Last updated: 17 September 2026
Quick Answer: Thick description records what a person did and the meaning that made the act intelligible. Gilbert Ryle coined the term in 1968; Clifford Geertz carried it into fieldwork in 1973. Thin description says a man contracted his eyelid; thick description says he winked, and what the wink was for.
In 1968, Gilbert Ryle asked an audience at the University of Saskatchewan to picture two boys rapidly contracting the eyelids of their right eyes. One was an involuntary twitch, the other a conspiratorial wink. From a film there might be no telling which was which, Ryle wrote, and yet there remains what he called the immense but unphotographable difference between them.
Who Coined Thick Description, Ryle or Geertz?
Ryle coined it, in his 1968 lecture "The Thinking of Thoughts: What is 'Le Penseur' Doing?", reprinted in the second volume of his collected papers. Clifford Geertz borrowed it five years later and credited him by name, citing two Ryle essays rather than one.
Ryle was not writing about method. He was asking what counts as a complete report of an action, a question Stanford's account of his later work traces through Rodin's Thinker.
Geertz opened The Interpretation of Cultures in 1973 by defining ethnography as an elaborate venture in, to borrow a notion from Gilbert Ryle, thick description. He named the two essays he drew on, "Thinking and Reflecting" and "The Thinking of Thoughts", then retold the wink almost line for line.
Getting the order right is not pedantry: Ryle asked whether a description is adequate, not whether it is detailed, which is why thick description is a test you can fail with a thousand words and pass with forty.
What Is the Difference Between Thick and Thin Description?
Thin description reports the movement. Thick description reports the movement plus the public code that makes it a signal. Geertz gave the pair: thin is "rapidly contracting his right eyelids"; thick is "practicing a burlesque of a friend faking a wink to deceive an innocent into thinking a conspiracy is in motion".
Both describe the same half second of video. Only one tells you what happened. Thick does not mean long, either: Ryle's image is a sandwich of many layers, of which only the bottom slice is the thinnest description. The layers below run in his build order.
| Layer | On camera | What the writer must establish |
|---|---|---|
| Twitch | Eyelid contraction | Nothing was intended, and no recipient existed |
| Wink | Eyelid contraction | A shared code exists, and this recipient was the target |
| Parody of a wink | Eyelid contraction | Someone is being mimicked, and the audience is in on it |
| Rehearsal of the parody | Eyelid contraction | There is no audience yet, and this is practice |
What Does a Thick Description Look Like in Practice?
Geertz's worked example in the thick description chapter is not the Balinese cockfight, which is a separate and later chapter of the same book. It is a raw excerpt from his field journal about a Jewish trader named Cohen in the highlands of central Morocco in 1912.
Cohen is robbed and two visiting traders are killed. He claims his 'ar, a compensation payment, under a trade pact the French had abolished on paper and nobody had stopped using, and rides out with the Marmusha sheikh who held it. He returns with five hundred sheep, which the French seize, calling him a spy.
What makes it thick is Geertz's reading. Three frames of interpretation, Jewish, Berber and French, sat in one situation, and his diagnosis of the failure is a confusion of tongues: those sheep were at once a lawful settlement, a tribal concession and evidence of sedition. What we call our data, Geertz writes, are really our own constructions of other people's constructions.
Writing one means working outward from the thinnest true statement, naming the code that made the act meaningful and ruling out the rival reading, without losing the path back to the raw moment. Alchemic's end-to-end consumer research ties every flagged moment to the frame, clip and transcript line it came from.
How Does Thick Description Support Transferability?
It does not prove a finding generalizes. It gives a reader enough context to decide whether the finding travels to their own case, and that judgment belongs to the reader rather than the author. Transferability is one of four trustworthiness criteria Lincoln and Guba proposed in 1985, alongside credibility, dependability and confirmability.
A practical guide to qualitative trustworthiness puts it in one line: the researcher facilitates the transferability judgment by a potential user through thick description. Statistical generalization runs from a sample to a population and the researcher owns the claim; transfer runs from one case to another, and whoever owns the second case makes the call. So "n equals 30, so it isn't projectable" is the wrong objection: projectability was never the claim, and the broader qualitative research toolkit rests on a different warrant. Nor is the criterion itself settled: a 2025 conceptual analysis argues it collapses applicability, theoretical engagement and resonance into one question, and thick description serves only the first well.
What Kind of Fieldwork Produces Material Thick Enough to Interpret?
Thickness is decided in the field, not in the write-up. You cannot interpret what was never recorded, and no analysis recovers a code nobody captured. Two properties of the record decide, in the end, how thick a description it can support:
- The respondent's own words, in their own language. Meaning lives in idiom, register and irony, the first casualties of translation.
- The signal beyond the words. Hesitation, self-correction and tonality are code; a typed survey box records none of them.
Reach and interpretive depth are one problem: a study fielded only where a browser link and a shared written language reach has already narrowed whose meanings can be recovered. Voice notes as qualitative data and multilingual interviewing change what the record contains, not what analysis does with it. Alchemic publishes 57+ languages including Spanish, Arabic and Mandarin, and runs AI-moderated interviews inside WhatsApp with no link and no app, or as an outbound phone call.
Where Thick Description Goes Wrong
Thickness is not validity. A thick description can be thickly wrong, and Ryle said so first: he describes alternative possible thick descriptions of one act, so interpretation selects a single reading from several the evidence permits. Selecting confidently is not selecting correctly. Three failure modes account for most bad practice:
- Detail mistaken for depth. Pages of texture attached to a claim that stays thin, when the test is whether a sentence establishes a code.
- The single confident reading. An interpretation with no rival named reads as description but functions as assertion, and the interpretive bias that enters qualitative analysis is hardest to see in the readings that feel obvious.
- Deductive disclosure. Thickness accumulates particulars, and particulars identify people. The ICC/ESOMAR International Code requires researchers to ensure an inferred identity is not traceable by deductive disclosure, naming small samples and cross-analysis as the routes.
Some questions are better answered without it: counting is a job for a well-built thin description, and a setting you can live inside for months still rewards a trained ethnographer.