OCR A-Level Psychology: Research Methods and the Statistical Tests
OCR A-Level Psychology: Research Methods and the Statistical Tests
Research methods are the backbone of OCR A-Level Psychology, and Component 01 -- worth 90 marks and 30% of the A-Level over a two-hour paper -- is where they are examined most directly. Many students arrive fearing this component because it mixes exam technique with something close to statistics. Yet it is often the highest-scoring part of the course, precisely because it is rule-governed: once you understand why a design is chosen or how to select a statistical test, the marks follow reliably. There is far less to "remember" than in the core studies, and far more to understand.
This guide covers the whole of Component 01: experimental and non-experimental methods; aims, hypotheses and variables; sampling and experimental design; levels of data and descriptive statistics; and -- the part students most want a clear method for -- the five named non-parametric inferential tests, with a decision table that tells you exactly which to use. It finishes with reliability and validity and the BPS ethical framework. Throughout, statistical ideas are explained in plain language and standard symbols so you understand the logic, not just a recipe.
Everything here references the OCR H567 specification descriptively. For the wider map of the qualification, see our complete guide to H567. This article pairs directly with the research methods course and, for timed practice, the exam prep course. The full sequence is on the OCR A-Level Psychology learning path.
Why Methods Matter Across the Whole Course
Although research methods are assessed most heavily in Component 01, the skills are genuinely synoptic. You cannot evaluate Milgram, Loftus and Palmer, or Maguire without understanding experimental design, sampling and ethics -- and questions in Components 02 and 03 routinely ask you to critique a study's method or apply a methodological concept to a novel source. So methods are not a self-contained 30%; they are the toolkit you use in every paper. Keep them warm all year, not just in the run-up to the methods exam.
The content divides into four blocks, which this guide follows in turn:
- Designing research -- experimental and non-experimental methods, hypotheses and variables, sampling and design.
- Handling data -- levels of measurement and descriptive statistics.
- Drawing inferences -- significance and the five inferential tests.
- Quality and ethics -- reliability, validity and the BPS code.
Block 1: Designing Research
Experimental Methods
An experiment manipulates an independent variable (IV) and measures its effect on a dependent variable (DV), while controlling other variables so that any change in the DV can be attributed to the IV. OCR distinguishes three experimental methods by how much control the researcher has and where the study takes place:
| Method | IV manipulated by | Setting | Strength | Weakness |
|---|---|---|---|---|
| Laboratory | Researcher | Controlled | High control; replicable | Lower ecological validity |
| Field | Researcher | Natural | Higher ecological validity | Less control of extraneous variables |
| Quasi | Pre-existing (e.g. age, sex) | Either | Studies participant variables | Cannot randomly allocate; weaker causal claim |
The key distinction with a quasi-experiment is that the IV is a fixed characteristic of the participants (such as gender or a clinical diagnosis) rather than something the researcher creates, so participants cannot be randomly allocated to conditions and causal claims are weaker.
Non-Experimental Methods
Not all research manipulates a variable. OCR requires you to understand three families of non-experimental method:
- Observations. Watching and recording behaviour. Observations vary along several dimensions: structured versus unstructured; naturalistic versus controlled; participant versus non-participant; overt versus covert. Designing an observation means defining behavioural categories (a coding frame) and choosing a sampling method -- time sampling (record what is happening at fixed intervals) or event sampling (record every occurrence of a target behaviour).
- Self-report. Asking people directly, via questionnaires or interviews (structured, semi-structured or unstructured). Questions may be open (free response) or closed (fixed choices), and closed items may use rating scales such as Likert scales or semantic differential scales.
- Correlation. Measuring the relationship between two co-variables without manipulating either. A correlation can be positive, negative, or show no relationship. Crucially, correlation does not establish causation -- a third variable, or reverse causation, may explain an association.
Aims, Hypotheses and Variables
Research begins with an aim (the general purpose) and a research question. From these you derive testable hypotheses:
- The alternative hypothesis (sometimes called the experimental hypothesis in experiments) predicts an effect or relationship.
- The null hypothesis predicts no effect or relationship -- any observed difference is due to chance.
Hypotheses are also either directional (one-tailed) -- predicting the direction of the effect (e.g. "participants recall more words in the quiet condition") -- or non-directional (two-tailed) -- predicting a difference without specifying its direction. You choose a directional hypothesis when previous research gives you good reason to predict which way the effect will go.
To test an IV-DV relationship you must operationalise your variables -- define them in measurable terms. "Aggression" is not measurable; "the number of aggressive acts recorded in a ten-minute observation" is. You must also control extraneous variables (anything other than the IV that could affect the DV), because an uncontrolled extraneous variable that systematically varies with the IV becomes a confounding variable and undermines the study.
