Ethnography and Participant Observation

Everything in Topic 7.4 answered questions of the form how much and what causes what. This Topic answers a different family: what is this, how does it work, what does it mean to the people doing it, and what actually happens as opposed to what the rules say happens.

The most demanding of these methods asks you to go somewhere, stay for a long time, take part, and write down what you see.

It is also the method with the highest ratio of misunderstanding to attention. People assume it is easy because it has no equations, and that it has no standards because its standards are not statistical. Both are wrong, and this lesson is about what its standards actually are.

"You've slowed me up plenty since you've been down here."

In 1937 a young researcher called William Foote Whyte moved into an Italian-American district of Boston that he would later call Cornerville, to study a neighbourhood that outsiders described almost entirely in terms of what it lacked — organisation, order, respectability.

He stayed nearly four years. He learned Italian. He lived with a local family. He bowled, badly and often, with the men whose street corner he was studying.

And the study exists because of one person. A local man Whyte called Doc — a natural leader with an unusual analytical mind — took him on. Doc introduced him, vouched for him, explained what he had just seen, and once told him, memorably, that if he asked questions people would clam up, but if he simply came along nobody would bother him.

What Whyte found overturned the standard picture. Cornerville was not disorganised. It was differently organised — densely, with elaborate structures of obligation, reputation and leadership that no outside agency could see because none of it was written down. His most famous demonstration was almost embarrassingly concrete: the men's bowling scores tracked their standing in the group. Not their skill — their standing. A low-ranked member who bowled well in practice would fall apart in a match against a leader, under a barrage of good-natured commentary that everyone understood and nobody had planned.

Then Doc said the thing that every ethnographer has thought about ever since.

"You've slowed me up plenty since you've been down here. Now, when I do something, I have to think what Bill Whyte would want to know about it and how I can explain it. Before I used to do things by instinct."

Read that carefully, because it contains the method's central problem and its central strength in one breath.

The problem: Whyte changed the thing he came to observe. Doc's behaviour was now partly performed for an audience of one.

The strength: Doc's reflection — the effort to explain to an outsider what had previously been instinct — is a large part of what made the analysis possible at all. The presence that distorts is the same presence that elicits.

And there is a coda that belongs in an honest course. Fifty years later another researcher, Marianne Boelen, returned to the district, interviewed surviving residents including Doc's family, and published an account contradicting Whyte on several points — the neighbourhood's character, Doc's standing, whether people recognised themselves in the book. Whyte replied at length and disputed her methods and her informants.

The dispute was never resolved, and that is the point. Two accounts of the same place, decades apart, neither verifiable against a shared record — because the fieldnotes are the data, and the data belonged to the fieldworker. Whatever you conclude about Whyte, this exchange defined the discipline's problem: ethnography's evidence is not, in the ordinary sense, inspectable. Everything modern ethnographers do about rigour is a response to that fact.

Three problems arrive together and they cannot be solved separately.

One — you change what you are there to see. Not occasionally: continuously, and most in the settings where your presence is most unusual.

Two — your access is partial and it was granted by someone. Whoever let you in has a position, and by entering through their door you have taken part of it. The faction that sponsored you is the faction whose enemies will not talk to you.

Three — the analysis rests on judgements only you were present to make, and the reader cannot check them.

Every technique in this lesson is an answer to one of those three. And the reason ethnography survives them is that it is the only method that reaches certain things at all: what people do rather than what they say they do; the practices they cannot articulate because they are not aware of them; processes that unfold over months; the unofficial system that runs alongside the official one; and the categories participants actually use, as opposed to the ones a questionnaire offered them (see 7.4.1).

What the method actually involves

Immersion, participation, observation, and an enormous amount of writing.

Prolonged immersion. Months at minimum, more often a year or several. Time is not a luxury here; it is the instrument. Short visits capture the setting's front stage (see 5.3.2). Only duration gets you past the performance, through the seasons of a setting's life, and to the point where people stop explaining themselves for your benefit.

