Topic I: Psychological skills

Quantitative and qualitative describe the form of data; primary and secondary describe where data come from. These are two separate decisions, so a dataset can be both qualitative and secondary, or quantitative and primary.

Choose from the inference backwards. A CBT researcher might collect primary rating-scale scores for comparable change and primary interviews for how change was experienced; a review could instead analyse secondary trial reports. Record who produced secondary data, when, how and for what purpose.

A sample represents a defined target population. The sampling frame and selection rule determine who could be included and therefore how far findings may generalise.

A complete answer names the frame, eligibility, numbers and selection action. For example, stratifying a school sample by age still requires random selection within each age group; merely choosing equal numbers is not proportional stratification.

Experimental design states how participants enter conditions. Its central trade-off is participant-variable control against order effects and practicality.

Choose by the claim and task. A before-and-after therapy comparison is naturally repeated measures, whereas exposure that permanently changes a person may require independent groups. State how the design works in the scenario, not only its name.

A hypothesis turns an aim into a falsifiable prediction by naming the population and operationalised variables: exactly how each variable will be manipulated, grouped or measured.

Directional experiment: 'Participants hearing traffic noise will correctly recall fewer words from a 20-word list than participants hearing silence.' Null: 'There will be no significant difference in the number correctly recalled...' A correlation instead predicts an association between two co-variables; it does not use an IV.

Questionnaires and interviews gather self-report data. Their structure controls comparability, while question form controls whether the response is numerical, categorical or descriptive.

Use neutral, single-focus wording, exhaustive non-overlapping response options and a time frame. Pilot for ambiguity and accessibility. Standardise instructions, setting and recording; protect privacy so social desirability is less likely. A semi-structured interview can combine a 0–10 rating with an open follow-up asking why that rating was chosen.

An experiment manipulates an independent variable (IV), measures its effect on a dependent variable (DV), and controls plausible alternatives so a causal inference may be tested.

Operationalise the IV as the exact condition difference and the DV as a replicable score, count or measurement. Hold extraneous variables constant or randomise them; if one varies systematically with the IV, it becomes a confounding variable.

Observation records behaviour using declared categories and sampling rules. Design choices affect naturalness, ethics, depth and reliability.

Operationalise each category with observable start/stop rules, pilot it, select times and locations, train two observers, independently code overlapping periods and calculate agreement. Record context without changing categories after seeing a striking case.

A correlation asks whether two measured co-variables vary together. Plot paired scores first: an upward pattern is positive, a downward pattern negative, and no consistent pattern indicates little or no correlation.

Strength describes how closely points follow a monotonic pattern, from values near 0 to values near +1 or −1. The sign gives direction, not strength. A scatter diagram also reveals outliers, restricted range and non-linear patterns that one coefficient can conceal.

Correlations are valuable when manipulating a variable is impossible or unethical, for prediction and for identifying questions for controlled or longitudinal research. Spearman's rank is suitable when paired data are at least ordinal and the prediction concerns association.

Additional methods answer different psychological questions; their labels do not confer automatic credibility. Match the evidence form to the claim.

Control is purposeful management of variables and response effects that could offer a rival explanation for the observed result.

Operationalisation converts a construct into a repeatable manipulation or measure. Randomising condition order is not the same as random allocation; standardisation does not control a badly chosen measure.

Descriptive statistics summarise the sample; they do not decide whether a population effect is statistically significant.

Build a frequency or summary table, check missing/impossible values and sample size, calculate with units and requested precision, then compare like with like. Interpret using actual values: name the group, direction and size of the pattern. Formulae need not be memorised, but the mathematical steps must be competent.

Inferential testing asks how compatible the observed sample result is with the null hypothesis. Choose the test from the research question, design and level of measurement before calculation.

State a one- or two-tailed hypothesis, calculate the observed value, identify N or degrees of freedom, select the matching significance level and critical value, then apply the table's rule. For Wilcoxon a sufficiently small T is significant; for chi-squared a sufficiently large observed value is significant; Spearman tables use coefficient magnitude and direction.

At p≤.05, a result at least this extreme would occur no more than 5% of the time under the null model. Rejecting a true null is Type I; retaining a false null is Type II. A stricter alpha reduces Type I risk but raises Type II risk when other factors are fixed.