Sampling
A study's target population is the whole group the researcher wants to generalise to; the sample is the subset actually studied. OCR requires four sampling techniques:
| Technique | How it works | Strength | Weakness |
|---|---|---|---|
| Random | Every member of the population has an equal chance of selection | Unbiased in principle | Needs a full list of the population; can still be unrepresentative by chance |
| Opportunity | Use whoever is available and willing | Quick, convenient | Often unrepresentative; researcher bias in who is approached |
| Self-selected (volunteer) | Participants respond to an advert or call | Easy access to motivated participants | Volunteers may differ systematically from non-volunteers |
| Snowball | Existing participants recruit others they know | Reaches hard-to-access populations | Sample is networked, so not representative |
The recurring exam skill is to identify a sampling method from a scenario and to evaluate its representativeness and hence the generalisability of the findings.
Experimental Design
Design refers to how participants are allocated to conditions:
- Independent measures -- different participants in each condition. No order effects, but participant variables may differ between groups.
- Repeated measures -- the same participants in every condition. Controls participant variables, but risks order effects (practice or fatigue) and possible demand characteristics.
- Matched participants -- different participants paired on key variables, one of each pair per condition. Reduces participant variables without order effects, but matching is time-consuming and never perfect.
Counterbalancing (for example an ABBA order) is used with repeated measures to spread order effects evenly across conditions, and randomisation of trial order or materials reduces systematic bias.
Block 2: Handling Data
Levels of Measurement
Before you can choose a statistical test you must identify the level of data. OCR uses three levels:
| Level | What it is | Example |
|---|---|---|
| Nominal | Data in named, separate categories (frequencies/counts) | Number of people who helped versus did not help |
| Ordinal | Ranked data, where order matters but intervals are not equal | Positions in a rating scale; ranked preferences |
| Interval | Data on a scale with equal intervals | Reaction time in milliseconds; temperature |
This ordering matters enormously for test choice, so learn it cold: nominal is the "weakest" (categories only), ordinal adds rank order, and interval adds equal spacing. You should also distinguish quantitative from qualitative data, and primary (collected by the researcher) from secondary (pre-existing) data.
Descriptive Statistics
Descriptive statistics summarise a data set. OCR expects two families:
Measures of central tendency (the "typical" value):
- Mode -- the most frequent value. Usable with nominal data; unaffected by extreme scores; but there may be no single mode.
- Median -- the middle value when data are ordered. Robust to extreme scores; suits ordinal data; but ignores the actual values either side.
- Mean -- the arithmetic average. Uses every value and is the most sensitive; but is distorted by extreme scores and needs interval data.
Measures of dispersion (how spread out the data are):
- Range -- highest minus lowest value. Quick to calculate, but based only on the two extreme scores.
- Variance and standard deviation -- measures of the average spread of scores around the mean. The standard deviation is the more informative measure of dispersion because it uses every score; a larger standard deviation means more variability in the data.
You should also be comfortable with ratios, percentages and fractions, frequency tables, and the standard graphs: line graph, pie chart, bar chart, histogram and scatter diagram (used to display correlations). A quick rule: bar charts display discrete categories (bars have gaps), histograms display continuous data (bars touch), and scatter diagrams show the relationship between two co-variables.
Block 3: Drawing Inferences
Significance and Probability
Descriptive statistics summarise your sample; inferential statistics let you decide whether a result is likely to reflect a real effect in the population, or could plausibly have arisen by chance. The logic runs through probability and a significance level.
The conventional significance level in psychology is p ≤ 0.05 -- a result is called statistically significant if the probability of it occurring by chance (if the null hypothesis were true) is 5% or less. A stricter level, p ≤ 0.01, is used where a false positive would be especially costly. To decide significance you compare a calculated (observed) value from your test against a critical value from a statistical table, using the significance level, the number of participants, and whether the hypothesis is one- or two-tailed.
Two errors are always possible:
- A Type 1 error is a false positive -- rejecting the null hypothesis when it is actually true (claiming an effect that is not real). Using too lenient a significance level increases this risk.
- A Type 2 error is a false negative -- retaining the null hypothesis when it is actually false (missing a real effect). Using too strict a significance level increases this risk.
You should also recognise the standard symbols: =, <, ≪ (much less than), ≫ (much greater than), >, ∝ (proportional to) and ~ (approximately). And you should know the criteria for a parametric test -- roughly, interval-level data, a normal distribution, and similar variance between conditions -- because when those criteria are not met, you use one of the five non-parametric tests below. OCR's five named tests are all non-parametric.
The Five Named Inferential Tests
This is the part of the paper students most want a clean method for. OCR names five non-parametric inferential tests, and you must be able to choose the right one for a given study. The choice depends on three questions:
- Are you testing for a difference (between conditions) or an association/correlation (between variables)?
- Is the design related (repeated measures or matched pairs) or independent (independent measures)?
- What is the level of data (nominal, ordinal or interval/ratio)?
