Short answer: The critical incident technique (CIT) is a qualitative research method, introduced by John Flanagan in 1954, that collects participants' accounts of specific events that clearly helped or hindered an outcome. Researchers then extract each incident, sort the incidents into inductive categories, and report how often each category appears. It suits questions about what makes a practice succeed or fail.
Last updated: October 27, 2026
\n\nThe critical incident technique is one of the oldest structured interview methods in qualitative research, and it is still widely used in counselling psychology, nursing, education, service marketing and UX research. Its appeal is simple: people are poor at describing what they \"usually\" do, but they can describe one concrete event in detail. This guide walks through Flanagan's five steps, gives a ready-to-use CIT interview guide with probes, shows a short worked example, and lists the credibility checks reviewers now expect.
\n\nFlanagan described the method in The critical incident technique (Psychological Bulletin, 1954) as a set of procedures for collecting direct observations of human behavior that have critical significance for a defined aim. It grew out of the U.S. Army Air Forces Aviation Psychology Program in World War II, where it was used to identify the behaviors that separated effective from ineffective performance.
\n\nAn incident is any observable activity complete enough to let you draw inferences about the person doing it. It is critical when its purpose is clear to the observer and its consequences are definite enough that there is little doubt about its effect. In modern practice, most CIT studies replace expert observation with participants' own retrospective accounts, collected in interviews or open-ended questionnaires.
\n\nFifty years on, Butterfield, Borgen, Amundson and Maglio (2005, Qualitative Research) reviewed how the method had moved from task analysis toward exploring psychological constructs and experiences. They also warned that its flexibility had produced inconsistent terminology and studies that drift from Flanagan's analysis steps.
\n\nUse CIT when your research question is about what helps or hinders a specific outcome, and when the people involved have recent, first-hand experience of that outcome. It fits well when the topic is hard to discuss in the abstract but easy to illustrate with a story.
\n\n| Your question looks like… | Best-fit method | Why |
|---|---|---|
| What helps or hinders X? What makes X succeed or fail? | Critical incident technique | Concrete events map directly to helping and hindering factors. |
| How do people experience or make sense of X? | Semi-structured interview with thematic analysis or IPA | Meaning matters more than event-level causes. |
| Why do people prefer one option over another? | Laddering interview | Links attributes to consequences and values. |
| What do people do routinely, day to day? | Diary study or observation | CIT misses the ordinary; diaries capture it in the moment. |
| What triggered a purchase or a switch? | Jobs-to-be-done interview | Timeline of one decision, with push and pull forces. |
In service research, Gremler (2004, Journal of Service Research) reviewed how CIT had been applied to satisfying and dissatisfying service encounters. The same review is a good model for reporting standards, because it criticizes studies that omit how incidents were classified and checked.
\n\nWrite one sentence that states what the activity is for, in terms that participants and experts would agree on. Example: \"The aim of doctoral supervision is to help a student become an independent researcher.\" Every incident will be judged against this aim, so a vague aim produces vague incidents.
\n\nDecide who can report incidents (people with direct experience), what counts as an incident, which situations are in scope, and how recent the event must be. Many studies limit recall to the past 6 to 12 months to reduce memory distortion. Write these rules down before the first interview; they become your inclusion criteria for extraction.
\n\nInterviews give the richest incidents; open-ended questionnaires reach more people but often return thinner accounts. A December 2025 methods article in Frontiers in Psychology (Petiot et al.) compared both and recommends following written incidents with an interview that walks through each one in chronological order. Record and transcribe every interview.
\n\nAnalysis has three moves. First, choose a frame of reference based on how the results will be used. Second, extract incidents from transcripts and group similar ones into inductively formed categories. Third, decide the level of specificity: categories broad enough to be useful, narrow enough to be distinct. Butterfield et al. (2009) cite Borgen and Amundson's rule that a category should be supported by at least 25% of participants to stand on its own.
\n\nReport each category with a definition, example incidents, the number of incidents, and the participation rate. State the limits of the frame of reference and how credibility was checked.
\n\nThis guide follows the structure used in the enhanced critical incident technique: context first, then helping incidents, hindering incidents, and a wish list. Ask for one incident at a time and probe until you have the three elements Flanagan needs: the situation, the action, and the outcome.
\n\nNielsen Norman Group's guide to CIT in UX research (Rosala, 2020) adds two practical cautions: ask for a critical incident rather than \"the last time,\" and pilot the wording, because small changes in phrasing change which events people recall. For a structured approach to piloting, see our guide to pilot-testing an interview guide.
\n\nThe example below is invented to show the mechanics; it is not real data. Suppose a research lab studies what helps or hinders first-year PhD students in working with their supervisors, and interviews 15 students.
\n\n| Step | What the researcher produces |
|---|---|
| Raw incident | \"Two weeks before my first progress review, my supervisor blocked an hour, read my draft line by line and told me which section to cut. I went in with a clear argument and passed.\" |
| Extraction | Situation: upcoming review. Action: supervisor gave dedicated, specific feedback. Outcome: clearer argument, passed review. Type: helping. |
| Category | \"Timely, specific feedback on written work\" |
| Report line | Category name, definition, number of incidents, participation rate (e.g., 9 of 15 participants, 60%), and one or two representative quotes. |
The researcher logs, after each interview, how many new categories appeared. When several consecutive interviews add none, the category scheme is exhaustive. In the Butterfield et al. (2009) ECIT study, no new categories emerged from the sixth through the twentieth interview.
