Ask a new colleague what the rules are around here and you get a handbook. What you don't get: the first person to speak at the weekly meeting is always one of the same two people, and nobody ever decided that, and nobody has noticed it.
This page is about the least scientific-looking method in social science — move in for a year or two, work alongside people, eat with them, write down what you see. Why it counts as evidence, what it can get that a survey never will, and the weakness it cannot design away: the person writing is also standing in the scene.
In 1922 a book called Argonauts of the Western Pacific opened with a rule that sounds obvious now and turned a page then: pitch your tent in the middle of the village, not at the mission station or the trading post; learn the local language rather than working through an interpreter; stay a year or two. Its author, Bronisław Malinowski, had done two such stretches of fieldwork in the Trobriand Islands (now part of Papua New Guinea) between 1915 and 1918.
The practice came to be called participant observation. What he was after he gave a name of its own: the imponderabilia of everyday life — how a day starts, how food gets divided, what people say when they meet on a path, what restores the mood after an argument. A questionnaire has no box for any of it. You can only be there.
But being there is not itself a method. It asks you to do two things that undercut each other: get close enough that people stop adjusting for you, and stay distant enough that you still notice which things are worth writing down.
In the late 1930s William Foote Whyte moved into an Italian-American neighbourhood in Boston and stayed more than three years. He spent his time with a group of young men who hung around a particular street corner, and went bowling with them. Eventually he saw something: each man's score on a given night matched his position in the group. One man who bowled well elsewhere, but ranked low, fell apart whenever he was up against men who ranked above him — not throwing the game, genuinely unable to bowl.
No questionnaire could have asked about this, because nobody knew it existed, the bowlers included. That is the method's hardest claim: it finds the question you would never have thought to ask.
Nail down the strength of the evidence: this is a single, long, deep case. It is good at "so that's how it runs" and silent on "in what fraction of people does this hold" — the exact complement of what the statistical rates in the piece on Durkheim and social facts can do, and neither substitutes for the other. Nor should it be romanticised: staying long does not automatically mean seeing straight. Two ethnographers have written irreconcilable accounts of the same society; the argument over Samoa ran for decades without a verdict.
An eyelid closes fast, once. On camera that is a few frames. But it could be an involuntary twitch, or a wink letting you in on something, or someone imitating a wink to get a laugh, or a person practising in front of a mirror at home. The physical movement is identical and these are four different acts.
The example is the philosopher Gilbert Ryle's. In 1973 Clifford Geertz borrowed it to set a standard for ethnography: recording the movement of the eyelid is thin description; saying which act it was requires knowing what a wink means among these people, when it can be used and with whom — that is thick description.
In April 1958 Geertz and his wife arrived in a Balinese village. For days nobody acknowledged them; people walked past as though past empty air. On the tenth day they went to a cockfight — the activity was banned by the authorities at the time, and this one was raising money for a school. The police came, the crowd scattered, and the couple ran with a household into its courtyard; the man of the house lied to the police, saying the two had been drinking tea with him all afternoon. The next day the whole village talked to them. The barrier came down not through explanation but through one shared escape.
His analysis afterwards ran like this: in the matches with the largest stakes, what is being wagered is not money. People never bet against a cock from their own side; matches are typically between parties standing in some kin or village opposition; and the deeper the match, the closer the odds are pushed to even, which makes it a bad deal as arithmetic. The money here is a measuring stick, and what it measures is standing — the same sum of money is a bet under thin description and a public act of allegiance under thick.
The piece on Weber and rationalisation described interpretive sociology (Verstehen): to explain why someone did something you first have to know what they take themselves to be doing. Thick description is that sentence turned into a craft you can practise.
The objection is hard and belongs on the table: thick description has no obvious route to falsification. The reader holds only the anthropologist's text and cannot check whether the reading is over-reading. In the 1986 collection Writing Culture the cockfight essay was taken on directly — the charge being that the "Balinese point of view" it constructs is not one Balinese would necessarily recognise; the argument has never closed. The honest formulation: thick description yields a candidate interpretation that other interpretations can challenge, not a measurement.
In 1967, twenty-five years after Malinowski's death, his private diary was published. The man in it is unrecognisable next to the composed observer of Argonauts: irritable, homesick, pages on his own body and moods, and at times harsh and racially prejudiced language about the people around him.
The discipline was unsettled for a long time. What mattered was not his character but that the diary punctured an unstated warrant: ethnographic credibility had always rested on "I was there, I saw it", while that "I" never appeared in the text. The diary made the "I" visible, and with it a question that could no longer be avoided: who brought this material back, and from what position?
This became known as reflexivity: the observer's identity, language competence, route in and source of funding are themselves part of the data, to be written in rather than scrubbed out. Ethnographies today often open with a section on who the author is and how they got into the field — that section grew out of this shock.
The ethical lessons come from elsewhere. In 1958 a study of a small town in New York State appeared with the town and everyone in it under pseudonyms. Residents read it and recognised themselves and their neighbours at once, and felt they had been turned into specimens. The following year the town's Fourth of July parade carried a string of floats mocking the book, the last one bearing an effigy of the author. The pseudonyms protected nobody — in a place of a few hundred people, a description is an identity document.
A heavier case came in 1965: a military-funded social science project on unrest in Latin America was exposed, having recruited researchers without disclosing its funding or actual purpose, and was cancelled that year. The episode pushed the American Anthropological Association to adopt principles of professional responsibility in 1971 — obligations to those studied come first, and the aims and sponsorship of research must be disclosed.
A longer history has to be stated plainly too: this discipline grew up entangled with colonial administration. Much early fieldwork happened in colonies, some funding and freedom of movement came from colonial authorities, and some findings were put to administrative use. There is a large literature reckoning with that period, and it comes from inside the discipline rather than from its critics outside.
So the real difficulty of the craft was never "how do I stay objective". The wholly uninvolved observer does not exist — the moment he walks into the village, the village has one more person in it. What can be done is to state your own position clearly: readers cannot audit the author's interpretation, but they can at least audit the author's situation.