Participation, in some degree. Doing what members do, to the extent possible and appropriate. Participation gives you the feel of a practice — the physical difficulty, the boredom, the timing — which observation alone does not.

Observation with a question in mind, which is what separates fieldwork from hanging around. The question changes constantly (see 7.2.1), and that is legitimate; but there is always one.

Learning the members' categories. Not what you would call this, but what they call it, and what distinctions matter to them. This is the core interpretive task (see 7.1.1 on second-order constructs), and it is why ethnographers pay obsessive attention to vocabulary.

And fieldnotes, which are the actual work. The rule that every experienced fieldworker gives: write them the same day, before sleeping. Jottings in the field — fragments, phrases, times — expanded into full notes within hours, because detail decays fast and what decays first is exactly the concrete specificity that makes notes usable a year later.

Good notes keep three things separate: what was observed (he stood up and left without speaking), what is inferred (he seemed angry), and the fieldworker's own reaction (I felt I had caused it). Collapsing these is the commonest failure in novice fieldnotes, and separating them is what lets you go back later and re-interpret without having to re-remember.

Thick description — Clifford Geertz's term, borrowed from the philosopher Gilbert Ryle.

Ryle's example: consider a rapid contraction of one eyelid. A thin description records the physical movement: the eyelid contracted.

But that movement might be an involuntary twitch; a conspiratorial wink; a parody of someone else's wink; a rehearsal of a parody; or a fake wink designed to make a third party think a conspiracy exists. These are entirely different social acts and physically identical.

Thick description records the act within the web of meaning that makes it that act rather than another. It is not description plus adjectives, and it is not writing at length. It is description at the level at which participants themselves distinguish one thing from another — which is precisely what a coding frame or an observation checklist cannot do, because it must decide in advance what the categories are.

Roles, access and their costs

Raymond Gold's four field roles, and what each buys and loses.

Complete participant — fully involved, identity as researcher concealed. Maximum access to what really happens, no reactivity from the research role. And covert, with everything that implies ethically.

Participant-as-observer — involved, and known to be researching. The standard position, and Whyte's. You take part, people know why you are there, and after enough time they mostly stop noticing.

Observer-as-participant — present and known, participating minimally. Common in organisational settings.

Complete observer — watching without interaction, usually in public space. Minimal reactivity, minimal understanding.

And access has a structure. Gatekeepers grant it and can withdraw it, which shapes what you dare write. Key informants — Whyte's Doc — teach you the setting, and you inherit their position in it. A researcher sponsored by management sees an organisation from management's doorway; one sponsored by the union sees a different organisation. The remedy is not to have no sponsor — that is not available — but to record whose door you came through and to work deliberately at the edges of their network.

One more thing about access that people learn the hard way: it is not granted once. It is renegotiated continuously, with every new person, at every new stage, and it can be lost in a week over something you did not notice.

Covert research, and the study that changed the rules.

In the 1960s Laud Humphreys studied anonymous sexual encounters between men in public toilets. He took the role of "watchqueen" — a lookout, a genuine role in the setting — and observed covertly.

Then he did something that is still discussed as a limiting case. He recorded participants' car registration numbers, obtained their home addresses through a police contact, waited a year, altered his appearance, and interviewed them at home as part of an unrelated health survey.

His findings were substantively important: the men were mostly not who the public imagined — a majority were married, and many were socially and politically conservative. The work challenged a stereotype with evidence that could not have been gathered by asking.

And the method was indefensible on several counts at once: no consent, deception compounded by a second deception, and the creation of a document linking named individuals to criminalised behaviour, in a period when exposure meant prosecution, dismissal and ruin. The risk of harm was not hypothetical.

The case is why modern ethics review exists in something like its present form (see 7.8.1). And it does not resolve into a simple rule. Covert observation is still occasionally approved — for studying powerful institutions that would otherwise refuse access, or settings where announcing yourself would destroy the phenomenon. What has changed is that the burden is now on the researcher to show why nothing else would work, and that the people observed cannot be harmed by it.

