Most development evidence comes from what people tell us — surveys, interviews, self-reports. Observation is the disciplined practice of learning by watching what people actually do, in the real settings where they do it. It is a method with its own rigour, not a coffee break.

People do not always do what they say — not from dishonesty, but because memory is faulty, norms shape answers, and habits run below awareness. The say–do gap is the distance between reported and observed behaviour.

Time in the field builds tacit understanding — the feel of a place, its rhythms, its tensions. Much of what matters in development is relational and spatial, and only becomes visible to someone who lingers.

Watching is interpretation. Two observers can see the same scene and record different things. The rest of this course is about making observation systematic, transparent and ethical — so it earns the word 'evidence'.

Every observation sits on three axes. Decide each one deliberately before you enter the field — they shape what you can see and what you owe the people you watch.

The observer takes part in the life of the setting — joining a self-help group meeting, working a shift, sharing a meal — while observing. The anthropologist's classic stance.

The observer stays apart — watching a school classroom from the back, timing a clinic queue, mapping a market — without joining in. More detached, more easily structured.

Stand in a busy market, an anganwadi, a gram sabha — a thousand things happen at once. Without a framework you record randomly, miss patterns, and tire fast. You need a lens that tells you what to look at.

Spradley advised starting with a grand-tour: a broad sweep that takes in the whole setting before zooming in. Walk the space, map it, get the lay of the land — then narrow to a mini-tour of one corner in detail.

A simple sketch map — doorways, seating, the queue, where the worker stands — captures spatial patterns words cannot. Who sits near the front? Who is pushed to the margins? Space encodes power.

Observation funnels over time: from descriptive (take in everything) to focused (track the dimensions that matter) to selective (count specific, well-defined behaviours).

A framework focuses attention — but a fixed framework can also blind you to what it leaves out. Hold your lens firmly enough to be systematic, loosely enough to be surprised.

Observation produces nothing without a record. Memory decays fast and reshapes itself to fit our expectations. Field notes are the raw data of observation — the single most important craft in this course.

Thin: 'a man nodded.' Thick: 'when the upper-caste landlord entered, the labourer stopped speaking mid-sentence and nodded — a gesture of deference everyone present understood.' Thick description carries the meaning.

When someone says something striking, write it down word for word, in their language, in quotation marks. A real quote is worth a paragraph of paraphrase — and it is honest evidence.

In observation, there is no machine between you and the data — you are the instrument, and your notebook is its output. Treat it with the seriousness a surveyor gives a calibrated gauge.

Once you know what matters, structured tools let you record it consistently — across observers, sites and time — so you can count, compare and track. Structure trades richness for reliability.

A checklist records whether defined things are present or absent: Is the growth chart displayed? Is the toilet functional? Is the teacher present? Simple, fast, and brutally clear.

A rating scale records degree, not just presence: classroom engagement on a 1–5 scale, cleanliness from poor to excellent. Powerful, but only if every point is anchored with a description.

Between a free notebook and a rigid checklist sits the observation guide: a structured prompt list of things to attend to, with room for open description. It steers without straitjacketing.

A structured tool is only as good as its definitions. Before you can tally 'on-task' or 'aggressive interaction', you must write down exactly what counts — an observable rule any observer can apply the same way.

Have two observers code the same session independently, then compare. Wide disagreement means your definitions are loose, not that one observer is 'wrong'. Tighten the tool and re-test.

After fieldwork you hold pages of notes, tallies and quotes. Sensemaking is the disciplined work of turning that pile into patterns, and patterns into insight. It is analysis, and it deserves as much rigour as the observing did.

Coding means tagging stretches of notes with short labels — 'waiting', 'gatekeeping', 'workaround', 'deference' — so you can gather everything about each idea across all your notes.

A theme is a pattern that recurs across observations and carries meaning for your question. Themes are built, not found: you assemble them from repeated codes, then test whether they really hold.

The instinct is to smooth over the observation that breaks your pattern. Resist it. The anomaly — the negative case — is where understanding deepens: it shows the limits of your pattern, or reveals a better one.

