Qualitative Research: The Best Books on Interviews, Coding and Analysis, in Order
This is a craft path. It is about the actual work of qualitative research — sitting across from someone and asking a question well, being in a setting long enough to see something, turning a shelf of transcripts and fieldnotes into codes, and turning codes into a claim you can defend. The hard part is almost never getting people to talk; it is the middle passage from four hundred pages of transcript to a finding that is more than a set of quotations arranged under headings, and most of the books here exist because that step is where projects fail. A boundary worth stating, because the site carries a companion path on social science research methods: that one covers research design across the whole methodological range — sampling, causal inference, survey construction, experiments, mixed designs — and is where to go for the question of what design your study should have. This path assumes you have decided to do qualitative work and want to do it competently. It teaches interviewing, fieldwork, coding, analysis and the standards by which the result is judged. Many of these are long-running texts catalogued at earlier editions; where an approach has genuinely moved on, the note says so.
Choosing a Tradition and Designing the Study
IntermediateEstablish that qualitative research is not one method but several, each with its own logic, and choose deliberately rather than by default. By the end you should be able to distinguish ethnography, grounded theory, phenomenology, narrative and case study by what each treats as data and what each produces as a finding — and be able to state your study's design as an interactive whole rather than a linear plan, since qualitative designs change as the fieldwork teaches you things.
▸ Study plan for this stage
Pace: Eight to ten weeks for 1,032 pages. Maxwell's Qualitative Research Design is only 192 pages and is the best first book: design as an interactive model in which goals, framework, questions, methods and validity constrain each other, which is how qualitative projects actually develop. A fortnight, and
- Qualitative research as several methods rather than one, each with its own logic about what counts as data and what counts as a finding
- Maxwell's interactive design model: goals, conceptual framework, questions, methods and validity as mutually constraining rather than sequential
- Why a qualitative design changes as the fieldwork teaches you things, and how to write a plan that survives that
- The five traditions distinguished by what each produces: a story, a description of an essence, a theory, an account of a culture, an analysis of a bounded case
- The trustworthiness criteria - credibility, transferability, dependability, confirmability - and what each one asks you to have done
- Transferability as the reader's judgement rather than the researcher's claim, which is why thick description is a methodological obligation
- Choosing a tradition deliberately rather than by disciplinary default, and being able to say what the alternatives would have produced instead
- Name the five traditions and, for each, say what it treats as data and what it produces as a finding.
- Why does Maxwell insist design is interactive rather than linear? Give a case where a fieldwork discovery should change the question.
- What are the four trustworthiness criteria, and what concrete practice satisfies each?
- How does transferability differ from generalisability, and what does the difference require of your writing?
- For your own study, which tradition are you working in and what would the study have looked like in the tradition you rejected?
- Lincoln and Guba replaced reliability and validity rather than adapting them. What is the argument for that move, and what does it cost you when a quantitative reviewer reads your work?
- Draw Maxwell's interactive model for your own project on a single page, with the five components and the arrows between them. Then change one component - say, the access you can actually get - and redraw. Watching the other four move is the whole argument of the book.
- Take Creswell's worked example and write, in one paragraph each, the finding that each of the five traditions would produce from it. Five paragraphs, one page, and it is the most efficient way to choose a tradition rather than inherit one.
- For each of Lincoln and Guba's four criteria, write down the specific thing you will do in this study to satisfy it - the member check, the audit trail, the thick description, the negative case search - and when in the timeline it happens. Do this now rather than at write-up, because two of the four cannot be added retrospectively.
- Write a one-paragraph description of a setting or an interview you already have access to, then rewrite it as thick description in Lincoln and Guba's sense. The difference between the two paragraphs is what transferability actually costs.
Next up: With a tradition chosen and the standards known, the next stage takes up the method almost every qualitative project uses and almost nobody is taught properly.
The best first book: Maxwell treats design as an interactive model where goals, framework, questions, methods and validity constrain each other, which is how qualitative projects actually develop. Short, and it will save you from writing a plan you cannot follow.
The standard comparison of five traditions — narrative, phenomenology, grounded theory, ethnography, case study — running the same worked example through each so the differences are concrete. Read second, to choose.
Lincoln and Guba's 1985 classic, and the source of the trustworthiness criteria — credibility, transferability, dependability, confirmability — that qualitative work is still judged by in place of reliability and validity. Dated in places and foundational; read it for the standards, which you will be held to in stage five.
