Qualitative Data Analysis is a methods sourcebook, not a monograph. Its central claim is that qualitative analysis should be treated as a craft with visible procedures: data condensation, data display, and conclusion drawing/verification. Coding is not preparation before analysis; it is analysis. Matrices and networks are not decoration; they are the working surface on which comparison, pattern detection and conclusion testing happen. The book also insists that conclusions must be tested against evidence and against rival explanations.
The book is written for researchers in all fields who work with qualitative data, and for the teachers and students who need a disciplined vocabulary for analysis. It does not advance a new theory of knowledge. It orders an established tradition of systematic qualitative analysis into a usable sequence and requires conclusions to be tested against evidence and rival explanations.
This is the third edition. Johnny Saldaña prepared it with the late Matthew B. Miles and A. Michael Huberman. Saldaña states in the preface that he streamlined the original text, deleted material, added newer genres and reorganised the display chapters around five purposes: explore, describe, order, explain and predict.
The central claim is stated in the book’s phrase, “You know what you display.” A display is an organised, compressed assembly of information that allows conclusion drawing and action. Building a display is analysis. The book does not require every researcher to use the same matrices or networks. It argues that creating, testing and revising simple, practical analysis methods is a priority for qualitative researchers.
— Miles, Huberman & Saldaña, Ch. 1
Qualitative analysis can be made explicit, systematic and auditable. It is not a mysterious leap from field notes to conclusions. It is a craft with visible procedures: data condensation, data display, and conclusion drawing/verification.
Coding and memoing are not preparation before analysis. They are analysis. Matrices and networks are not decoration. They are the working surface on which comparison, pattern detection and conclusion testing happen.
The book’s task is to share that craft. It is a methods sourcebook for researchers in all fields who work with qualitative data, and for the teachers and students who need a disciplined vocabulary for analysis.
— Miles, Huberman & Saldaña, Ch. 1
Key point: The book’s argument is practical and methodological. It does not advance a new theory of knowledge. It orders a recognised tradition of systematic qualitative analysis into a usable sequence and insists that conclusions be tested against evidence and rival explanations.
A tight design starts with a clear conceptual framework, defined research questions and a bounded sample. It suits researchers working with well-delineated constructs, multiple cases or confirmatory aims. The analysis is more economical and comparable, but it risks bending local data into cross-case categories.
A loose design keeps pre-structuring to a minimum. Concepts emerge from the field. It suits experienced researchers exploring unfamiliar cultures or understudied phenomena. The risk is data overload and a long wait before meaningful patterns appear.
The distinction matters because it shapes sampling, instrumentation, coding and what counts as a defensible conclusion. The book recommends neither extreme. It asks researchers to choose deliberately and to justify the choice.
A conceptual framework comes first. Research questions are focused, sampling is bounded, and instruments are semi-structured. Analysis can move quickly to cross-case comparison.
The empirical sequence is more deductive. Theory and prior research guide the coding scheme. The researcher justifies each step by consistency between framework, questions, data and conclusions.
Example (Ch. 2): a multiple-case school improvement study with a pre-structured case outline and a common interview guide.
Sensitising concepts come first. The framework emerges during fieldwork. Research questions may be reformulated as understanding grows, and sampling is sequential.
The empirical sequence is more inductive. Coding schemes develop from the data, and early analysis feeds back into later data collection. The risk is overload; the reward is local groundedness.
Example (Ch. 2): an exploratory study of an unfamiliar setting where the key variables are not yet known.
The model (Ch. 1) treats qualitative analysis as four interrelated activities: data collection, data condensation, data display, and conclusion drawing/verification. The researcher moves among them throughout the study, not in a single sequence.
Data condensation is not reduction. It is selecting, focusing, simplifying and transforming data so that conclusions can be drawn and verified. It happens before, during and after data collection.
Data display is an organised, compressed assembly of information that allows conclusion drawing and action. Extended text is a poor display; matrices and networks are better.
The model explains why analysis is iterative. A new display suggests a new coding decision; a new conclusion sends the researcher back to the data. The loop is not a failure of procedure; it is the procedure.
