Overview – Observations

An observation is a non-experimental method (i.e. there is no independent variable that is manipulated) where researchers watch and record naturally occurring behaviours (or behaviours that happen in controlled settings) as they happen.

In psychology, observations can be categorised in 4 ways:

Observations can collect qualitative data (detailed descriptions of behaviours and experiences) or quantitative data (numerical measurements, such as frequency counts of how many times a specific behaviour occurred).

Researchers often use behavioural categories and coding schemes to operationalise complex behaviours into clear and observable actions that can be recorded consistently. Researchers may record behaviour every time the target behaviour occurs (event sampling) or at pre-set time intervals (time sampling). For more on these details, see observational designs.

Types of observation

There are 4 ways to categorise types of observation:

Structured vs. unstructured observations

The structured vs. unstructured distinction is about how data is operationalised and recorded by the researchers during the observation. In short:

  • Structured observations are about using a rigid system of pre-defined behavioural categories to collect specific quantitative data.
  • Unstructured observations are about recording everything you see to collect rich qualitative data.
Structured Observation Unstructured Observation
Summary Researchers decide pre-determined behavioural categories to record and then count those. No pre-set behavioural categories are used. Researchers just record all behaviours as they happen.
Key Features Behavioural categories & coding schemes: Clear definitions set in advance to standardise data collection.
Sampling techniques: Uses event sampling (tallying specific actions) or time sampling (instantaneous scan, predominant activity, one-zero sampling).
Yields quantitative data.		 Continuous recording: Observers write down descriptive notes on everything taking place.
No restrictive checklists or sampling intervals.
Yields rich qualitative data.
Examples Bandura et al (1961): Used pre-determined behavioural categories (e.g. imitative physical aggression, verbal aggression) recorded at 5-second intervals (time sampling) via a one-way mirror.
Piliavin et al (1969): Used structured tallying by two female observers to record total helpers, speed of response, and demographic details of bystanders on the subway carriage.		 Freud (1909): Unstructured qualitative observations recorded through ongoing notes and letters sent by Little Hans’ father regarding the boy’s phobia.
Rosenhan (1973): Researchers made unstructured qualitative observations of psychiatric hospital conditions. They recorded notes on staff behaviour, patient treatment, and daily experiences without using pre-determined behavioural categories.
Strengths High inter-rater reliability: Uses pre-set categories, so observations are consistent and easier to replicate across different observers.
Efficient: Produces quantitative data that can be quickly summarised and statistically analysed. Also, observers waste less time noting irrelevant actions and only focus on coded behaviours.		 In-depth detail: Notes all actions, including unexpected or complex behaviours that would not be anticipated if relying on pre-set coding.
High validity: Captures complex and natural behaviours without forcing actions into rigid boxes.
Weaknesses Limited detail: Important unexpected or complex actions might be ignored if they were not included in the pre-determined coding scheme.
Lower validity: May oversimplify behaviour by forcing observations into fixed categories that do not fully reflect reality.		 Inefficient: Produces qualitative data that takes longer to compare, summarise, and analyse systematically. Also it’s hard to note everything down accurately – especially when observing multiple participants.
Subjectivity: Open-ended notes are more subjective, so different observers may record or interpret behaviour differently.

Naturalistic vs. controlled observations

The naturalistic vs. controlled distinction is about where the observation takes place and how much control the researchers have over the environment. In short:

  • Naturalistic observations are about watching participants in their real-world environment without altering the setting.
  • Controlled observations are about watching participants in an artificial or manipulated setup where outside influences are regulated.
Naturalistic Observation Controlled Observation
Summary Behaviour is observed in the participant’s natural environment without any intervention or manipulation by the researcher. Behaviour is observed in an artificial, controlled setting (such as a lab) where extraneous variables are regulated.
Key Features Natural environment: Takes place where the target behaviour normally occurs.
No researcher intervention: The environment is left entirely as it is.
Focuses on spontaneous behaviour in everyday life.		 Manipulated environment: Setup is controlled or artificial to trigger specific responses.
Standardised procedures: Conditions are kept identical across participants.
Focuses on controlled responses under specific conditions.
Examples Piliavin et al (1969): Observed spontaneous helping behaviour in a real NYC subway carriage during normal, everyday journeys.
Levine et al (2001): Observed non-emergency helping behaviours towards strangers in real-world public spaces across 23 international cities.		 Bandura et al (1961): Observed children’s aggression in a standardised, artificial room containing specific toys through a one-way mirror.
Ainsworth (1978) – Strange Situation: Observed infant-attachment behaviours in a controlled playroom with scripted entrances and exits.
Strengths High ecological validity: Behaviours are authentic and true-to-life because participants are in their normal surroundings.
Reduced demand characteristics: Participants are less likely to change their behaviour because they are in a familiar environment.		 High control over extraneous variables: Prevents outside distractions or confounding factors from skewing the results.
Easy to replicate: Standardised environments and setups allow other researchers to easily repeat the study.
Weaknesses Low control over extraneous variables: Uncontrolled background events (e.g. weather, crowds) can distort the results.
Hard to replicate: Natural environments change constantly, making exact repetition difficult.		 Low ecological validity: The artificial environment can lead to unnatural and unrealistic behaviour.
Risk of demand characteristics: Participants may realise they are in a study and alter their normal behaviour (Hawthorne effect).