Interviewing
IntermediateLearn the single most-used qualitative method properly. By the end you should be able to build an interview guide that opens rather than closes, ask a real follow-up instead of moving to the next question, handle silence, and recognise when you are getting an account the person has already rehearsed. The stage moves from practical craft to the interview as a knowledge-producing conversation to the specifically ethnographic interview, where the goal is to learn the informant's categories rather t
▸ Study plan for this stage
Pace: Seven to nine weeks for 842 pages, and this stage should be read with a recorder running rather than only at a desk. Weiss's Learning from Strangers is 251 pages and is the most practical book on qualitative interviewing in English - how to recruit, how to open, how to probe, what to do when it goes
- Building a guide that opens rather than closes: how a question's form determines the length and kind of answer it can get
- The real follow-up - pursuing what was just said - versus moving to the next item on the guide, which is the difference between an interview and a questionnaire read aloud
- Handling silence, and why the interviewer's discomfort is the usual reason material is lost
- Recognising a rehearsed account: the story the person has told before, and how to get underneath it
- The interview as a knowledge-producing conversation with two participants, which is Kvale's frame and has consequences for how you report it
- Transcription as an interpretive act rather than a clerical one, and the decisions it silently makes
- Spradley's three question types - descriptive, structural, contrast - as a procedure for learning someone else's categories rather than confirming your own
- The ethics of the relationship: what you owe someone who has told you something they did not intend to
- Take three questions from your own guide and say, for each, what kind of answer its form permits. Which ones can be answered in a sentence?
- What does Weiss do when an interview goes wrong? Name three of his specific repairs and the situation each addresses.
- How do you tell a rehearsed account from a considered one, and what do you do next?
- Kvale treats the interview as jointly produced. What follows for how you write up a quotation?
- What decisions does a transcript make on your behalf - about pauses, overlaps, dialect, false starts - and which of them could change an analysis?
- Explain Spradley's contrast questions. What are they for, and why can you not ask them at the first interview?
- Record and transcribe one of your own interviews, then annotate it in Weiss's manner: mark every point where you moved on and could have followed up. Most people find several per page in their first interview, and this single exercise does more than the rest of the stage combined.
- Rewrite your interview guide after reading Weiss and before reading Kvale, then rewrite it again afterwards. Keep all three versions; the drift between them is a record of what you learned.
- Take one page of your own transcript and transcribe it a second time to a different convention - including pauses and false starts. Then code both versions lightly and see whether anything changes. Kvale's claim that transcription is interpretation is easiest to believe once you have made it fail.
- Build a Spradley-style question sequence for your setting: five descriptive, three structural, two contrast, in the order you would actually ask them. Then use it and note where the sequence broke down.
- Sit in silence for a full five seconds after an answer, deliberately, in three consecutive interviews, and write down what came after the silence. It is the cheapest technique in the stage and the hardest to make yourself do.
Next up: Interviews give you accounts of a setting; the next stage is about being in one, and about the writing that turns presence into data.
The most practical book on qualitative interviewing in English: how to recruit, how to open, how to probe, what to do when it goes wrong, with long extracts from real interviews annotated by what the interviewer was doing. Start here — it is craft, not theory.
Kvale's account of the research interview as a construction of knowledge between two people, covering the whole arc from thematising through transcription to analysis and reporting. The theoretical complement to Weiss, and the standard text. Catalogued under the display title Interviews rather than Kvale's intercapped InterViews.
The specifically ethnographic version, built around eliciting a culture's own categories through descriptive, structural and contrast questions. Read last in this stage; it is the interview form that leads directly into the fieldwork below.
Fieldwork and Fieldnotes
IntermediateLearn what to do when the data is a place rather than a conversation. By the end you should be able to describe the participation-observation continuum and where your project sits on it, know what to write down and when, and understand why fieldnotes are already analysis rather than raw material — the decision about what is worth recording is the first interpretive act in the study.
▸ Study plan for this stage
Pace: Four to five weeks for 434 pages, the shortest stage on the path and one you should be reading while actually in the field rather than before. Spradley's Participant Observation is 195 pages - how to select a setting, what degree of participation to adopt, and the descriptive question sequence that
- The participation-observation continuum, from complete observer to complete participant, and where your project sits on it
- Selecting a setting: access, boundedness, richness, and the tension between the setting you want and the one that will have you
- Spradley's descriptive question sequence as the move from unstructured presence to systematic observation
- Jottings in the field versus full notes the same evening, and why the gap between them is where most data is lost
- Fieldnotes as already analysis: the decision about what is worth recording is the first interpretive act of the study
- Writing concretely - what was said and done - rather than in summary characterisations that cannot be re-analysed later
- The researcher's own position and reactions as data to be recorded rather than noise to be suppressed
- Where on the participation continuum does your project sit, and what does that position let you see and prevent you from seeing?
