You are viewing a free preview of this lesson.
Subscribe to unlock all 10 lessons in this course and every other course on LearningBro.
Edexcel A-Level Psychology requires two things of each topic that go beyond describing theories and studies. The first is a key question — a question of contemporary societal relevance that the topic's concepts can be used to analyse — and the second is a practical investigation, a small-scale study that you design and could carry out using the research methods the topic teaches. Both are examinable, and both are places where the top marks go to students who can apply learning theory rather than merely recite it. This lesson does both. In the first half it takes a genuinely contested social question — does exposure to media violence cause aggression? — and analyses it rigorously through the concepts of the learning approach, weighing what social learning theory can and cannot show. In the second half it models, in full, an ethical observational study of imitation: a structured, non-experimental observation of whether children imitate a peer's play behaviour, complete with a hypothesis, operationalised variables, behavioural categories, a plan for inter-observer reliability, a sampling strategy, a full ethical protocol, and an analysis plan naming both a descriptive statistic and an inferential test (the chi-square test, χ2). Together these show the learning approach doing what it is for: explaining behaviour that matters, using methods that are rigorous and humane.
This lesson addresses the two applied requirements of Edexcel 9PS0 — Paper 1, Topic 4: Learning Theories: the topic's key question of societal relevance and its practical investigation. For the key question you are required to describe a contemporary issue and analyse it using the concepts and theories of the learning approach (here, social learning theory, conditioning and their limits). For the practical you are required to be able to design and report a small-scale study using a method appropriate to the topic — for a learning-approach practical this is naturally an observation — including an aim and hypothesis, operationalised variables, a structured-observation design with behavioural categories and inter-observer reliability, a sampling method, an ethical protocol, and an analysis using an appropriate descriptive statistic and inferential test. In assessment-objective terms, you should be able to describe the key question and the design of a practical (AO1), apply learning-theory concepts and research-methods knowledge to both (AO2), and evaluate the strength of the analysis and the ethical and methodological quality of the design (AO3).
Connects to…
Few questions in applied psychology have a longer public life or higher stakes. From video "nasties" in the 1980s to violent video games such as Grand Theft Auto and Call of Duty, to the ready availability of graphic content on smartphones, the claim that watching or playing at violence makes people — especially children — more aggressive has driven film classifications, game age-ratings (PEGI), broadcasting codes, and periodic moral panics after high-profile crimes. Governments regulate on the assumption that a link exists; the entertainment industry, worth hundreds of billions globally, disputes it. Because the answer shapes law, parenting and industry regulation, and because it bears directly on child welfare, it is a question of clear societal relevance — and the learning approach, with its theory of imitation, is one of the main psychological frameworks brought to bear on it.
The learning approach offers a specific and testable mechanism by which media violence could increase aggression, and equally specific reasons for caution.
| Concept | Application to the question |
|---|---|
| Observational learning / modelling | On-screen characters act as models; viewers can acquire aggressive "scripts" by watching, exactly as Bandura's children acquired mallet-strikes by observation. |
| Identification | Imitation is stronger when the viewer identifies with the model — an admired, attractive or similar protagonist — so a glamorous violent hero is a more potent model than a punished villain. |
| Vicarious reinforcement | When on-screen violence is rewarded (the hero wins, gains status, faces no consequence), viewers vicariously learn that aggression pays; when it is punished, imitation is inhibited. |
| Mediational processes | Whether a viewer imitates depends on attention, retention, reproduction and motivation — media that is salient, repeated and rewarding maximises each, which is why interactive violent games may model more strongly than passively watched scenes. |
| Disinhibition | Repeated exposure may disinhibit aggression already in the repertoire, and desensitisation may reduce the emotional response that normally restrains it — extending the effect beyond simple copying. |
Against this, the same framework supplies the grounds for doubt. Bandura's own studies show imitation of doll-directed acts in a laboratory; generalising to real interpersonal violence is exactly the inferential leap the ecological-validity criticism warns against. Learning theory also predicts that punished on-screen violence should reduce imitation, so blanket claims that "all screen violence increases aggression" misuse the theory. And crucially, most real-world evidence linking media violence to violent behaviour is correlational: heavy consumers of violent media may be more aggressive because aggressive dispositions draw people to violent media (a selection effect), or because a third factor — a chaotic home, peer influence — drives both. Social learning theory establishes that imitation is possible and specifies when it is likelier; it does not, by itself, establish that media violence is a cause of societal violence.
It is worth being clear about what kinds of evidence bear on the question, because a top-band key-question answer distinguishes them rather than treating all "studies" as equal. Laboratory experiments (Bandura's tradition, and later short-term studies exposing participants to violent clips or games and then measuring a proxy for aggression) can establish causation because the exposure is manipulated and confounds are controlled — but they buy that internal validity at the cost of ecological validity, since a laboratory proxy for aggression (blasts of noise, hot-sauce allocation, a Bobo doll) is not real violence, and the effect measured is immediate and short-lived. Correlational studies and surveys can examine real media diets and real aggression in large samples, gaining ecological validity, but cannot show direction of causation and cannot rule out third variables. Longitudinal studies, which follow the same people over years, are the design best placed to separate socialisation (media making people aggressive) from selection (aggressive people choosing violent media), because they can ask whether early media exposure predicts later aggression once earlier aggression is held constant — and their findings are genuinely mixed, with some reporting small predictive effects and others none. The honest summary is that the psychological community has not reached consensus: professional bodies have at times issued cautionary statements about violent media, while other researchers argue the effects are small, inconsistent, and confounded. A strong answer names this disagreement rather than pretending the science is settled, and uses learning theory to explain why an effect is plausible while conceding why it is so hard to prove.