Associated with the book’s pragmatist orientation and with its insistence that analysis be documented.
The researcher must keep a record of decisions: which data were condensed, which displays were built, which conclusions were tested. The model is only as good as the audit trail that supports it.
Associated with the book’s verification standards in Ch. 11.
A display is an organised, compressed assembly of information that allows conclusion drawing and action. The book’s central practical claim is that good displays are a major route to robust qualitative analysis (Ch. 5).
Displays are organised by purpose: to explore, describe, order, explain and predict. The third edition reorganises the display chapters around these five purposes rather than around within-case and cross-case divisions.
The display is not a report of analysis. Building it is analysis. Deciding on rows, columns, nodes and links forces the researcher to specify what is being compared and why.
Rows and columns intersect two lists. Matrices suit comparison across cases, roles, time periods or variables. They condense large amounts of data into a single page.
The book offers many forms: checklist matrices, role-ordered matrices, time-ordered matrices, conceptually clustered matrices, effects matrices and meta-matrices. The names sound technical; the book insists they are convenient labels, not standards.
Risk: a matrix can look more orderly than the data. The book warns that cell entries must be traceable to field notes.
Nodes and links show streams of actions, events and processes. Networks suit case-oriented accounts that preserve chronology and complex interrelationships.
The book uses networks for causal chains, causal networks, event-state networks, cognitive maps and composite sequence analysis. Networks are particularly useful for explanation and prediction.
Risk: a network can suggest more causal order than the data support. The book advises one-way arrows and an accompanying narrative.
The book is a methods sourcebook, not a report of new research. Its evidence is pedagogical and illustrative: worked displays, coding schemes, checklists, short cases and references. Many examples come from the Miles–Huberman school improvement study; others come from Saldaña’s projects.
That fits the genre. The guidance is clear enough to use, and the displays are shown in enough detail that a reader can see how the analysis was built. But the examples illustrate the advice; they do not test it.
Claims about method — such as the claim that display-driven analysis produces better conclusions — rest on professional judgement and illustration. The book does not claim empirical proof, and readers should not supply it.
The coding advice, display templates and verification tactics are clearly presented, internally coherent, and illustrated with examples a researcher can follow. The verification chapter gives the book its clearest quality standards.
The book is honest about its scope. It says it does not cover entry, field notes or interviewing, and it recommends other texts for those matters.
The book asserts the value of its display methods. It does not demonstrate their superiority over other forms of qualitative analysis. The software advice is tied to 2014 tools. Visual and arts-based analysis are acknowledged but not developed.
Novices may find the display repertoire overwhelming, even though the book warns them not to copy the displays mechanically.
The book succeeds as a practical guide to systematic qualitative analysis. It is less successful as a complete methods course. Researchers and students should use it with data in hand, and supplement it with fieldwork texts and current software guidance.
The selection has costs. The book is not a fieldwork primer, a statistics text or a current software manual. Its examples are largely educational and multi-case. It asks for trust where it could have demonstrated the effects of its own methods.
These are limits of scope, not failures of argument. The authors present the displays as examples and heuristics, not as standards. For the audience they have chosen, that is enough.
The strongest chapter is Ch. 11 on drawing and verifying conclusions. It gives the book its clearest account of what makes a qualitative conclusion trustworthy. The weakest material is the dated software discussion and the thin treatment of visual and arts-based data.
Graduate students and early-career researchers, who need a disciplined vocabulary for coding, displaying and verifying analysis.
Methods teachers, who can use the display models and exercises in courses.
Applied researchers and team leaders, who need templates for making analysis auditable and comparable.
Curious general readers can learn what qualitative analysis involves, though the book assumes a research project. This benefit is a reading, not a claim the authors make.
Readers seeking a fieldwork primer: entry, field notes and interviewing are outside the book’s scope.
Readers seeking current software advice: the CAQDAS discussion is tied to 2014 tools.
Readers seeking a light introduction: the display repertoire is demanding and assumes a project.
Bottom line: Read it for discipline. It gives researchers an audit trail from data to conclusions and a repertoire of displays for making analysis visible. What it does not give is a complete methods course. That is a fair trade for a sourcebook of this scope.