- What does Spradley's descriptive question sequence actually do to your observation? Give an example from your own setting.
- Why are fieldnotes analysis rather than raw data? What is the practical consequence for how you write them?
- What is the difference between 'the meeting was tense' and a note you could re-analyse in six months? Rewrite the first as the second.
- How soon after leaving the field must full notes be written, and what specifically is lost by waiting?
- What do you record about your own reactions, and how do you keep that separate from what you observed?
- Spend one hour in your setting taking jottings, then write full notes that evening following Emerson, Fretz and Shaw. The next morning, mark every place in the notes where you wrote a characterisation instead of a description, and rewrite those passages. Repeat weekly; it is the core discipline of fieldwork.
- Run Spradley's descriptive question sequence through a single observation session and write down what you noticed that you would not have noticed otherwise. The sequence is procedural and the difference is usually obvious.
- Take one set of your own fieldnotes and highlight everything that is a quotation, everything that is a described action, and everything that is your inference. If the third category dominates, the notes will not support an analysis later.
- Write a page about your own position in the setting - who they take you to be, what that makes available and unavailable - and file it with the notes rather than in a separate diary. It belongs to the data.
Next up: You now have transcripts and fieldnotes and the problem this whole path exists for: the passage from several hundred pages of material to a finding you can defend.
Spradley's companion volume: how to select a setting, what degree of participation to adopt, and the descriptive question sequence that gets you from wandering around to systematic observation. The clearest procedural account of fieldwork available.
The book on the part nobody teaches: how to actually write the notes, from jottings in the field to full accounts the same evening, and how that writing shapes what you will later be able to say. Emerson, Fretz and Shaw is the standard and there is no real substitute.
Coding and Analysis
IntermediateThe core of the path and the stage most projects underestimate. By the end you should be able to run a first and second coding cycle and say why you chose those methods, build and defend a codebook, know what makes grounded theory different from thematic analysis rather than treating them as synonyms, and use displays to reason across cases instead of only within them. A recurring warning across these books, worth taking seriously: coding is not analysis. Sorting quotations into labelled bins pr
▸ Study plan for this stage
Pace: Four to five months for 1,833 pages, the core of the path and the stage every project underestimates. Saldana's The Coding Manual for Qualitative Researchers is 368 pages - dozens of coding methods, each defined, exemplified and matched to the kind of question it suits, organised into first and seco
- First-cycle and second-cycle coding as distinct operations, and choosing a method for each rather than defaulting to descriptive codes throughout
- The codebook: how a code is defined, what its boundaries are, and how you would demonstrate you applied it consistently
- Why themes do not emerge: the incoherence of the passive construction, and what Braun and Clarke say you are actually doing instead
- A theme as a pattern of shared meaning organised around a central concept, versus a topic summary masquerading as one
- What makes grounded theory different from thematic analysis: theoretical sampling, constant comparison, and building toward a theory rather than a description
- Charmaz and Corbin and Strauss's disagreement about prescriptiveness, and what each position implies about the researcher's role
- Memo-writing as the actual site of analysis, which is the answer to the coding-is-not-analysis warning
- Displays - matrices and networks - as a way of reasoning across cases rather than only within them
- What is the difference between a first-cycle and a second-cycle code? Name a method for each from Saldana and say what your question needs.
- Braun and Clarke object to the phrase 'themes emerged'. What is the objection, and what should the sentence say instead?
- Distinguish a theme from a topic summary using two examples from your own data. Which of yours are actually summaries?
- What does theoretical sampling require you to do that ordinary sampling does not, and at what point in the study does it start?
- Where exactly do Charmaz and Corbin and Strauss disagree, and which position does your own study need?
- What is a cross-case display for, and what claim does it let you make that within-case analysis does not?
- Code the same five pages of your own transcript twice, using two different first-cycle methods from Saldana. Compare the two code lists. The exercise is not to pick a winner but to see how much of what you find is determined by the method you chose.
- Write a codebook entry for three of your codes: definition, inclusion criteria, exclusion criteria, and an example of a borderline case decided each way. Then have someone else apply them to a page. Almost every code fails this the first time.
- Take your best-looking theme and test it against Braun and Clarke's own criteria. If it is a topic label with quotations under it, rewrite it as a claim about a pattern of meaning with a central organising concept - and note that this usually means fewer themes, not more.
