What an observation is
Definition
Observation
A method that collects data by watching and recording what participants do, rather than asking them about it.
- An observation collects data by watching what people do and recording it.
- Nobody is asked anything, so you record the behaviour itself rather than what someone says about their behaviour.
- Participants may or may not know they are being watched, and that single decision changes both the data and the ethics.
Behavioural categories
Definition
Behavioural categories
Clear, separate and observable actions that a behaviour is broken into so it can be recorded consistently, for example on a tally chart.
- Behavioural categories break the behaviour you are studying into separate, clearly defined actions.
- Each category has to be operationalised, so that anyone watching can tell instantly what counts and what does not.
- Observers then tally each category on a chart, which turns what they see into quantitative data.
Example
- Suppose you want to record helping behaviour in the school canteen.
- "Being kind" is useless as a category, because two observers would never agree on when it had happened.
- "Picks up a dropped item", "holds a door open" and "offers a seat" can each be tallied without argument.
Checking that observers agree
Definition
Inter-observer reliability
How far two or more observers watching the same behaviour agree on what they record; close agreement means the observation is more reliable.
- Two observers watch the same behaviour at the same time and record it separately.
- If their tallies closely match, the recording is consistent and the observation is more reliable.
- If they do not match, the categories are almost always too vague and need rewriting before the study runs for real.
Strengths
- It sidesteps the problem with self-report, because people are poor at describing their own behaviour and here it is recorded as it happens.
- Behaviour can be natural, because participants who do not know they are being watched act exactly as they normally would.
Weaknesses
- You see the behaviour but never the reason, since a tally tells you a pupil walked away from a group but not whether they were upset, bored or late.
- Observer bias is a real risk, because what an observer expects to see affects what they record, especially when a category leaves room for judgement.
Common Mistake
- Do not write a behavioural category that needs an opinion, such as "acting aggressively", because observers will disagree about it.
- Do not confuse two observers agreeing with the observation being worthwhile, because a consistent record can still be measuring the wrong thing.
When to choose an observation
- Choose it when you want behaviour as it actually happens rather than an account of it given afterwards.
- Choose it when the participants could not report reliably on themselves, as with very young children.
- Choose it when asking questions would itself change the behaviour you are trying to study.
- Choose a self-report method instead when the reasons behind the behaviour are what you actually want.
Turning this into marks
- Write every behavioural category as a visible action that could genuinely be tallied.
- Offer a specific fix when asked to improve one, such as sharpening a named category or adding a second observer.
- The mark most often lost is a category expressed as a feeling or a judgement.
Self review
- What does an observation record, and what can it never tell you?
- Why must behavioural categories be operationalised?
- How would you check that two observers agree, and what does close agreement show?
- Give a situation where an observation would beat a questionnaire, and say why.