The three problems, and what discipline does about them

Reactivity, over-rapport, and generalisation.

Reactivity cannot be eliminated. It can be managed and it can be reported. Managed: by time, so that sustaining a performance for months becomes impossible; by participating rather than watching, which makes you unremarkable; by attending to what happens when you are treated as absent. Reported: by writing what your presence changed, rather than implying it changed nothing.

Over-rapport — Miller's term for what happens when a researcher becomes so embedded in a group that questioning it becomes a betrayal. The tell is that the fieldworker stops recording anything unflattering, and the account becomes advocacy in analytic prose (see 7.1.2 on Gouldner's reply to Becker).

And the opposite failure gets far less attention and is at least as common: never achieving rapport, and producing an account in which the participants are exotic, deficient or opaque. Both errors are the same error — a failure to hold analytic distance and human recognition at once.

Generalisation is the objection ethnography meets most often and understands best. One factory, one ward, one street corner. The answer is 7.3.1's: this is theoretical, not statistical, generalisation. The claim is not this is what neighbourhoods are like but here is a mechanism, documented in operation, with the conditions under which it operates. A mechanism established in one site is a claim about anywhere the conditions hold, and the way to test it is to go somewhere else — which is what Burawoy's extended case method formalises (see 7.2.3).

How modern ethnography answers the verification problem.

Because the fieldnotes are private, the discipline has developed a set of practices that substitute for inspectability. These are its rigour standards, and a serious ethnography exhibits most of them.

Extended and specified time in the field, stated precisely — how long, how often, where, doing what.

An explicit account of role and access. Who let you in, what you were taken for, what you could and could not go to, what you were excluded from. A study that does not tell you this is withholding the information most relevant to reading it.

Deliberate negative case analysis. Actively hunting the cases the emerging account cannot handle, and reporting them (see 7.2.3). The single strongest signal of a careful ethnography is the presence of material that complicates its own argument.

Recording where possible. Audio and video make the evidentiary base far more solid and can be re-analysed, at the cost of consent complexity and reactivity.

Team ethnography, where more than one fieldworker observes the same setting — expensive, and it converts an unverifiable judgement into a comparable one.

Member checking, with a clear-eyed view of its limits. Showing participants your account can catch factual errors and misunderstood terms. It cannot validate an analysis, because participants may disagree with an interpretation precisely when it is correct — an analysis of unacknowledged hierarchy will be rejected by the people at the top of it (see 7.7.2).

Archiving fieldnotes where consent permits, so a future researcher could check.

And the practice Mitchell Duneier has pressed for: seeking out the cases that are inconvenient for your argument — not the accessible, articulate, cooperative people who make good copy, but the ones whose lives do not fit. The vivid case and the representative case are rarely the same person, and the whole temptation of the method is to write from the vivid one.

Four temptations, and they are stronger than they look.

Selecting the vivid. Fieldnotes contain thousands of hours of dullness and a handful of extraordinary moments. The extraordinary moments write themselves, and they are by definition unrepresentative. A good ethnography reports the boring regularity that the striking incident departed from.

The quote that carries the argument. One participant said the thing that expresses the analysis perfectly. Ask how many said it, and how many said something else.

Composite characters and altered details. Sometimes necessary for anonymity, and every alteration weakens verifiability. The obligation is to state clearly what has been changed and on what principle — not to leave the reader assuming they are reading a record.

The ethnographer as protagonist. Accounts in which the fieldworker's journey — the difficulty of access, the personal transformation — displaces the setting. Reflexivity is required and is not the subject (see 7.1.2 on Bourdieu's objection to confessional reflexivity).

Because there are things no other method can reach, and they are not marginal.

What people do, as against what they say they do. The gap is large and systematic (see 7.4.1 on LaPiere).