- Write three analytic memos in Charmaz's sense - not summaries of codes, but arguments about what a code is doing - and keep them dated. The memos, not the codes, are where the finding will come from, which is the coding-is-not-analysis warning in practice.
- Build one Miles and Huberman display for your data: a matrix with cases as rows and a dimension as columns. Fill it in and note which cells you cannot fill. The empty cells usually tell you what to collect next.
- Run one round of constant comparison properly - a new incident against an existing code, deciding whether it fits, stretches or breaks it - and record what you did. It is a small procedure and it is the engine of grounded theory.
Next up: An analysis only exists once it is written, and the final stage treats the choice of voice as a methodological decision rather than a matter of style.
The indispensable reference: dozens of coding methods, each defined, exemplified and matched to the kind of question it suits, organised into first and second cycles. Start the stage here even if you already know which tradition you are working in.
Braun and Clarke's book-length treatment of reflexive thematic analysis and, importantly, of its widespread misuse — themes that are really topic summaries, and the incoherence of claiming themes emerged from the data. The most current book on the path and the one that will most change your practice.
The constructivist reworking of grounded theory: theoretical sampling, memo-writing and constant comparison, with the claim that the researcher constructs rather than discovers the theory. The most usable entry into grounded theory specifically.
Corbin and Strauss's procedural version of grounded theory, with open, axial and selective coding set out step by step. Read alongside Charmaz: the two disagree about how prescriptive the method should be, and seeing that disagreement is more instructive than adopting either on faith.
Miles and Huberman on analysis as data condensation, display and conclusion-drawing, and the origin of the matrix and network displays that make cross-case comparison possible. The most systematic approach here, and the best answer to the charge that qualitative analysis is unaccountable.
Writing It Up
IntermediateFinish the study. Qualitative research is made public as prose, and the choice of narrative voice — the neutral realist account, the confessional methods story, the impressionist tale — is a methodological choice rather than a stylistic one, because it determines what the reader can check. By the end you should be able to name the convention you are writing in and defend it, and to make your analytic procedure visible enough that a reader can judge the finding.
▸ Study plan for this stage
Pace: Two to three weeks for a single 173-page book, and the length is deliberate. Van Maanen's Tales of the Field treats ethnographic writing as a set of genres with different claims to authority - the neutral realist account, the confessional methods story, the impressionist tale - and it is the book th
- Realist, confessional and impressionist tales as distinct genres with distinct claims to authority
- The choice of narrative voice as a methodological decision, because it determines what a reader is able to check
- How the researcher's presence in the text relates to the trustworthiness criteria from the first stage - the audit trail made visible or invisible
- Making the analytic procedure legible: what you did to the data, in enough detail that a reader can judge the finding rather than only the prose
- The uses and limits of quotation: evidence for a claim versus decoration standing in for one
- Naming your convention and defending it, which is the concrete outcome of the whole path
- Name the three genres Van Maanen distinguishes and say what authority each claims and how.
- Which convention are you writing in, and what does it permit your reader to check? What does it hide?
- Go back to Lincoln and Guba's four trustworthiness criteria from the first stage. Which of them does your chosen voice make it easier to satisfy, and which harder?
- What would a reader need to see about your coding and analysis to judge your finding rather than merely believe it?
- Take three quotations in your draft. For each, is it evidence for a claim or a substitute for one?
- Write the same two pages of your findings three times - once as a realist tale, once as a confessional one, once impressionistically. Then choose, and write down the reason. That written reason is the methodological defence of your voice and it belongs in your methods section.
- Audit your draft for procedure: mark every place where you assert an analytic result and check whether the reader can see how you got there. Add what is missing. This is the practical form of the coding-is-not-analysis warning at write-up.
- Take every quotation in a finished section and delete it, leaving only your claims. Read what remains. If the argument disappears, the quotations were carrying it, which is Van Maanen's warning about authority in a form you can check in an afternoon.
- Reread your earliest fieldnotes after finishing Van Maanen. Most researchers find their own notes were already written in a genre they never chose, which is the strongest argument for choosing one now.
Next up: This closes the craft path. If the open question is now what design your next study should have - sampling, comparison, causal inference, mixed methods - that is the subject of the companion path on social science research methods.
The classic on ethnographic writing as a set of genres with different claims to authority, and the book that made the choice of voice a legitimate object of methodological argument. A short, sharp close to the path, and it will make you reread everything you have drafted.
Discussion
Keep reading
Paths that share books, cover the same subject, or open a related topic.