The learning approach also has recognised rivals and supplements for this question that a strong answer can acknowledge. Catharsis theory makes the opposite prediction — that acting out aggression vicariously through media might discharge aggressive impulses and reduce real aggression — though the evidence for catharsis is weak. Cognitive priming and script theory propose that media violence works less by simple imitation than by making aggressive thoughts, expectations and problem-solving "scripts" more accessible, so that ambiguous social situations are read as hostile. Desensitisation research suggests repeated exposure blunts the physiological and emotional response to violence, weakening a normal restraint. These are not contradictions of social learning theory so much as elaborations of the cognitive-mediational side of it — and citing them shows the examiner that you can situate the learning-approach analysis within the wider debate.
graph TD
MV[Exposure to media violence] --> SLT{Learning-approach mechanisms}
SLT --> A[Modelling of aggressive scripts]
SLT --> B[Identification with admired model]
SLT --> C[Vicarious reinforcement if violence rewarded]
SLT --> D[Disinhibition / desensitisation]
A --> AGG[Increased aggression?]
B --> AGG
C --> AGG
D --> AGG
CAUTION[Cautions] --> E[Evidence largely correlational]
CAUTION --> F[Selection: aggression draws viewers to violent media]
CAUTION --> G[Doll-directed lab findings may not generalise]
E --> AGG
F --> AGG
G --> AGG
style AGG fill:#e67e22,color:#fff
style CAUTION fill:#2980b9,color:#fff
The defensible conclusion is not a flat yes or no but a conditional one. The learning approach shows that media violence can contribute to aggression — through modelling, identification and vicarious reinforcement — and that its influence is greatest where the violence is rewarded, performed by an identified-with model, and repeated or interactive. But because the strongest real-world evidence is correlational and vulnerable to selection effects, and because laboratory imitation of a doll cannot be equated with real violence, media violence is best understood as one risk factor among many rather than a sufficient cause. This nuanced, theory-grounded answer — explaining how an effect could occur, when it is strongest, and why causal certainty is elusive — is what the key-question assessment rewards.
The topic's classic study, Bandura's Bobo doll, is ethically compromised because it may have taught aggression to children. A well-designed learning-approach practical keeps Bandura's observational-learning logic while removing the ethical objection: instead of modelling aggression, we observe whether children imitate a neutral, prosocial play behaviour. The following is a complete, examinable design.
Aim: to investigate whether children imitate a novel, neutral play behaviour they have seen an adult model perform, compared with children who have not seen the model.
This practical uses a structured (systematic) observation within an independent-groups design. It is a controlled observation, not a true experiment in the everyday sense, but it manipulates one factor (model present vs absent) and observes a naturally occurring behaviour, which is exactly the design Edexcel expects for a learning-approach observation.
| Variable | Operationalisation |
|---|---|
| Independent variable (IV) | Whether the child observes the model: model-present condition vs no-model (control) condition |
| Dependent variable (DV) | Whether, within a fixed 5-minute free-play period, the child performs the target action (e.g. placing a toy block on their head before stacking it) — coded as a yes/no (nominal) outcome per child |
| Controlled variables | Same room, toys, model, script, time of day, 5-minute observation window, and standardised instructions for every child |
The target behaviour must be novel and unambiguous — a specific, arbitrary action the child is unlikely to produce spontaneously (so that its appearance signals imitation), and one that is neutral so that no harm can result. "Placing a block on the head, saying 'up it goes', then stacking it" is the kind of distinctive, harmless act that works, mirroring how Bandura used distinctive acts to show specific imitation.
Structured observation requires operationalised behavioural categories decided in advance, so that observers record objectively rather than interpreting. A partial coding scheme:
| Category | Definition (what counts) |
|---|---|
| Full imitation | Child performs the complete target action (block on head + phrase + stack) |
| Partial imitation | Child performs part of the target action (e.g. block on head but no phrase) |
| Non-imitative play | Child plays with the toys but does not perform the target action |
| No engagement | Child does not play with the target toys in the window |
For the primary analysis the DV is collapsed to a binary category — imitation present (full or partial) vs absent — recorded once per child. Recording is by event sampling (noting each occurrence of a category during the 5-minute window) rather than time sampling, because the target act is discrete and may occur only once.
Because "did the child imitate?" involves a judgement, the observation must be checked for inter-observer reliability: two independent observers, blind to the child's condition where possible, code the same children using the same categories. Agreement is quantified — for example as a percentage agreement (agreements/total observations×100), with a conventional threshold of at least 80%, or more robustly using Cohen's kappa, which corrects for chance agreement. Piloting the categories on a few children first lets the observers refine ambiguous definitions before the main study, which raises reliability. High inter-observer reliability is the methodological feature that turns a subjective impression into defensible data — and, as the classic-study lesson noted, it is exactly what gave Bandura's observation its rigour.
Designing this study ethically is the whole point of choosing a neutral behaviour, and the protocol should map onto the BPS principles explicitly.
Subscribe to continue reading
Get full access to this lesson and all 10 lessons in this course.