Participant vs. non-participant observations

The participant vs. non-participant distinction is about the role of the researcher within the social setting during the observation. In short:

  • Participant observations are when the researcher actively joins the group being studied to gather insider data.
  • Non-participant observations are when the researcher is separate from the group and gathers data without getting directly involved.
Participant Observation Non-Participant Observation
Summary The researcher actively joins participants and interacts with them while collecting data. The researcher remains separate from the participants and observes from a distance without joining in.
Key Features First-hand experience: The observer immerses themselves in the environment to gain insider insight (verstehen).
Direct interaction: Researchers engage with participants, which can influence natural group dynamics.
Yields rich qualitative data.		 Clear separation: Observers keep their distance (e.g. behind a one-way mirror or from across a room).
Objective recording: Observers do not participate, reducing personal influence on the group.
Yields mostly quantitative data.
Examples Rosenhan (1973): Researchers gained admission as pseudopatients into psychiatric hospitals, interacting directly with patients and staff to observe treatment.
Freud (1909): Little Hans’ father acted as an observer of Hans and recorded detailed observations of the boy’s behaviour and experiences while interacting directly with him.		 Bandura et al (1961): Observers watched children through a one-way mirror without interacting with them.
Ainsworth (1978): Researchers observed mother-child interactions through a one-way mirror in a controlled playroom setup.
Strengths Increased insight: Provides unique, deep understanding of participant experiences and reasons behind behaviours.
High validity: Captures authentic group dynamics that an outside observer might misinterpret or miss entirely.		 High objectivity: Researchers maintain professional distance, lowering the risk of personal bias or emotional involvement.
Easier to record: Researchers are free from the distraction of participating, which allows for continuous and accurate note-taking.
Weaknesses Risk of losing objectivity (‘going native’): Researchers may identify too closely with participants, biasing their data collection.
Researcher influence: The researcher’s presence within the group may alter genuine participant behaviour.		 Lack of insider depth: Researchers may misunderstand or misinterpret the true meaning of observed actions without context.
Lower validity: The researcher might miss subtle social cues or motivations that only people in the group would pick up on.

Overt vs. covert observations

The overt vs. covert distinction is about whether participants are aware that they are being observed by the researcher. In short:

  • Overt observations are when participants know they are being observed and have given their consent.
  • Covert observations are when participants do not know that they are being observed.
Overt Observation Covert Observation
Summary Participants are fully aware that they are being studied and observed by the researcher. Participants are completely unaware that they are being observed or studied.
Key Features Open setup: The researcher’s presence and intentions are made clear in advance.
Ethical alignment: Prior consent is obtained from all participants.
Risk of participants altering behaviour due to being watched.		 Hidden setup: Uses hidden cameras, one-way mirrors, or undercover observers.
Deception: Participants are unaware of data collection occurring.
Ensures entirely uninhibited and natural behaviour.
Examples Bandura et al (1961): Children were brought into a lab setting where adults and experimenters were present, knowing they were being watched.
Ainsworth (1978): Mothers gave full consent to be observed with their infants in a staged playroom scenario.		 Piliavin et al (1969): Subway passengers had no idea they were part of a study on helping behaviour.
Rosenhan (1973): Psychiatric hospital staff were unaware that the pseudopatients were collecting data on their conduct.
Strengths Ethical: Respects participant autonomy by obtaining informed consent and avoiding deception.
Easy data collection: Researchers can openly take notes or use recording equipment without hiding.		 Reduced demand characteristics: Participants behave naturally because they do not know they are being observed.
High validity: Participants act genuinely because they do not feel watched or judged.
Weaknesses Demand characteristics / Hawthorne effect: Participants may alter their behaviour to act more favourably or as expected.
Lower validity: Behaviour may become less natural because participants know they are being observed.		 Ethical issues: Raises ethical concerns regarding lack of informed consent, deception, and invasion of privacy.
Practical difficulties: Hard to record notes discreetly without raising suspicion or missing details.

Observation design

Observation design is about how researchers structure and record observations, including:

Behavioural categories

Behavioural categories are when researchers break down a broad, complex target behaviour into a clear, specific, and operationalised list of observable actions (a coding scheme).

For example, when observing aggression in children playing, researchers must decide how to identify aggressive behaviour and how they will collect data on it. They need to define aggressive behaviour – this may mean breaking it down into distinct categories such as:

  • punching,
  • kicking,
  • shouting,
  • and so on.

By doing this each aggressive behaviour can be consistently identified and recorded and, if necessary, quantified.

For example, researchers might organise observational data in the following way:

Participant Sex Age
 Observed aggressive behaviours Aggression rating (1-10)
A F 2 none 1
B F 1 none 1
C M 3 P 7
D F 2 S 4
E M 2 S 5

Rather than writing complete descriptions of behaviours, the behaviours can be coded into categories. For example, P = punching, and S = shouting. Researchers can also create numerical ratings to categorise behaviour, like the aggression rating example above.

Time vs. event sampling

This is about how and when behaviours are sampled during the observation:

Time Sampling Event Sampling
What behaviour is happening at this point in time? How many times has this behaviour happened?
Time sampling is a method where researchers record behaviour at pre-set time intervals rather than continuously. Event sampling is a method where researchers record every occurrence of a specific target behaviour during the observation period.
For example, when observing aggression in children playing, a researcher might check every 5 minutes and record whether each child is showing aggressive behaviour at that moment. For example, when observing aggression in children playing, a researcher might tally every time a child hits, kicks, pushes, or shouts throughout a one hour observation.

<<< Previous: The experimental method

Next: The self-report method >>>

<<< Back to OCR psychology revision notes
<<< Back to AQA psychology revision notes