The four research methods

• Experiment — the only method that can establish cause and effect. Laboratory (controlled setting, high internal validity, low ecological validity), field (natural setting, more realistic, less control), and quasi (the IV is a pre-existing characteristic such as age or a diagnosis, so it is not manipulated — no causal claim).
• Observation — structured or unstructured, naturalistic or controlled, participant or non-participant, overt or covert. Real behaviour, but observer bias and (for covert work) consent problems.
• Self-report — questionnaires and interviews (structured, semi-structured, unstructured). Large samples or rich depth, but social desirability.
• Correlation — measures the relationship between two co-variables: positive, negative or no correlation. It can never show cause and effect.

Watch the trap: a quasi experiment looks like an experiment but the IV was not manipulated, so a difference between the groups cannot be attributed to the IV. Casey et al. (2011), comparing high and low delayers, is a quasi design.

Hypotheses, sampling, designs and variables

• Hypotheses: the null predicts no difference or relationship; the alternative predicts one. One-tailed (directional) if previous research points one way; two-tailed (non-directional) if it does not.
• Sampling (OCR's list): random (every member of the target population has an equal chance), snowball (participants recruit others — useful for hard-to-reach groups, but very unrepresentative), opportunity (whoever is available) and self-selected (volunteers responding to an advert).
• Experimental designs: repeated measures (order effects — counterbalance), independent measures (participant variables — randomly allocate) and matched participants.
• Variables: the IV and DV must be operationalised, and extraneous variables controlled before they become confounding.
• Designing observations: behavioural categories, coding frames, time sampling and event sampling. Designing self-reports: open and closed questions, Likert and semantic differential rating scales.

Which sampling technique?

Levels and types of data

• Nominal — data in categories, counted as frequencies (helped / did not help). No order, so you can only use the mode.
• Ordinal — data that can be ranked, but the gaps between ranks are not equal (a Likert score, a rating of confidence).
• Interval — a scale with equal intervals and a standardised unit (reaction time in ms, number of words recalled).
• Quantitative data is numerical (objective, easy to analyse, but loses meaning); qualitative data is descriptive (rich and valid, but hard to compare and open to interpretation).
• Primary data is collected first-hand for the purpose; secondary data already exists (Gould's re-analysis of the Army IQ data is secondary).

What level of data?

Tap a measurement, then tap its level of data.

Summarising the data

• Central tendency: the mean uses every score but is distorted by outliers; the median is unaffected by extremes; the mode is the only measure available for nominal data.
• Dispersion: the range (highest minus lowest) is crude; the standard deviation uses every score and tells you how far, on average, scores lie from the mean; the variance is the square of the standard deviation.
• Also required: ratio, percentages and fractions; raw data tables; standard and decimal form; significant figures; and estimation.
• Graphs: bar chart (categories, with gaps), histogram (continuous data, no gaps), line graph, pie chart, scatter diagram (correlations only), plus frequency (tally) tables.

Distribution: in a normal distribution the mean, median and mode coincide and the curve is symmetrical. In a skewed distribution they separate, and the median is usually the fairer summary.

Calculate the mean

Calculate the range

Choosing the right test

Ask: (1) difference or relationship? (2) independent or related design? (3) what level of data?

• Chi-square — difference/association, independent, nominal data.
• Mann-Whitney U — difference, independent, ordinal or above.
• Wilcoxon signed ranks — difference, related, ordinal or above.
• Binomial sign test — difference, related, nominal data (each participant simply improves or gets worse: you count the pluses and minuses).
• Spearman's rho — relationship, ordinal or above.
• Parametric tests require interval data, a normal distribution and homogeneity of variance. If those criteria are not met, use a non-parametric test.

Critical values: for Mann-Whitney U, Wilcoxon and the binomial sign test, the observed value must be equal to or LESS than the critical value. For chi-square and Spearman's rho, it must be equal to or GREATER than the critical value. Always check N (or df), the significance level and one- or two-tailed.

Which test?

Degrees of freedom

Percentage from real data

Match the test to the design

Tap a test on the left, then the design it fits.

Reliability, validity and bias

• Reliability = consistency. Internal (split-half — do the two halves of a test agree?), external (test-retest — does it give the same result on another occasion?) and inter-rater (do two observers agree?).
• Validity = measuring what you claim to measure. Internal (was it really the IV?), face (does it look right?), construct, concurrent (does it agree with an established measure?), criterion, population (can it be generalised to people?) and ecological (can it be generalised to real settings?).
• Threats: demand characteristics, social desirability, researcher/observer bias and researcher effects, and sampling bias.
• Ethics — BPS Code: Respect (informed consent, right to withdraw, confidentiality), Competence, Responsibility (protection of participants, debrief) and Integrity (avoiding deception).

Reliability or validity?

Report writing, referencing and peer review

• Sections of a report: abstract (a short summary of the whole study), introduction (previous research funnelling down to the aim and hypotheses), method (design, sample, materials/apparatus, procedure — enough detail to replicate exactly), results (descriptive then inferential), discussion (interpretation, limitations, implications, future research), references and appendices.
• Harvard referencing: e.g. Milgram, S. (1963) Behavioral study of obedience. Journal of Abnormal and Social Psychology, 67, (4), 371–378.
• Peer review — before publication, other experts in the field assess the work's quality, originality and validity. It validates new knowledge, protects the integrity of the discipline and guides funding — but it can be slow, reviewers may be biased against findings that challenge their own work, and anonymity is hard to guarantee.

Report sections

The principles of scientific enquiry

OCR requires you to know how psychology works as a science, and how society uses it.

• Cause and effect — only established by manipulating an IV while controlling everything else.
• Falsification (Popper) — a scientific theory must make predictions that could be shown to be wrong.
• Replicability — the procedure must be repeatable so findings can be checked (which is why the method section must be so detailed).
• Objectivity, control and standardisation, and quantifiable measurements.
• Induction (building a theory up from observations) and deduction (deriving a testable hypothesis from a theory), then hypothesis testing.

The economy and society: OCR wants you to appreciate that psychological research affects real decisions — police interviewing, hospital design, road safety, mental health provision — and that society, not just scientists, decides which questions get funded.

How science works

How Component 01 is assessed

Component 01: Research methods — 90 marks, 2 hours, 30% of the A level.

• Section A — short-answer questions on research methods and techniques.
• Section B — research design and response: you are given a scenario and must design a study, justifying every decision (method, sample, design, variables, controls, ethics).
• Section C — data analysis and interpretation: read a table or graph, calculate, choose a test, interpret a result against a critical value.

Justify, do not just state. "I would use an independent measures design" scores little; "I would use independent measures because the same participant cannot be tested twice without an order effect from having already seen the word list" scores well.

The big ideas to know

That is Component 01 — worth 90 marks and 30% of the A level. Press Finish to see your score.