Methodological evaluation is a chain: identify a concrete feature, explain its effect on evidence, and revise the exact conclusion.

Standardise instructions, operationalise variables, pilot materials, use inter-rater or test-retest checks, control confounds and sample across the target population. For qualitative research, retain an audit trail, compare sources, check interpretations and acknowledge researcher positioning.

Thematic analysis identifies patterned meaning across qualitative data while preserving enough context to show how the interpretation was produced.

1. Transcribe and repeatedly familiarise yourself with the data. 2. Code relevant units using clear labels. 3. Group related codes into candidate themes. 4. Check themes against coded extracts and the full dataset. 5. define, name and distinguish each theme. 6. Report the pattern with brief evidence and a conclusion tied to the research question.

Use a coding guide, reflexive notes and an audit trail. A second coder can independently code a shared subset and disagreements can refine definitions. Frequencies may convert themes into quantitative summaries, but prevalence alone does not establish meaning or importance.

Published conventions separate question, method, evidence and interpretation so another researcher can judge and replicate the work.

Editors send suitable manuscripts to knowledgeable reviewers, who examine originality, method, analysis, ethics, clarity and fit. Authors revise or answer objections and the editor decides. Replication, corrections and later synthesis continue scrutiny after publication. Preregistration and accessible materials can make selective reporting easier to detect.

When reading a paper, trace every discussion claim back to a stated result and then to the measure and sample that produced it.

Human research begins with competence, a proportionate risk assessment and respect for autonomy, dignity, privacy and scientific value.

Under the UNCRC, children have a right to express views in matters affecting them, with weight appropriate to age and maturity, alongside protection from harm. Seek child assent and appropriate adult consent, design accessible choices and make refusal real; protection should not silence participation.

UK animal procedures are bounded by the Animals (Scientific Procedures) Act 1986 and Home Office regulation. A potentially useful question does not itself justify animal use.

Researchers must justify species, numbers, procedure and expected benefit; work under the relevant licences, trained personnel and veterinary/welfare oversight. A harm–benefit assessment considers severity, duration, cumulative effects and whether reliable knowledge is realistically obtainable.

A key question for society uses concepts, theories or research from Topics A–H, excluding optional Topics F and G, to explain a real decision and its consequences.

1. Define the social question and stakeholders. 2. Select two relevant psychological routes, such as biological and learning accounts. 3. Explain each mechanism accurately. 4. apply it to the question with specific evidence. 5. compare evidence quality, ethics, feasibility and unintended effects. 6. reach a conditional judgment stating when an intervention or policy is justified.

For reducing internet addiction, operant conditioning explains how variable social rewards maintain checking and suggests changing cues and reinforcement; dopamine-reward evidence offers a biological risk mechanism but does not prove inevitability. Evaluate whether support improves functioning without pathologising ordinary use or transferring control to platforms.

Ethical evaluation asks whether a study's knowledge could justify its method, which beings bear risk, and what safeguards make participation or animal use proportionate.

Use a harm–benefit analysis that is study-specific: severity and duration of harm, vulnerability, scientific validity, alternative methods and likely value. Poorly designed research is ethically weak because burdens cannot yield reliable benefit.

A feasible investigation aligns aim, operationalisation, method, sample, controls, ethics, analysis and resources. One weak link can make the final inference unusable.

1. Convert the topic into a focused aim and justified hypothesis. 2. define the target population and recruit through a named frame. 3. choose method/design and operationalise variables. 4. pilot instructions, materials, timing and data capture. 5. control rival explanations without destroying the target behaviour. 6. complete risk/ethics review and consent materials. 7. preselect descriptive and inferential analysis. 8. standardise, record deviations, securely store data and debrief.

Time, access, equipment, researcher competence and participant burden constrain design. A larger representative sample may cost more; tighter laboratory control may reduce realism; repeated measures may save participants but create order effects. Improvements should target the most consequential threat.

Reductionism explains behaviour through smaller components; holism studies the interacting person and context. They are complementary levels of analysis, not simply bad versus good.

