcripts, field notes, and other source material involving textual data to understand the breadth of what participants said and the context behind it.

This stage isn't just passive reading. Researchers typically jot down preliminary ideas and initial impressions as they go—early notes that will directly inform the themes identified in the next stage. Skipping or rushing familiarization is one of the most common ways researchers undermine the rest of the process.

2. Identifying a thematic framework

This stage builds the index of themes that will eventually become the columns of your matrix. Key themes come from two directions: some are predefined based on the research questions (deductive), and others emerge directly from the data during familiarization (inductive).

The resulting framework doesn't need to be perfect on the first pass. It's a working structure that gets tested—and refined—once you start applying it to real data in the next stage.

3. Indexing

Indexing is the systematic application of your thematic framework across the entire dataset. Every relevant piece of data gets coded to the theme (or sub-theme) it corresponds to, ensuring consistent treatment across all cases.

This is where consistency matters most, particularly for teams. If different researchers are indexing different transcripts, clear, explicit code definitions are what keep the resulting analysis comparable.

4. Charting

Charting is the stage where indexed data gets summarized into the framework matrix itself: cases in rows, themes in columns, and each cell holding a condensed summary of that case's data for that theme. It's the step that gives framework analysis its name and its defining structure.

Charting is also, by a wide margin, the most labor-intensive step in the method. Done by hand, it means manually re-reading indexed excerpts for every case-theme combination and writing a concise, accurate summary for each cell—work that scales directly with the number of cases and themes in a study. This is precisely where software built for the task earns its keep, a point we'll return to below.

5. Mapping and interpretation

With the matrix populated, mapping and interpretation is where analyzing qualitative data pays off. Researchers read across rows to compare themes within a case, and down columns to compare a single theme across all cases, looking for patterns, associations, and explanations that answer the original research questions.

This is the step where the structured groundwork of the previous four stages pays off. Because every case has been summarized consistently against the same themes, patterns across the dataset are visible in a way that's difficult to achieve with unstructured qualitative notes.

A worked example: Framework analysis in practice

To make the method concrete, here's a fictional, illustrative example—not a real study—showing how framework analysis might move from indexed data to a filled matrix to an interpretation.

Imagine a research team studying patient experience across three outpatient clinics. Their research questions focus on communication, wait times, and follow-up care. After familiarization with interview transcripts, the team builds a thematic framework with those three themes as starting columns, then indexes each transcript accordingly.

The resulting matrix might look like this, with each cell holding a condensed summary of that case's coded data for that theme:

Reading down the "follow-up care" column reveals a pattern the team might have missed case by case: follow-up is inconsistent across all three clinics and, where it happens at all, it's often patient-initiated rather than proactive. That single observation—visible only once the matrix made cross-case comparison possible—becomes a concrete, evidence-backed recommendation for the clinics involved.

How to run framework analysis in NVivo (the Framework Matrix)

NVivo's Framework Matrix is a purpose-built feature for the charting stage of framework analysis, letting researchers build a case-by-theme matrix directly from their cases and coded data—available on both Windows and Mac, with AI-assisted summarization to speed up the most time-consuming step.

Setting one up starts with cases and themes you've likely already established earlier in your project: cases (participants, sites, or other units of analysis) become the matrix rows, and theme codes—the codes you developed during indexing—become the columns. From there, NVivo offers two ways to populate each cell:

• Manual summarization, where researchers write each cell summary themselves while viewing the underlying coded excerpts in a side-by-side preview, keeping the connection to source data visible throughout.
• AI-assisted summarization, where NVivo's AI Assistant drafts a summary directly within a matrix cell for a given case and theme, which researchers then review, edit, or reject before it's finalized.

And that's where NVivo's Framework Matrix saves you critical time. Charting by hand means writing a fresh summary for every case-theme intersection in a study—work that grows quickly as cases and themes multiply. NVivo's Framework Matrix doesn't remove that analytical judgment, but it does remove the mechanical overhead: AI-generated summaries are editable, the original coding stays visible for verification, and teams working under strict IRB or data-privacy requirements can disable AI features at the admin level while still using the matrix manually.

Once populated, matrices export cleanly for reporting and write-up, giving researchers a publication-ready, auditable summary of their cross-case analysis rather than a set of disconnected notes.

Framework analysis best practices and tips

Framework analysis rewards discipline, but a few common missteps can undermine even a well-designed study.

• Don't lock the framework too early.
  Treat your initial thematic framework as a draft; revisit and refine it as indexing surfaces patterns you didn't anticipate.
• Keep indexing consistent across a team.
  Define codes explicitly and check for drift between researchers, especially on larger, multi-coder projects.
• Watch for over-summarizing in cells.
  A chart cell should condense data, not strip away the nuance that made it meaningful in the first place.
• Never lose the link back to raw data.
  Every summary should be traceable to its source excerpt so findings can be verified or revisited.
• Pilot the framework before full-scale indexing.
  Testing it against a handful of cases first catches structural problems before they multiply across the dataset.
• Maintain a documented audit trail throughout.
  This is the feature that makes framework analysis defensible to funders, reviewers, and policymakers—don't treat it as an afterthought.