Observation

Uncontrolled: conducted in the natural setting without such prearrangement, recording whatever occurs. Its merit is flexibility and the capacity to notice the unanticipated; its limitation is that records are hard to compare and to quantify.

Structured and unstructured

The same distinction expressed in terms of the instrument: structured observation uses a schedule with defined categories; unstructured observation records descriptively in field notes.

Direct and indirect

Direct observation watches the behaviour itself; indirect observation examines its traces: wear on a floor, the contents of a waste bin, the state of a register. Indirect observation is entirely unobtrusive and is a useful check on what people report.

Its merits as a tool

1. It records behaviour, not reports of behaviour, which is its decisive advantage over every other tool in this module.
2. It reaches what respondents cannot report: habits they do not notice, contradictions between what they say and do, and the ordinary detail nobody thinks worth mentioning.
3. It does not depend on the respondent's willingness or ability to answer, so it reaches the very young, the ill, and the unwilling.
4. It is in the natural setting.
5. It produces the detail from which good questions for other tools are built, which is why it belongs to exploratory work in [Types of Methodology].

Its limitations

1. The observer effect. People who know they are watched behave differently. It diminishes with time as the observer becomes familiar, which is the standard remedy.
2. Subjectivity. What is noticed and how it is interpreted depend on the observer, and two observers of one scene produce different records.
3. It cannot reach the past, the future, or the private. A great deal of social life, including most of what happens inside a household, cannot be observed at all.
4. It cannot reach opinions, attitudes and motives, which have to be asked about.
5. Small coverage, and therefore no generalisation.
6. Time and cost.
7. Recording problems: writing during observation disrupts, and writing afterwards loses accuracy.
8. Ethical problems where the observation is covert.

Reliability and validity, and how they are improved

Reliability is consistency: would another observer, or the same observer on another occasion, record the same thing? Validity is whether what is recorded is what the researcher intends to measure.

How they are improved: defining the categories precisely in advance; using more than one observer and measuring their agreement; training observers together; recording immediately; using audio or video where the setting and ethics permit; prolonged presence, so that the observer effect wears off; and triangulating with interviews and documents.

A worked example

The question: what actually happens to matters listed for a day.

Non-participant, controlled observation is the right design. The observer sits in the public area with a schedule recording, for each matter called: the time called, the time taken, who appeared, what order was made, whether the matter was reached at all, and the stated reason if it was not.

What this yields that nothing else does. The register records "adjourned at the request of the parties" for a matter that was in fact never reached because the court rose. The observation and the record disagree, and the disagreement is the finding, exactly as [Research Methods: Documentary, Empirical and Survey] predicts.

The observer effect, in this setting, is small: the court is public, the observer is one of many people present, and proceedings do not alter for them. This is worth noting because it is a setting in which observation is unusually reliable.

Its limits are equally visible. The observer sees one courtroom of many; cannot see the conversation in the corridor that produced the adjournment; and learns nothing about what the litigant thinks or what the day cost them. For those, the interview and the survey are needed, which is the next two chapters.

Quick revision

1. Observation is the systematic, purposive watching and recording of behaviour in its natural setting. P. V. Young: a deliberate study through the eye.
2. Scientific observation is purposive, planned, systematically recorded, checked and theoretically related; casual seeing is none of these.
3. Participant observation reaches meaning and the group's own view; its dangers are the observer effect, going native, non-generalisability and, where covert, ethical objection.
4. Non-participant observation gains detachment and accuracy of record, and loses access to meaning.
5. Controlled observation uses a schedule and defined categories, giving precision and comparability; uncontrolled observation is flexible and notices the unanticipated.
6. Direct observation watches behaviour; indirect observation reads its traces and is wholly unobtrusive.
7. Its decisive merit: it records behaviour rather than reports of behaviour.
8. Its decisive limits: it cannot reach the past, the private, or attitudes and motives; coverage is small; and the observer alters what is observed.

Test yourself

1. What distinguishes scientific observation from ordinary seeing? That it serves a purpose formulated in advance; that it is planned as to what, where, when and for how long; that it is recorded systematically and immediately by a stated procedure rather than remembered; that it is checked for reliability, ideally by a second observer, and for validity against other sources; and that it is related to a body of theory, so that what is seen is interpreted rather than merely listed.

2. Distinguish participant from non-participant observation, with merits and limits. In participant observation the researcher joins the group and takes part in its life, which gives access to meanings, motives and the group's own point of view and reaches groups that would answer no questionnaire; its dangers are that the observer alters the group, may go native and lose analytical detachment, cannot easily record without disrupting, and, if concealed, observes people who have not consented. In non-participant observation the researcher watches without taking part, gaining detachment and accurate recording while losing access to meaning and risking misinterpretation for want of insider knowledge.

3. Distinguish controlled from uncontrolled observation. Controlled observation is conducted under prearranged conditions with a schedule of defined categories and often a time sample, so that records are comparable and countable; its cost is that it can record only what was anticipated. Uncontrolled observation is conducted in the natural setting without such prearrangement and records whatever occurs; it is flexible and can notice the unanticipated, but its records are difficult to compare or to quantify.

4. State the decisive merit and the decisive limitations of observation. Its decisive merit is that it records behaviour rather than reports of behaviour, so it reaches what respondents cannot or will not tell an interviewer, including habits they do not notice and contradictions between what they say and what they do. Its decisive limitations are that it cannot reach the past, the future or the private, that it cannot reach attitudes and motives, which must be asked about, that its coverage is too small to support generalisation, and that those observed behave differently for being watched.

5. How are the reliability and validity of observation improved? By defining the categories to be recorded precisely before observation begins; by using more than one observer and measuring the agreement between them; by training observers together so that they apply categories alike; by recording immediately rather than from memory; by using audio or video where the setting and ethics permit; by remaining in the field long enough for the observer effect to wear off; and by triangulating the observations with interviews and documentary records.