Practices nobody can articulate. Ask a nurse how she decides which patient to see first and she will describe a protocol. Watch for a week and you will see a completely different, highly skilled, entirely unwritten system. She is not lying; the protocol is what is sayable. This is Bourdieu's practical logic (see 5.8.1) and it is only visible in operation.

Processes over time. How a rule becomes a routine. How a conflict escalates. How a newcomer is made into a member. A survey takes a photograph; a panel takes a sequence of photographs; ethnography watches.

The gap between the official and the real system. Every organisation has one, nobody writes it down, and it is where most of the actual work happens.

And for a reader, the criteria are usable without any technical training. How long were they there, in what role, admitted by whom? Do they report anything that complicates their own argument? Do they tell you what they could not see? Does the account let you imagine the setting well enough to picture what would have counted as a surprise?

An ethnography in which everything confirms the thesis and nobody is inconvenient is not a study. It is an essay with fieldwork in it.

The instrument is a person, and the person is in the room. Whyte's four years in Cornerville overturned the "disorganised slum" account by documenting an elaborate unwritten order — and Doc's remark that he now had to think how to explain what he had done by instinct states the method's problem and its strength together. The later Boelen restudy dispute defined the discipline's difficulty: the fieldnotes are the data, and they belong to the fieldworker.

Three problems arrive together: you change what you observe; your access is partial and sponsored; and your judgements are unverifiable.

The method is prolonged immersion, participation, observation with a question, learning the members' categories, and same-day fieldnotes that separate observation, inference and reaction. Thick description — Geertz via Ryle's wink — records an act at the level at which participants distinguish it from the physically identical acts it is not.

Gold's four roles trade access against ethics and understanding. Gatekeepers and key informants grant access and transmit their position with it — so record whose door you came through.

Humphreys' study is the limiting case on covert work: important findings, indefensible method, and the reason modern ethics review exists.

Reactivity is managed by time and reported honestly; over-rapport and its opposite are the same failure of analytic distance; and generalisation is theoretical rather than statistical — a mechanism plus its conditions, testable by going elsewhere.

Rigour substitutes for inspectability: specified time in the field, an explicit account of role and access, deliberate negative cases, recording, team fieldwork, member checking within its limits, archived notes, and the deliberate pursuit of inconvenient cases. The strongest signal of a careful ethnography is material that complicates its own argument.

Ethnography / participant observation — prolonged immersion in a setting, taking part while observing, to describe practice and meaning from within.

Fieldnotes — jottings expanded into full notes the same day, keeping observation, inference and reaction separate.

Thick description — describing an act within the web of meaning that makes it that act.

Gold's roles — complete participant, participant-as-observer, observer-as-participant, complete observer.

Gatekeeper / key informant — who grants access; who teaches you the setting and whose position you inherit.

Reactivity — the effect of the researcher's presence on the setting.

Over-rapport — identification with the group strong enough to suppress unflattering observation.

Negative case analysis — actively seeking and reporting cases the account cannot handle.

Member checking — returning the account to participants; catches errors, cannot validate an analysis.

Theoretical generalisation — inference to a mechanism and its conditions, not to a population.

Extended case method — using a deliberately awkward case to reconstruct existing theory.

Inconvenient cases — deliberately including participants whose lives complicate the argument.

One — write fieldnotes on somewhere you already are. Spend forty minutes in a familiar setting — a shop, a waiting room, a canteen — and write everything down afterwards. Keep observation, inference and reaction in separate columns. You will find how much of what you "saw" was inference.

Two — find the unwritten system. In a place you know well, describe the official procedure and then the real one. Note who taught you the second, and how.

Three — audit an ethnography. Take any ethnographic book or article and find: how long they were there, in what role, who admitted them, and one piece of evidence that complicates their argument. If the last is missing, note it.

Four — test the wink. Observe one ambiguous gesture in a real setting and write down three different acts it could be. Then note what would tell you which.

Five — name your sponsor. For any setting you might study, work out who would have to let you in — and write down which people that would make unavailable to you.