Unit 1: Social and cognitive psychology AS

Obedience is a change in behaviour in response to a direct order from someone perceived to have authority. Agency theory explains a change in responsibility; social power theory explains why the authority can influence the target.

Milgram used a standardised laboratory task in which a volunteer 'teacher' believed they were increasing electric shocks to a learner after errors. In the original study, 65% reached the 450 V maximum, showing that ordinary adults could obey a legitimate experimenter despite visible conflict.

The variations isolate situational influences: reduced prestige, proximity or legitimate authority increased dissent. Their standardisation supports comparison, but an artificial shock task, a male US sample and serious deception and distress limit generalisation and ethics.

Obedience and dissent reflect an interaction between the person, the immediate situation and the culture. A factor is supported when changing or comparing it produces a reliable difference, not merely because it sounds plausible.

Dissent becomes easier when responsibility is salient, the authority's legitimacy is weakened or another person models resistance. These routes can work together rather than competing as single explanations.

Conformity is changing behaviour or belief because of real or imagined group pressure. Its depth depends on whether the change is public only, tied to group membership, or genuinely accepted.

Normative social influence produces acceptance or avoidance of rejection, so it often supports compliance. Informational social influence occurs when others are treated as evidence about reality, especially in ambiguous or difficult situations, so it can produce internalisation.

Asch tested whether a unanimous majority could make a person publicly give an obviously wrong line judgement. A real male student answered after confederates who had been instructed to choose the same incorrect comparison line on critical trials.

About three quarters conformed at least once, although many responses still resisted the majority. Later variations showed that conformity rose as a small majority formed, fell sharply when one partner broke unanimity, and changed with task difficulty and whether answers were public.

The controlled line task isolates majority size, unanimity and response conditions, supporting a situational explanation and replication. However, the artificial task, awareness of being studied and restricted US male student sample limit everyday and population validity; historical norms may also change conformity.

Minority influence occurs when a smaller group changes majority judgement. Moscovici's account proposes a conversion process: an unexpected, coherent minority creates conflict, makes the majority examine the issue more deeply and may produce private internalisation.

The same three levels of explanation organise conformity and minority influence: individual differences alter susceptibility, the situation changes pressure and credibility, and culture shapes the norms against which agreement is judged.

A good questionnaire or interview turns a construct into questions that respondents can understand and answer without being steered. The wording, order and researcher behaviour must serve the aim and produce analysable data.

1. Define the target construct and population. 2. Write one clear idea per question, avoiding leading wording, jargon, double negatives and double-barrelled items. 3. Choose open or closed responses to match the evidence needed. 4. Pilot the instrument and revise ambiguous wording or an exhausting length. 5. Standardise instructions, consent, privacy and recording.

An interviewer's tone, appearance, prompts or reactions can change responses; social desirability and acquiescence can do the same. Neutral wording, consistent prompts, anonymity where possible and reflexive recording of researcher decisions reduce these effects.

Primary data are gathered first-hand for the current research aim. Secondary data already exist because another researcher or organisation collected or compiled them for an earlier purpose.

Judge fitness rather than assuming one source is superior: check operational definitions, population, date, missing data, collection method and ethical permission. A new analysis of an existing dataset remains secondary data for the new researcher.

Interview structure controls how far the researcher can adapt the conversation; question format controls the kind of response produced. These are separate design decisions.

An alternative hypothesis states the predicted relationship or difference between operationalised variables. It must be testable with the planned data and population; it may be directional when prior evidence justifies a direction, or non-directional when only a difference or association is predicted.

Directional: 'Students shown a unanimous group response will give more conforming answers than students shown a mixed response.' Non-directional: 'There will be a difference in conforming answers between the two conditions.' Both define what is measured and compared.

The target population is everyone to whom the claim is intended to apply; the sampling frame is the accessible list or source; the sample is the people actually studied. Define inclusion and exclusion criteria before recruitment, choose a sampling technique, record non-response and compare sample characteristics with the target population.

A sampling technique is the rule used to move from a target population or sampling frame to the people studied. The rule determines which selection biases are likely.

Quantitative analysis should reveal the pattern without hiding the scale of measurement or the spread. First identify whether values are categories, discrete counts or continuous measurements; then choose compatible summaries and displays.

Fractions, ratios and percentages express the same part-whole information in different forms. Keep the denominator explicit: 12 conforming responses out of 30 is $12/30$, a conforming-to-nonconforming ratio of $12:18$, and 40%.

A distribution describes how scores are spread. Its shape affects which measure represents a typical score and how extreme values should be interpreted.

