The last row is the commonest failure in development practice. Images are gathered, captioned and published without ever being read as data — which makes them decoration with a methods section attached.

The last row is the strongest argument for the method. A photograph records the ordinary arrangement of a room, which no interview would produce because nobody considers it worth saying.

Most projects use images as output and call the work visual ethnography. The method sits in the middle row: the act of making the image is where the knowledge is generated.

The last row is the one practitioners actually do. The distinction is not the equipment or even the consent form — it is whether the person photographed had any say in how they appear.

The found category is under-used and cheap. What a household chooses to frame and hang tells you about aspiration, kinship and religion without you having taken a single photograph.

The fourth row is a research site hiding in plain sight. Whether a muster board is displayed, updated and legible is an observable fact about transparency that requires no interview.

Consent therefore has to cover circulation, not just capture. "May I take your photograph" is a smaller question than "where may this end up".

The order is deliberate. The ethics section comes last not because it matters least but because it is unreadable until you know what the methods actually involve.

The camera arrived with anthropology and was shaped by it. These were not aesthetic choices but methodological ones, expressing a theory about what people are and what could be known from looking.

Photographic classification helped make colonial categories real. Recording people as instances of a caste or tribal type contributed to the fixity those categories acquired in administration.

Salvage logic did not end in 1930. A grant proposal promising to document a "disappearing craft" or a "last generation of speakers" is using the same frame — and it still tends to photograph the practice rather than the pressure destroying it.

Bateson and Mead’s contribution was treating a photographic corpus as analysable rather than as evidence that the fieldwork happened. That is the founding move of the method.

Collier wrote from within a documentary tradition that took the camera's objectivity largely for granted. The critique that followed — Sontag, Berger, and the reflexive turn in anthropology — did not overturn the method so much as strip it of that innocence.

The turn is real and incomplete. Handing over the camera changes who frames; it does not by itself change who edits, who publishes, or who holds copyright — which is slide 78’s subject.

Neither argued against photography; both argued against treating it as transparent. The photograph is a made object with an author and an interest, and reading it as a window is the error.

Nobody intends this and the grammar is inherited. The conventions of the donor photograph were formed in the same century as the colonial archive, and they encode the same relation.

Answer all three before fieldwork, in writing. Projects that decide the medium first and the purpose later end up with a large image archive and no analytic use for it.

Mixing is normal and improves the work: participant photographs surface a theme, your survey documents its extent, and the archive — strongest on change over time, weakest in that you control neither what was kept nor destroyed — shows how long it has been there.

Repeat photography is the under-used technique here. The same frame at intervals — a season, a year, a decade — shows change that no single image and no recollection can establish.

The third row is the finding people overlook. Ask participants what they chose not to photograph and why — the refusals map risk, shame and privacy more precisely than any question.

An NGO’s own image archive is an excellent and uncomfortable research site. Twenty years of donor photographs record how the organisation has seen the people it serves, and the shift is usually visible.

Because elicitation is a technique rather than a source, its ethics travel with the image. An archival photograph shown in an interview carries whatever the archive did to the people in it, and the person you show it to may recognise a relative.

The talk is the data, and the photograph is the instrument that produces it. Studies that analyse the images and discard the transcript have used the technique backwards.

The third row is underrated. A shared object gives both people somewhere to look, which removes the pressure of eye contact and makes long pauses tolerable — and pauses are where the material is.

The mechanism most often cited is that photographs recruit a different kind of memory than questions do — episodic rather than summary. People narrate a scene rather than characterising a life.

The risk column is the reason to use both where possible. Your images make responses comparable and constrain what can be said; theirs open the field and cannot be aggregated.

The closing question is the most productive one in the sequence. "What would you have photographed that is not here?" reliably surfaces the thing the image set was blind to.

Prints work better than screens and it is not a nostalgic preference. A print can be held, sorted, set aside and returned to, which turns the interview into something the participant is doing rather than watching.

The third row is the specific hazard of cross-cultural work. An image you read as poverty may be read as ordinary, or as a particular season, or as somebody’s house — and the difference is the finding.

Wang and Burris set out the method in 1997 out of work with village women in Yunnan, drawing on Freirean critical consciousness, feminist theory and documentary photography.

Wang and Burris formalised it in 1997 from work with rural women in Yunnan. The lineage matters: photovoice was designed to reach policymakers, not to produce an archive.

The third goal is what distinguishes photovoice from a photography workshop, and it is the one that requires organising a room with people who can act in it.

The W is the hinge. A group that stalls at "why does this exist" with an answer about individual failings has not yet moved to structural analysis, and the D that follows will be a list of self-improvement tasks.

The W is where the method earns its lineage. Moving a group from "this is what it looks like" to "this is why it is like this" is the critical-consciousness step, and it is the one facilitators rush.

Ethics training for participants is not optional. They will photograph neighbours, children and strangers, and the consent obligations transfer to them along with the camera.

The third row is the one that gets people hurt. A participant photographing an illegal dump, a contractor or a police interaction is taking a risk you have asked them to take.

Photovoice produces depth and cannot produce representativeness, which is worth stating in any report. Twelve participants’ photographs are twelve people’s priorities, not the community’s.

Video is roughly an order of magnitude more work per hour of material than stills — logging, transcribing, syncing and editing all land after the fieldwork is over, when the budget is usually spent.

"Fly on the wall" is an aspiration rather than a description. A camera in a room is an event in that room, and observational film-making manages the effect rather than eliminating it.

