What Is Framework Analysis?
Framework analysis, also called the Framework Method, organizes qualitative data in a matrix. Each row is a case, usually one interviewee. Each column is a code or category. Each cell holds a short summary of what that person said about that topic, with a pointer to the transcript. A reader can scan down a column to compare people, or across a row to see one person's account whole.
Ritchie and Spencer developed it for large-scale policy research in the late 1980s. Gale and colleagues describe it as a method that is not tied to one epistemological position, so a team can use it with deductive, inductive or mixed coding. That flexibility is also a warning: the method organizes material, but the quality of the findings depends on the analytic choices made inside it.
Decision This Supports
Use it when several people must analyze the same material against questions fixed early, and a commissioner will want to see how each finding was reached. It suits an evaluation with a defined set of questions, such as why participants left a program, where a decision is due on a known date. It is less natural when the aim is to build a new theory from the data (see grounded theory) or to keep a single researcher's interpretive voice at the center of a reflexive analysis (see thematic analysis).
Before You Start
- A clear question and a decision it informs.
- Transcripts or detailed notes, with a note on how they were produced. Notes from rapid work carry less detail; see rapid qualitative analysis.
- A plan for who codes, who reviews, and how disagreements will be recorded. See qualitative codebook.
The Stages
Gale and colleagues describe seven stages. In practice they overlap and loop back.
| Stage | What Happens | What to Keep |
|---|---|---|
| Transcribe | Produce a usable record of each interview | Source of each transcript and any translation notes |
| Familiarize | Read and listen before coding | First impressions, written as memos |
| Code | Label meaningful segments | Excerpt ID and code for each segment |
| Build a working framework | Group codes into categories after several analysts compare their coding | A dated version of the framework |
| Apply the framework | Index all transcripts with it | Revised versions when the framework changes |
| Chart | Summarize each case under each category in the matrix | A link from each cell to its excerpts |
| Interpret | Look for patterns, contrasts and explanations | Memos that connect cells to claims |
A Small Worked Example
This example is fictional. It shows the reasoning, not a finding about any real program.
Three interviews with participants in a hypothetical adult learning program are charted under two categories.
| Case | Session Timing | Care Responsibilities |
|---|---|---|
| P1 (left in week 3) | Shift ends as the session starts; arrives late and feels behind (E1) | Not mentioned |
| P2 (stayed) | Also arrives late; facilitator catches him up (E2) | Not mentioned |
| P3 (left in week 2) | Not mentioned | Sister can watch children only until 5:30 (E3) |
Reading down the first column suggests timing matters. Reading across row P2 shows the same timing problem without leaving, so the working claim must change: timing conflict appears in two of these three accounts, but support at the session appears to matter for whether it ends in leaving. The empty cells matter too. "Not mentioned" does not mean "not true"; it may mean the topic was not asked about. A claim about all participants would need to check that.
Limits and Cautions
- A matrix summary is already an interpretation. Write cells close to the participant's words and keep the excerpt ID so a reviewer can return to the source.
- A tidy matrix can make counts look like findings. Cases in a purposive sample are not a population; a column count describes the people interviewed. See purposive sampling.
- Charting can flatten context if cells become one-word labels.
- Inter-coder agreement statistics fit some designs and not others. Describe the process your approach actually uses; see qualitative codebook.
- Keep cases that do not fit. See negative case analysis.
Frequently Asked Questions
How Is Framework Analysis Different from Thematic Analysis?
Framework analysis is a structured way to manage and compare coded data, usually with a team and often with a framework set early. Reflexive thematic analysis, as Braun and Clarke describe it, treats coding as the researcher's interpretive work and does not use a fixed codebook. Both can produce themes, but they rest on different assumptions about how knowledge is made. See choosing a qualitative analytic tradition.
Can AI Fill the Matrix?
A language model can draft cell summaries, but each summary must be checked against the excerpt it cites. See checking AI-assisted qualitative coding. People keep responsibility for interpretation.
M&E Studio also does this work for clients.
The team that writes and reviews this library takes on M&E system design, evaluation quality reviews, AI advisory and workflow automation.
Framework Method, matrix analysis, charting