Answer those three and the test is determined. Here is the decision table -- the single most useful thing to memorise for Component 01:
| Test | Difference or association? | Design | Level of data |
|---|---|---|---|
| Binomial sign test | Difference | Related | Nominal |
| Wilcoxon signed-ranks | Difference | Related | Ordinal (or higher) |
| Mann-Whitney U | Difference | Independent | Ordinal (or higher) |
| Spearman's rho | Association (correlation) | Paired scores | Ordinal (or higher) |
| Chi-square (χ²) | Association (independence) | Independent | Nominal (frequencies) |
A few practical notes to make the table stick:
- The two "difference + related" tests are separated only by data level. If your related-design data are merely nominal (e.g. each participant just "improved" or "did not improve"), use the sign test; if they are at least ordinal, use Wilcoxon.
- Mann-Whitney U is the independent-design counterpart of Wilcoxon -- both are for differences at ordinal level or higher, but Mann-Whitney is for independent groups and Wilcoxon for related ones. Confusing these two is the single most common test-selection error, so anchor them by the design.
- Spearman's rho is the only test for a correlation -- if the question describes measuring two co-variables and asks about a relationship, it is Spearman's.
- Chi-square is for frequency data in categories and tests for an association (or difference in distribution) between independent groups. If your data are counts in a contingency table, it is chi-square.
A reliable exam routine: read the scenario, decide difference or association, then design, then data level, and read the test straight off the table. Do not try to recognise the test by "feel" -- work the three criteria every time.
A memory hook some students find useful for the difference tests is that the sign test handles the simplest (nominal) related data, Wilcoxon the ranked related data, and Mann-Whitney the ranked independent data -- so as your data get richer and your design shifts from related to independent, you step across the table.
Block 4: Quality and Ethics
Reliability
Reliability is consistency. A reliable measure gives the same result under the same conditions. OCR distinguishes:
- Internal reliability -- consistency within a measure. Checked by split-half (comparing two halves of a test).
- External reliability -- consistency over time or observers. Checked by test-retest (same test, same people, later) and inter-rater/inter-observer reliability (do two observers agree?).
Validity
Validity is accuracy -- whether a study measures what it claims to. It comes in several forms:
| Type | Question it answers |
|---|---|
| Internal validity | Did the IV (not a confound) cause the change in the DV? |
| Face validity | Does the measure look, on the surface, like it measures the right thing? |
| Construct validity | Does it measure the underlying theoretical construct? |
| Concurrent / criterion validity | Does it agree with an established measure or outcome? |
| External validity | Do the findings generalise beyond the study? |
| Population validity | Do they generalise to other people? |
| Ecological validity | Do they generalise to other settings and real life? |
A recurring exam theme is the trade-off between control and ecological validity: a tightly controlled laboratory experiment has strong internal validity but may lack ecological validity, while a field study reverses the balance. You should be able to name the threat and suggest how a design controls it. Threats to validity you must recognise include demand characteristics (participants guessing the aim and changing behaviour), social desirability (answering to look good), and researcher/observer bias and effects.
Ethics: The BPS Framework
Psychological research on humans is governed in the UK by the British Psychological Society (BPS) Code of Ethics and Conduct, built on four principles. You should be able to apply each to a scenario:
| Principle | What it requires |
|---|---|
| Respect | Informed consent, the right to withdraw, and confidentiality. |
| Competence | Working within the limits of one's professional ability. |
| Responsibility | Protecting participants from harm and providing a debrief. |
| Integrity | Honesty; avoiding deception except where justified and managed. |
Many of the classic core studies -- Milgram above all, but also Piliavin and Freud's Little Hans -- are examined partly for their ethical issues, so the BPS framework is a tool you will use across all three papers, not only in methods questions. When you evaluate ethics, name the specific principle at stake and weigh the cost against the study's contribution, rather than simply asserting that a study was "unethical".
Report Writing and How Science Works
Component 01 also covers the conventions of a scientific report -- abstract, introduction, method (design, sample, materials/apparatus, procedure), results, discussion, references and appendices -- along with Harvard referencing and the purpose of peer review. And it asks you to understand how science works as a set of principles: cause-and-effect, falsification, replicability, objectivity, induction and deduction, hypothesis testing, the manipulation of variables, control and standardisation, and quantifiable measurement. These principles feed directly into the "psychology as a science" debate that runs through the whole qualification.
Because you conduct your own small-scale practical activities -- a self-report, an observation, an experiment and a correlation -- the best way to master report writing is to write up your own studies properly. Every section of your own report is a section you can be questioned on.
Bringing It Together
Component 01 rewards a very different kind of preparation from the core studies. Where the studies reward breadth of accurate recall, methods reward understanding a small number of rules deeply: what each method and design is for, how to operationalise and control variables, which sampling technique fits, how to describe data, and -- above all -- how to choose the correct inferential test from the three selection criteria. Master the decision table, keep the reliability/validity/ethics vocabulary precise, and practise applying all of it to unfamiliar scenarios, and Component 01 becomes the most dependable 30% on the whole course.
Work through the research methods course to build the content systematically, then rehearse under timed conditions with the exam prep course. For the mark-scheme logic behind each question style, see our exam technique guide.