Video is the most under-budgeted method in this deck. Teams plan the shooting days and not the logging, and the footage sits unanalysed because nobody costed the weeks it needs.

Sequence manufactures causation. Two shots in order are read as one leading to the other, whether or not they were related — which is the most powerful and least noticed editing effect.

Consent for video has to name the venues. Agreement to appear in a community screening is not agreement to appear on a website, and the second cannot be undone.

The third row is the reason to have shot video at all. If your analysis only uses the transcript, an audio recorder would have been cheaper and easier to consent.

Record the screening. What people say watching themselves and their neighbours is frequently the richest material the whole project produces, and it is routinely not captured.

Gatekeeper permission is not individual consent, and conflating them is the most common ethical shortcut in field photography. The sarpanch cannot consent on behalf of the woman in the frame.

Consent to be photographed is not consent to publication, and the two are routinely collected as one. The second is the one with consequences, and it needs its own conversation.

The last row is the donor-photograph convention. Removing the context removes the life, which is what makes the subject legible as an instance of need rather than as somebody in particular.

An image alone cannot distinguish these four, and it will be read as whichever the caption asserts. That is why metadata is a rigour requirement and not administration.

Record consent status against each image, not against the visit. Six months on, nobody can remember which photographs a particular person agreed to, and the safe default becomes publishing none of them.

The second row is the most valuable and the least recorded. What you were not permitted to photograph maps privacy, risk and shame more precisely than anything you did capture.

The fourth row is specific to development settings. A household that believes a photograph determines eligibility will present accordingly, which is a rational response and not deception.

The last row is the one nobody plans. Project photographs outlive the project, the staff and often the organisation, and consent given for one purpose does not travel to whoever inherits the drive.

Saturation is specific to a question, not a general property of the archive. You can be saturated on how a water point is used and nowhere near it on who never goes there.

Collecting is the easy part and the part that gets budgeted. Most visual projects end with an unanalysed archive, because analysis is slow, unglamorous and invisible in a workplan.

They answer different questions and are usually combined. Counting how many NGO images show women receiving something rather than doing something is content analysis with a semiotic point.

The third row is where Barthes is usually truncated. Connotation is not a property of the image; it is what a particular audience brings, so the same photograph signifies differently in the village and in a donor deck.

The last row is the archive-level question. Twenty years of an organisation’s photographs will contain no images of certain things, and the pattern of absence is a statement about how it sees.

The third is the one researchers underweight. You control the frame and can share control of the encounter, but you do not control how a stranger reads the image three years later or what caption an editor puts under it.

The second row is the reflexivity point people miss. What is remarkable to you is unremarkable locally, so a researcher’s archive is disproportionately composed of the locally ordinary.

A photograph is evidence of a moment, not of a pattern. One image of an empty clinic establishes that it was empty then, and nothing about how often it is.

The fourth row is why consent in these settings needs particular care. Refusal costs nothing only if the person believes it costs nothing, and a beneficiary asked by a service provider rarely does.

Collecting the two as one is the standard practice and the standard error. Someone comfortable being photographed at a training may not want that image in a fundraising campaign.

Photovoice projects that hand over the camera and keep the copyright have participated in the shooting only. Ownership is where the participation claim is tested and most often fails.

The donor gaze is the one practitioners produce without naming it. It is not malicious; it is what results when the audience is a funder and the image must demonstrate that help is required.

Assent matters as well as guardian consent. A guardian can permit a photograph; a child who does not want to be photographed should not be, and asking them is not a formality.

The harm arrives later, elsewhere, and you will not hear about it. That asymmetry is why anticipation has to be deliberate rather than left to whether anyone objected at the time.

Formal approval is a floor. An ethics committee cannot anticipate what a specific image will mean in a specific village three years after the project closed.

Publicly posted is not published for you. Legal availability and ethical availability are different questions, and terms of service settle neither.

It is worth being precise about why this is a research problem and not only a moral one: an image built to provoke pity is an image built to suppress analysis. It cannot carry a finding, because it has removed the context a finding would need.

Poverty porn works, which is why it persists. Guilt converts; it also produces donors who give once and a subject whose dignity was spent to achieve it.

Camera height is the change with the largest effect and the lowest cost. Photographing a seated person from standing produces diminishment automatically; crouching to eye level removes it.

"Nothing about us without us" entered wide use through the disability-rights movement and is now standard in participatory practice. Applied to images it has a concrete test: did the person see the photograph, and did they have a real option to say no to it?

The Dóchas Code and its international successors are widely signed and thinly audited. Signing is a commitment; the test is whether any image was ever rejected because of it.

The fourth row is the obligation most often promised and least often kept. Prints returned to the people photographed cost very little and are remembered long after the report is filed.

The second question is the sharper one. Most images that fail this test would not survive the author having to explain the caption to the person in it.

Plan the return to the community as an output with a budget line, not as a courtesy you will get to. Projects that leave it unbudgeted overwhelmingly do not do it.

Sequence is where the argument lives, and it is doing the same work as editing in film. Two images adjacent are read as related, which is a claim you are making whether or not you intend it.

Four captions, four different truths, one image. The caption is not description added to a photograph — it is what determines which of several possible readings the viewer arrives at.

Visual research extracts more visibly than most methods, because the object taken is recognisable. People notice that their photograph left and nothing came back.

Read Wang and Burris before running a photovoice project. Most implementations drop the policy-audience goal, and the original paper is short and explicit that it is not optional.