4.2.3.1 - Observational design

What observational design decides

An observation is only useful if the researcher has decided, before data collection, exactly what will be recorded and when it will be recorded. That is the purpose of observational design.

For this AQA requirement, the three core decisions are:

Imagine a psychologist wants to observe "prosocial behaviour" in a sixth-form common room. That phrase is too vague for reliable recording. One observer might count lending a pen; another might only count comforting a distressed friend. A better design would define specific categories such as "offers practical help", "shares equipment", and "comforts another student", then decide whether to record each event as it happens or sample behaviour at intervals.

This is where observational design becomes scientific. Instead of relying on a researcher's loose impression, the study uses a pre-planned recording system. That can improve objectivity because observers have clearer rules. It can also make the data easier to analyse because behaviours can be counted, compared and summarised.

However, design choices also create limits. If a behaviour is not included in the categories, it may not be recorded. If the sampling method misses short-lived behaviour, the data may not represent what really happened. A strong answer therefore treats observational design as a set of trade-offs, not as a perfect solution.

Behavioural categories

Behavioural categories are the foundation of a structured observation. They break a broad target behaviour into specific actions that an observer can see or hear and then record.

The most important rule is: if the observer has to guess a mental state, the category is not good enough yet. "Anxious" is not directly observable. "Bites fingernails", "looks away from speaker" or "asks to leave the room" are observable.

Good behavioural categories should be:

This is a form of operationalisation. The researcher turns an abstract idea such as aggression, helping or obedience into behaviours that can be recorded in the same way by different observers.

AO2 example: Dr Mensah is observing children's aggressive behaviour in a playground. "Acts aggressively" would be too vague. Better categories would be:

• "pushes another child with hands";
• "kicks another child";
• "throws an object at another child";
• "shouts an insult at another child".

These categories describe actions rather than inferred motives. Add recording rules: for example, count a new push only after hand contact has ended and begun again; define what happens if actions occur simultaneously. If categories are intended to be mutually exclusive, use a stated priority rule or define separate event units so one act is not counted twice. Pilot the schedule and compare independent observers’ records to find definitions needing clarification.

Observable indicators are not proof of a mental state: looking away may reflect distraction, a noise or a deliberate pause, not necessarily anxiety. Interpretation must stay within what the behaviour supports.

For example, a child can talk while smiling. These are two observable actions, so a schedule can legitimately tick both. That differs from counting the same smile twice as both a "smile" and a "grin". State whether the unit is a single action, a whole episode or the behaviour at one sampled instant.

The strength of behavioural categories is reliability. Clear categories reduce the chance that observers record only what fits their expectations. The limitation is reductionism. Complex social behaviour may be squeezed into a checklist, so the data can lose context and meaning. For example, "puts arm around another student" may be coded as comfort, but the coding system may miss whether the action was welcome, awkward or sarcastic.

Event sampling

Once categories are chosen, the researcher must decide how to sample behaviour. Event sampling records the target behaviour whenever it occurs.

Event sampling often uses a tally sheet. Each time the target behaviour occurs, the observer adds a mark beside the relevant behavioural category.

This illustrative tally summary contains 12 events in total, not necessarily 12 different children. One child can contribute several events. A frequency table describes recorded events; it does not identify their motives.

Worked recording decision

During a ten-minute observation, a student raises a hand, lowers it, and raises it again. If a new event begins after the hand has been lowered, record two events. A hand held up for 20 seconds is one event, not 20. The rule must be agreed before counting, otherwise observers could generate different frequencies from the same behaviour.

A comparison also needs equal observation time or a rate. Six requests in ten minutes and six in thirty minutes have equal counts but different rates: 0.6 and 0.2 requests per minute. The first observed rate is three times the second. This describes the observed sessions; it does not prove why the difference occurred.

Event sampling is useful when behaviours are:

• discrete, meaning each occurrence has a reasonably clear start and end;
• not too frequent, so the observer can keep up;
• central to the aim, so counting every occurrence is worth the effort.