The mean uses every value and is sensitive to extremes. For a strongly skewed distribution, the median usually gives a more representative centre; the range or interquartile pattern should be read alongside it. A histogram helps reveal the shape.

Thematic analysis is a systematic way to identify recurring patterns of meaning in qualitative data such as open questionnaire responses or interview transcripts.

1. Read the complete dataset repeatedly. 2. Code relevant segments with concise labels. 3. Group related codes into candidate themes. 4. Check each theme against its extracts and the whole dataset. 5. define and name distinct themes. 6. Report the pattern with brief anonymised evidence and explain how it addresses the aim.

A coding framework and a record of decisions make the analysis transparent. A second coder can test whether the definitions are applied consistently; disagreements should lead to clarification, not automatic majority voting. Reflexivity helps the researcher examine how expectations shaped coding.

The BPS Code of Ethics and Conduct (2009) requires researchers to protect dignity, autonomy, privacy and wellbeing while producing trustworthy knowledge. Ethical planning begins before recruitment and continues through storage, reporting and debriefing.

A risk assessment identifies hazards, who may be affected, likelihood and severity, preventive controls, stopping rules and follow-up. Extra safeguards are required for vulnerable participants, sensitive topics or unequal power relationships.

Moscovici et al. tested whether a consistent minority could shift a majority's colour judgements. Groups of six women contained four naïve participants and two confederates; each group judged 36 blue slides aloud.

The difference supports consistency as a cause of minority influence under these controlled conditions. Standardised slides and responses make the conditions comparable and replicable. However, an artificial colour task, female US student sample, deception and the small public response shift limit generalisation and do not directly measure lasting private conversion.

Burger asked whether Milgram-like obedience would still occur while reducing ethical risk. Seventy adults aged 20-81 were selected through a two-stage screening process; participants with relevant psychological or medical risks were excluded and were reminded repeatedly that they could withdraw.

In the base condition, a participant acting as teacher used an apparent shock generator after a confederate learner's errors. The procedure stopped at 150 V, the point at which Milgram's learner first strongly protested. In a modelled-refusal condition, another apparent teacher refused after 90 V before the real participant continued.

In the base condition, 70% continued past 150 V. The modelled refusal did not produce a clear reduction. The result suggests substantial obedience remained at the comparable decision point, but stopping early means it cannot show how many would have reached 450 V.

Standardisation supports reliability and comparison, and screening, a clinical psychologist, withdrawal reminders and an early stopping point reduced risk. Yet volunteers who passed screening were selective, the laboratory shock task lacked everyday realism, deception and distress remained, and the safeguards limit a direct replication claim.

Yi Huang et al. tested whether other people's opinions leave a lasting change in private judgement. Seventeen Chinese university students rated the attractiveness of 280 neutral-expression photographs of young Chinese women on a computer.

After each initial rating, participants saw an alleged average rating from 200 peers; on most trials it was experimentally set above or below their own. The faces were rated again after delays without showing the peer rating, so later movement toward the earlier group value could be measured without immediate public pressure.

Judgements shifted toward the group opinion shortly after exposure, but the effect was no longer detectable after three days. This supports a short-lived internal influence rather than an indefinitely stable conversion.

Computer presentation and many stimuli standardised exposure and reduced the effect of any one face. However, a very small Chinese student sample, an artificial attractiveness task and fabricated peer ratings limit population and ecological validity; repeated faces may also create memory or demand effects.

Haun et al. compared whether children and other great apes would abandon a previously rewarded choice after seeing peers choose differently. The sample contained 18 children, 12 chimpanzees and 12 orangutans.

Participants learned which of three coloured sections of a box delivered a reward, reaching a preference criterion on 8 of 10 trials. They then watched three demonstrators choose a different section twice and receive rewards. On three further rewarded choices, researchers recorded whether participants kept their preference, followed the majority or chose elsewhere.

Children were more likely than the apes to switch to the peer-endorsed option, whereas chimpanzees and orangutans usually retained the choice that had worked for them. The comparison suggests that human children are especially responsive to majority information in this setting.

A common apparatus, criterion and video coding support comparison and reliability. But small, age- and species-specific samples limit generalisation; the task may not have equal meaning across species, and copying demonstrators can reflect social learning rather than conformity to group pressure.

The practical must use a questionnaire to collect both quantitative and qualitative data about a social-psychology difference, while linking every design, analysis and ethical decision to the stated aim.

Standardise instructions and conditions, record non-response and missing data, and keep raw data securely. Improvements must target a demonstrated limitation: for example, clearer piloted wording improves validity, while a more representative sample improves generalisability.