\n\nWe built this checklist from the nine credibility checks in Butterfield, Borgen, Maglio and Amundson (2009, Canadian Journal of Counselling). Use it in your methods section or ethics application.
\n\n| # | Credibility check | What to do | Evidence to report |
|---|---|---|---|
| 1 | Audio-recording interviews | Record and work from recordings or transcripts. | Recording and transcription procedure |
| 2 | Interview fidelity | An expert in CIT reviews a sample of interviews against the guide. | Who checked, how many, and feedback given |
| 3 | Independent extraction | A second person extracts incidents from about 25% of transcripts. | Agreement rate and how differences were resolved |
| 4 | Exhaustiveness | Log new categories per interview. | Interview number after which no new categories appeared |
| 5 | Participation rates | Count participants per category. | Rate per category, with the 25% threshold |
| 6 | Independent placement | A judge places 25% of incidents into your categories. | Match rate (80% or better is the usual guideline) |
| 7 | Participant cross-checking | A second contact where participants review their incidents and categories. | Number of participants reached and changes made |
| 8 | Expert opinions | Two or more domain experts review the final categories. | Who reviewed and what they questioned |
| 9 | Theoretical agreement | State your assumptions and compare categories with the literature. | Which categories are supported, novel, or contradicted |
For checks 3 and 6, our guide to intercoder reliability explains percent agreement, Cohen's kappa and when each is appropriate. For check 7, see member checking.
\n\nQualitati is a European, budget-friendly qualitative research platform for universities and research companies, with GDPR and data privacy as central priorities. It does not have a CIT-specific mode, but it covers the parts of a CIT study that take the most time:
\nKeep extraction decisions, category definitions and the credibility checks with the research team. AI output is a draft for human review, not a substitute for the independent coder in checks 3 and 6.
\n\nUse CIT if you are a PhD student, lab or research company studying what helps or hinders a defined practice: supervision, onboarding, service recovery, clinical handovers, software adoption. Do not use it when you need to understand lived experience in depth (choose IPA or narrative analysis), when participants lack first-hand experience, or when the behavior is too routine for anyone to remember specific instances.
\n\nThe critical incident technique turns vague questions about \"what works\" into a set of concrete, countable events. Its strength comes from discipline: a clear aim, specific incidents, careful extraction, and documented credibility checks. Done that way, it gives qualitative research a level of auditability that reviewers recognize.
\n\nIt is a way of interviewing (or surveying) people about specific, memorable events that clearly helped or hindered an outcome, then sorting those events into categories. Instead of asking what people usually think, you ask what actually happened in one concrete situation, what they did, and what resulted.
\n\nCIT studies are sized by incidents, not people. Flanagan's rule is to keep collecting until new incidents stop producing new categories (exhaustiveness). In the Butterfield et al. (2009) example, no new categories emerged after the fifth of twenty interviews. Plan for more, and log new categories per interview so you can show when you stopped.
\n\nIt is usually treated as a qualitative method today, but it was designed to produce counts. Incidents are categorized inductively, and categories are then reported with frequencies and participation rates (the share of participants who mentioned each category). Many studies report both.
\n\nECIT, described by Butterfield, Borgen, Maglio and Amundson (2009), keeps Flanagan's five steps but adds wish-list items (what would have helped), contextual questions, and nine credibility checks such as independent extraction, participant cross-checking and expert review of the categories.
\n\nYes, for data collection, if the interview guide asks for one specific incident at a time and the moderator probes for context, actions and outcome. Pilot the guide, read the transcripts, and keep incident extraction and categorization under human control with an independent second coder.
\n\nIf your question is about what helps or hinders an outcome, the critical incident technique is a proven, well-documented choice. Define the aim, ask for one incident at a time, log exhaustiveness, and report the credibility checks. When you are ready to collect incidents, you can start free with 30 monthly credits, no credit card required, or view transparent pricing. For the analysis stage, our thematic analysis guide and codebook template pair well with CIT categories.
\n\n\n","related":[{"slug":"longitudinal-qualitative-research-guide-2026","title":"Longitudinal Qualitative Research: Designs & Analysis","description":"Longitudinal qualitative research explained: designs, wave spacing, attrition, trajectory vs recurrent cross-sectional analysis, and a wave-planning matrix.","category":"Qualitative Methods","date":"2026-10-19","readTime":"13 min read","author":"Qualitati Research Team"},{"slug":"interpretative-phenomenological-analysis-guide-2026","title":"Interpretative Phenomenological Analysis (IPA): 7 Steps","description":"What is interpretative phenomenological analysis? IPA steps in 2022 terms (PETs, GETs), sample size, a worked example, and a quality checklist for theses.","category":"Qualitative Methods","date":"2026-09-29","readTime":"11 min read","author":"Qualitati Research Team"},{"slug":"qualitative-research-questions-examples-2026","title":"Qualitative Research Questions: Examples by Design","description":"Qualitative research questions examples for phenomenology, grounded theory, case study and narrative designs, plus how many to ask and a free fit checklist.","category":"Qualitative Methods","date":"2026-09-23","readTime":"10 min read","author":"Qualitati Research Team"}]}