A drug study may isolate a receptor pathway to test efficacy, while interviews and longitudinal follow-up reveal adherence, relationships and quality of life. Strong explanation moves between levels and checks whether a component mechanism still predicts outcomes in context.

Psychological explanations select mechanisms at different levels. Comparing them means tracing what each predicts, what evidence could test it and whether their claims can coexist.

For addiction, reinforcement explains maintained behaviour, cognitive accounts explain expectancies, biology explains vulnerability and reward response, and social factors explain availability and norms. A combined model may predict more, but each link still requires evidence.

Psychology is scientific to the extent that claims are operationalised, tested systematically, exposed to falsification and revised through transparent evidence.

Laboratory control can isolate causes but reduce realism; qualitative work can be systematic and transparent without pretending meaning is observer-free. Measurement error, replication failure, publication bias and paradigm disagreement are reasons for stronger scientific practice, not automatic proof that the subject is non-scientific.

Cultural and gender bias can enter through who is studied, whose behaviour defines the norm, how constructs are measured and how findings are applied.

A cross-cultural difference may reflect sampling, language, response style, economic conditions or the construct itself. Researchers should involve communities, examine invariance, disaggregate data and avoid deficit labels.

Nature refers to inherited and biological processes; nurture to physical and social environments and experience. Modern evidence usually tests their interaction rather than allocating behaviour to one side.

Twin and adoption comparisons estimate whether greater genetic relatedness accompanies similarity, but shared environments, gene–environment correlation and unequal experiences complicate inference. Experiments and longitudinal studies can test environmental pathways, while epigenetic processes show experience can affect gene expression without changing DNA sequence.

A vulnerability–stress account predicts that inherited risk may be expressed under particular adversity or protection. People also select and evoke environments partly through heritable traits, so genes and environments are statistically entangled. Cross-fostering, longitudinal and genetically informed designs address different parts of this problem.

Compare concordance, effect size, temporal order and plausible environmental differences before deciding how much each pathway contributes in the studied population.

Development over time is shown when new methods, populations, evidence or ethical standards alter an explanation—not merely because a later study has a newer date.

Early introspection gave way to observable behaviourism; cognitive methods reintroduced mental processes as testable models; neuroscience linked cognition and behaviour to biological measures. Social explanations expanded from personality-only accounts to situational and identity processes. Clinical practice moved from moral/institutional judgments towards standardised diagnosis, evidence-based treatment, rights and cultural formulation.

Progress can mean better measurement, replication, broader samples, less harm and more accurate prediction. It may also create new problems: a diagnostic revision disrupts historical comparison, a brain image can invite biological reductionism, and formal ethics cannot eliminate power differences. Replication can reveal where an influential effect depends on context.

Use dates only to anchor the sequence; the evaluative work is to explain why the later evidence changed confidence, practice or the scope of a theory.

Social control is the deliberate or structural regulation of behaviour towards norms or goals. Psychology can make this control effective, but effectiveness does not settle legitimacy.

Assess the evidence for behaviour change, durability, side effects, fairness and who holds power. Beneficial coordination—such as a safety norm—can coexist with surveillance, stigma or unequal enforcement.

Application translates a supported psychological mechanism into action, then tests whether benefits transfer beyond the original study and whether harms are acceptable.

1. Define the social need and target outcome. 2. identify the mechanism and quality of supporting evidence. 3. co-design an intervention suited to population and culture. 4. pilot feasibility, consent and accessibility. 5. compare outcomes with an appropriate baseline/control. 6. measure adverse effects, equity and durability as well as average benefit. 7. revise, stop or scale with transparent accountability.

Memory research can improve interview questioning; learning principles can support phobia treatment; social-identity evidence can shape prejudice reduction; clinical evidence can guide treatment and anti-stigma practice. Each application changes context and therefore needs fresh outcome evidence.

Socially sensitive research concerns topics or interpretations that can affect identifiable people or groups beyond the immediate study—for example mental health, aggression, prejudice, trauma, crime, intelligence or childhood.

Sensitivity is not a reason to prohibit all research: avoiding a topic can leave harm invisible. Evaluate social value, scientific validity, power, foreseeable misuse and whether safeguards can reduce risk without silencing participants.