Atkinson and Shiffrin's multi-store model treats memory as information moving through three stores. Attention moves selected sensory input into short-term memory (STM); maintenance rehearsal keeps it active and can transfer it to long-term memory (LTM); retrieval returns stored information to STM for conscious use.

Serial-position findings support separate short- and long-term contributions, and preventing rehearsal rapidly reduces STM recall. HM's intact immediate memory but severe difficulty forming new declarative long-term memories also supports separable stores.

Baddeley and Hitch's working-memory model explains short-term memory as an active, limited-capacity system whose components can process different kinds of information at the same time.

Dual-task performance, the word-length effect and cases such as KF support partly separate components. The model also explains why listening to speech while writing verbal notes can overload the phonological system.

Bartlett's reconstructive account proposes that recall combines incomplete memory traces with schemas - organised knowledge built from prior experience and cultural expectations. The result is a plausible reconstruction, not a literal recording.

When details are missing or ambiguous, a schema guides selection and interpretation. Information may be omitted, rationalised, transformed into a more familiar form or added through confabulation. Assimilation changes new information so that it fits an existing schema, although schemas can also be updated by experience.

Two witnesses can encode the same event yet later describe it differently because their expectations about people, places or likely actions fill different gaps. Bartlett's War of the Ghosts and office-schema research found familiar substitutions and recall of schema-consistent items that were not present.

An experiment manipulates an independent variable (IV), measures a dependent variable (DV) and controls alternatives so that a causal explanation can be tested. The laboratory-field distinction concerns the setting and degree of control, not whether an IV exists.

Design from the hypothesis: operationalise IV and DV, choose a design and sample, standardise instructions and timing, control plausible confounds, pilot the task, obtain consent and manage risk, then record data consistently. A control condition provides a baseline when one is needed.

The independent variable is the condition the researcher changes or compares; the dependent variable is the measured outcome expected to respond. Both must be operationalised so another researcher could reproduce exactly what was varied and recorded.

A variable name such as 'noise', 'memory' or 'aggression' is not enough. State the conditions, dose or category for the IV and the observable score, unit, timing and scoring rule for the DV.

An experimental hypothesis predicts that an operationalised IV will affect an operationalised DV. A null hypothesis predicts no difference or relationship in the population and states that any sample pattern is due to chance.

Both hypotheses must identify the population where relevant, the exact conditions and the measured outcome. The experimental hypothesis can be directional or non-directional; the null remains a no-effect statement.

A directional hypothesis predicts which condition will score higher or whether a relationship will be positive or negative. A non-directional hypothesis predicts a difference or relationship without choosing its direction.

Choose the tail before seeing the results. A well-supported theory or consistent previous evidence can justify one-tailed testing; uncertainty or credible effects in either direction calls for two-tailed testing.

Experimental design determines which participants provide data in each condition. The best choice controls the most serious alternative explanation without creating a larger one.

Operationalisation turns an abstract construct into a replicable manipulation or measure. Control then protects the comparison from variables other than the IV.

Before data collection, list plausible situational, participant and material variables. Hold them constant, randomise them, match them or measure them, and standardise instructions and scoring. A pilot can reveal unclear items or ceiling and floor effects.

Controls create a fair comparison by holding an influence constant, distributing it unpredictably or providing a baseline against which the experimental condition can be interpreted.

Practice can improve later performance; fatigue or boredom can reduce it; carry-over means one condition changes response to the next. Counterbalancing estimates and distributes order effects, while independent groups avoids repeated exposure.

Situational variables arise from the environment or procedure; participant variables are stable or pre-existing differences between the people studied. Either can obscure an IV's effect on the DV.

If one memory group has previously played the computer task, prior practice can raise its scores. If one condition is tested in a noisy room, setting can lower its scores. Both become confounds when aligned systematically with the IV.

Research quality has distinct dimensions. Objectivity concerns freedom from researcher judgement; reliability concerns consistency; validity concerns whether the evidence supports the intended interpretation.

Experimenter effects occur when a researcher's expectations or behaviour alter procedures, scoring or participant responses. Demand characteristics occur when participants infer the aim and change their behaviour.

List A analysis matches the summary to the data rather than reporting every available statistic. Quantitative data show amounts; qualitative data show meanings and are analysed through systematic coding and themes.

Inferential statistics ask whether an observed pattern is unlikely under the null hypothesis. Choose a test from the research question, design and lowest level of measurement: nominal categories, ordinal ranks, or interval/ratio equal units.

For Wilcoxon, calculate paired differences, remove zero differences, rank absolute differences with tied ranks averaged, restore signs, total positive and negative ranks, and use the smaller total as $T$. Select the critical value using $N$, the chosen $p$ level and one- or two-tailed prediction.