Qualitative does not mean anti-number. Noting that a behaviour appeared in 9 of 12 observed sessions is honest and useful — far better than the vague word 'often', which lets readers imagine any figure they like.

Rigour in qualitative analysis is about transparency. Keep an audit trail: your codes, the rule for each, example quotes, and how a theme was built. Another analyst should be able to follow your path from notes to claim.

Observation tells you what people do; interviews tell you what it means to them; documents and data tell you the scale. Together they triangulate the truth.

Observe first, then ask. Seeing a practice lets you ask far sharper interview questions — about the very thing you watched — and lets you probe the gap between what you saw and what people say.

Registers, muster rolls, attendance sheets and meeting minutes are claims about reality. Observation tests them: does the attendance register match the children you counted in the room?

Survey and administrative data tell you how widespread; observation tells you how and why. A dashboard shows immunisation coverage dipped; observation of the session shows the cold-chain box was empty.

When observation, interviews and data all point the same way, your confidence rises sharply. Independent methods reaching the same conclusion are unlikely to be wrong in the same direction by chance.

Disagreement is not failure — it is information. If the register says 30 and you counted 18, if women report ease of access but you observe them turned away, the divergence itself is the finding.

Triangulation is not only across methods. Two observers watching the same setting — ideally differing in gender, language or background — will notice different things. Comparing their notes is itself a check on bias and a way to see more.

A pattern is not yet an insight, and an insight is not yet a decision. After every finding, ask the practitioner's question: so what? What should change because we now know this?

An insight is a finding that is actionable and non-obvious — it changes how you understand the problem, and it points somewhere. 'Attendance is low' is a fact; 'attendance collapses on ration-delivery days because mothers must queue elsewhere' is an insight.

A useful recommendation is specific, owned and feasible: who should do what, by when. 'Improve services' helps no one; 'shift the weighing session to avoid the ration-queue clash, with the supervisor confirming the new time' can be acted on tomorrow.

Insight that never returns to the people who generated it is extraction. Share findings back with frontline workers and the community — both to respect them and to check your interpretation against theirs.

The most powerful use of observation is not the one-off study but the feedback loop: routine, light-touch field observation that continuously corrects the programme — supervision visits that actually watch, and act.

Observation at a handful of sites cannot, on its own, prove a statement about a whole district. State what your observation can carry — rich explanation, mechanism, hypotheses — and be honest about what it cannot.

In observation there is no neutral camera. You select, perceive and interpret — and your presence changes the scene. Rigour is not the absence of bias; it is the disciplined management of it.

The eye cannot take in everything, so it filters — and the filter is shaped by what you expect and care about. The famous 'invisible gorilla' experiment showed observers can miss the obvious while counting something else.

Going native is the opposite risk to detachment: the observer so identifies with the group that they stop questioning, adopt its assumptions, and lose the outsider's vantage that made observation valuable.

Keep a reflexive thread in your notes: What surprised me? What irritated me? What did I expect to see? Your reactions are evidence about both the setting and yourself.

Who you are — gender, caste, class, age, language, the agency you represent — shapes what people show you and what you are able to see. Positionality is naming that honestly.

Observation enters people's spaces, routines and vulnerabilities — often without them gaining anything. Ethics is not a clearance form at the end; it is a duty that runs through every minute in the field.

People should know who you are, that you are observing, why, what you will do with what you record, and that they may decline — in a language and form they truly understand. Consent is harder in observation, because settings are fluid and people come and go.

A camera or recorder raises the stakes sharply: images and audio are identifiable, permanent and easily shared. They need their own, explicit consent — separate from consent to be observed.

The observer usually holds more power than the observed — literacy, institutional backing, the ability to leave. In settings of poverty, caste or gender hierarchy, a 'request' from an outsider can feel impossible to refuse.

Covert observation denies people the chance to consent. It can only be contemplated when the knowledge is important, no other method is possible, and harm is minimal — and even then it needs formal ethics approval. In most development work it is neither necessary nor defensible.

People give you their time, their space and their trust. Ethical observation asks what flows back: findings shared in usable form, recommendations that improve the service they use, respect for their time and their account of their own lives.