Interpretative Phenomenological Analysis (IPA): a practical guide

Phenomenological. The object of study is lived experience and what it means to the person living it. IPA does not ask whether an account is accurate; it asks what the experience was like and what sense the person made of it.

Hermeneutic. Interpretation is unavoidable and is done openly. The often-quoted formulation is the double hermeneutic: the participant is making sense of their experience, and the researcher is making sense of the participant making sense of it. The researcher's reading is a second-order activity and the write-up should show its working.

Idiographic. IPA is committed to the particular. Each case is analysed in full, in its own terms, before any cross-case work begins — and the write-up keeps individual participants visible rather than dissolving them into themes. This is the commitment most often abandoned in practice, and abandoning it turns IPA into thematic analysis with a phenomenological preface.

Together these produce a method suited to experiences that are existentially significant and where the interest is in meaning-making rather than in description: receiving a diagnosis, a change of identity, a bereavement, a decision someone is still living with.

IPA was developed by Jonathan Smith in health psychology in the mid-1990s, out of dissatisfaction with a discipline that had a great deal to say about populations and very little about individual experience. His early statement appears as Interpretative Phenomenological Analysis in Doing Social Psychology Research.

The theoretical grounding — how a phenomenological method sits alongside embodied and situated accounts of cognition — is worked out by Larkin, Eatough and Osborn in Interpretative phenomenological analysis and embodied, active, situated cognition.

The current book-length treatment is Smith, Flowers and Larkin's Essentials of Interpretative Phenomenological Analysis, whose opening chapter is the clearest short statement of what the method is and is not.

• Use it for experiences of existential significance. The method assumes the participant has been reflecting on this already. Applied to something mundane, it finds nothing to interpret.
• Use it when meaning-making is the question. If you want to know what people did, or how common something is, IPA is the wrong instrument.
• Use it with a small, purposive, homogeneous sample. Typically 3–10 participants who share the experience closely. A doctoral IPA study with six participants is normal; one with forty is a different method wearing IPA's name.
• Prefer reflexive thematic analysis for patterned meaning across a larger corpus. It is designed for that and IPA is not.
• Prefer descriptive phenomenology if you want the phenomenon, not the person. The Husserlian tradition brackets the researcher and seeks the essential structure; IPA does the opposite on both counts.
• Do not use it for evaluation or policy questions. The findings are not designed to generalise and should not be offered as though they were.

Six participants who had received an implantable cardioverter-defibrillator 12–24 months earlier were interviewed for 60–90 minutes each.

The first case produced an organising tension the researcher had not anticipated: the participant repeatedly described the device in the second person — "it decides for you, it does not ask you" — while describing his own actions in the first. His account was structured around having delegated something to a machine and being unable to withdraw the delegation. The linguistic noting is what surfaced this; a topic-coded reading would have filed the same passages under "device anxiety".

The second participant, analysed from scratch, produced something different: an account organised around gratitude and obligation, in which the device was a gift that imposed a duty to look after herself. Two later participants echoed the first case, one echoed the second, and the sixth described the device as unremarkable, which was itself informative because he was the only participant who had never had it discharge.

The group experiential themes reported convergence and divergence explicitly — "an agency ceded" was present in three of six accounts, and the write-up said so rather than presenting it as the experience of living with a defibrillator. The person who found it unremarkable was given his own section, because in an idiographic method a single divergent case is data rather than an outlier to be smoothed.

Nizza, Farr and Smith set out four markers of high-quality IPA in Achieving excellence in interpretative phenomenological analysis, and they are the most useful assessment tool available.

• Constructing a compelling, unfolding narrative. The analysis should read as a developing argument, not a list of themes with quotes attached.
• Developing a vigorous experiential and/or existential account. The findings should be about experience and meaning, not about topics the participants happened to mention.
• Close analytic reading of participants' words. Interpretation should be visibly built from specific language, including how something was said.
• Attending to convergence and divergence. Both what participants share and where they differ, with individual voices still identifiable.

• Sample too large and too varied. Thirty participants across four conditions cannot be analysed idiographically and will not be.
• Cross-case coding from the start. Destroys the idiographic commitment and produces thematic analysis.
• Description passed off as interpretation. "Participants found the experience difficult" is a summary. IPA asks what the difficulty consisted of and what it meant.
• No linguistic attention. Skipping the linguistic level removes most of what distinguishes IPA from careful thematic work.
• Themes as topics. "Communication with staff" is a topic. "Being managed rather than met" is an experiential claim.
• Participants disappearing. If the results section never says who said what, the idiographic commitment has been abandoned in the writing even if it survived the analysis.
• Claiming generalisability. Six people are six people. Theoretical transferability can be argued; empirical generalisation cannot.

IPA depends entirely on participants who can articulate their experience reflectively, which biases samples toward the verbally fluent and makes the method difficult with young children, people in acute distress, and anyone for whom the interview language is not their strongest.

It is also interpretation-heavy in a way that resists external checking. Two competent IPA researchers can produce different readings of the same transcript, and the method's answer — transparency and grounding in extracts rather than inter-rater agreement — is coherent but will not satisfy a reviewer who wants reliability.

And it is slow. Full idiographic analysis of six transcripts is a substantial piece of work, which is precisely why the sample must stay small and why studies that inflate the sample end up cutting the method.

Descriptive phenomenology in the Husserlian tradition brackets the researcher and seeks the invariant structure of the phenomenon — the opposite choice on interpretation.

Hermeneutic phenomenology shares the interpretative stance but is typically less procedurally specified and less idiographic.

Reflexive thematic analysis is the right alternative when the sample is larger or the question is about patterned meaning rather than individual sense-making.

Narrative analysis also attends closely to how an account is told, but treats the story's form as the object rather than the experience behind it.

IPA is the method where tooling helps least during analysis and most around it. The case-by-case work is close reading, and any tool that encourages coding all transcripts at once is working against the method.

What is genuinely useful is keeping the three levels of noting attached to the transcript line they belong to, keeping each case separate until cross-case work begins, and — at write-up — retrieving every extract behind a group experiential theme with its participant attached, which is what makes convergence and divergence reportable. Evidano supports IPA as a named methodology and follows the case-by-case structure rather than pooling transcripts. The interpretation is the researcher's work and cannot be delegated; a tool that offered to produce the themes would be removing the method.