At $p\le .05$, a result this extreme would occur by chance no more than 5% of the time if the null were true. A stricter $p$ reduces Type I risk - rejecting a true null - but increases Type II risk - failing to reject a false null. Sense-check impossible values, coding errors, outliers, sample size and whether the direction matches the hypothesis before concluding.

A case study investigates one unusual person or a small group in depth, often combining interviews, observations, memory tasks, records and brain imaging over time. Naturally occurring brain damage provides evidence that could not ethically be created experimentally.

After surgery involving the medial temporal region to treat epilepsy, Henry Molaison (HM) retained immediate memory and older memories but could not form normal new declarative long-term memories. Yet his mirror-drawing skill improved without conscious recollection, separating STM from LTM and procedural from declarative memory.

Memory impairment can limit continuing informed consent, and recognisable cases challenge anonymity. Researchers need capacity-sensitive consent, assent and proxy procedures where justified, privacy, minimal burden, withdrawal safeguards and sensitive handling of distressing information.

Bartlett tested how cultural schemas shape recall of an unfamiliar Native American story. Cambridge participants read War of the Ghosts and reproduced it after delays, either repeatedly themselves or through a chain of different people.

Across reproductions, accounts became shorter, more coherent and more conventional to the participants. Unfamiliar details were omitted or transformed: for example, canoes became boats and supernatural material faded. The pattern supports reconstructive memory because recall moved systematically toward familiar schemas.

Using the same unusual story helped reveal schema-driven changes, and the findings apply to eyewitness accounts and transmission of stories. However, the small culturally narrow sample, inconsistent recall intervals and limited procedural detail reduce generalisability and replicability; an unfamiliar story is also unlike many everyday memories.

Schmolck et al. examined whether the extent of temporal-lobe damage predicted loss of semantic knowledge. Six memory-impaired patients - including HM and people with damage extending into lateral temporal cortex - were compared with eight controls.

Participants completed nine tests using common objects and drawings, including naming, pointing, category judgements and verbal definitions. Responses were scored quantitatively; fourteen raters checked definition quality in one task, reducing subjective coding.

Patients whose damage was limited mainly to the hippocampal formation performed normally on semantic tests, while extensive lateral temporal damage was associated with severe, consistent semantic deficits. HM's modest language weakness was judged more consistent with interrupted education than the lesion pattern.

Multiple converging tasks and inter-rater checks strengthen reliability and the link between lesion extent and semantic performance. Yet six unique patients cannot represent all brain damage; lesion groups differed in more than one respect, and picture categorisation is an artificial sample of everyday semantic use.

Darling, Della Sala and Logie tested whether visual appearance and spatial location rely on separable parts of visuo-spatial working memory. Seventy-two adults from a volunteer panel were randomly allocated across combinations of memory and interference tasks.

Participants remembered either the visual appearance of a target on a black-and-white display or its spatial location. During retention they completed selective secondary tasks intended to interfere with appearance or location processing; response latency and accuracy indexed memory performance.

Interference showed an experimental double dissociation: appearance interference selectively disrupted appearance memory, while spatial interference selectively disrupted location memory. This supports separate visual and spatial subsystems rather than one undifferentiated sketchpad.

Random allocation, standardised displays and selective dual-task interference support internal validity and replication. However, a volunteer sample and simplified screen task limit population and ecological validity, and slower responses may reflect task coordination as well as memory storage.

Sacchi, Agnoli and Loftus tested whether altered photographs change memory for public events. Italian participants saw original or digitally doctored photographs of the 1989 Tiananmen Square protest and a 2003 anti-war protest in Rome, then answered questions about the events.

The Beijing image was altered to show a larger crowd; the peaceful Rome scene was altered to show confrontation. Photograph combinations were randomised and counterbalanced before participants completed event-memory and attitude questions.

The enlarged Beijing image led to estimates of more participants. The altered Rome image produced memories of greater violence, confrontation, damage and injuries, more negative ratings, and less willingness to join a future protest. Both younger and older adults were affected.

Real public events and media-like images give the misinformation effect practical relevance, while controlled versions support causal comparison. However, the student-heavy, female-skewed sample limits population validity; questionnaires invite social desirability, and some responses may reflect changed beliefs or source judgements rather than a detailed episodic false memory.

The practical is an ethical repeated-measures laboratory experiment that collects quantitative cognitive data and tests a difference with Wilcoxon signed ranks. Every decision must connect the hypothesis, manipulation, measurement and conclusion.

Suitable designs include a dual-task test of phonological or visuospatial working memory, or recall of acoustically similar versus dissimilar words. Use equivalent word lists and a counterbalanced order so practice, fatigue and material difficulty do not become confounds.