Research Methods — AQA GCSE Psychology Revision Notes

What you'll learn

• How to identify independent, dependent and extraneous variables in any study described to you
• The main research methods, and the trade-off every one of them makes
• Sampling: why the method matters more than the number of participants
• The difference between reliability and validity, and why a study can have one without the other
• Ethical principles, and what researchers do when following them conflicts with getting valid data
• How to calculate and interpret the mean, median, mode, range and percentages
• Why correlation can never establish cause, however strong it looks

Key terms and definitions

Hypothesis — a precise, testable statement predicting the outcome of a study.

Independent variable (IV) — the variable the researcher manipulates.

Dependent variable (DV) — the variable the researcher measures.

Extraneous variable — an unwanted variable that could affect the dependent variable.

Standardised procedures — keeping instructions and conditions identical for every participant.

Control group — a group not exposed to the independent variable, used for comparison.

Target population — the whole group a researcher wants their findings to apply to.

Sample — the smaller group actually studied, drawn from the target population.

Random sampling — every member of the target population has an equal chance of selection.

Opportunity sampling — using whoever is available at the time.

Reliability — consistency; a reliable study produces the same results when repeated.

Validity — accuracy; a valid study measures what it claims to measure.

Ecological validity — how well findings apply to real-life settings outside the study.

Quantitative data — data in the form of numbers.

Qualitative data — descriptive, non-numerical data such as interview answers.

Correlation — a relationship in which two variables vary together.

Social desirability bias — answering in a way that reflects well on yourself rather than truthfully.

Pilot study — a small-scale trial run before the main study.

Core concepts

Variables

In an experiment the researcher changes the independent variable and measures the dependent variable. A simple test identifies them: what did the researcher do differently between conditions (IV), and what number did they write down (DV)?

Extraneous variables are everything else that might affect the dependent variable — noise, time of day, participant age, how much sleep people had. They matter because if they are left to vary, the researcher cannot tell whether the independent variable caused the result or something else did. Controlling them is what makes an experiment able to establish cause.

Research methods and their trade-offs

Every method buys one advantage at the cost of another. Learning the pairs is more useful than learning lists.

Laboratory experiment. Variables are controlled and procedures standardised, so cause and effect can be established and the study can be replicated. The cost is artificiality: the setting is unnatural, participants know they are being studied, and the behaviour measured may be a stripped-down version of the real thing.

Field experiment. The independent variable is manipulated in a participant's natural surroundings. Ecological validity rises; control over extraneous variables falls, and consent becomes difficult.

Naturalistic observation. Behaviour is recorded as it genuinely occurs, which can reveal things nobody thought to ask about. But nothing is controlled so cause cannot be established, the observer's expectations can shape what gets recorded, and if participants know they are watched the behaviour is no longer natural. Covert observation solves that and creates a consent problem instead.

Questionnaires and interviews. These reach thoughts, feelings and reasons that no observation can. Questionnaires gather data from many people quickly; interviews go deeper. Both depend on participants being willing and able to report accurately, and social desirability bias, faulty memory and the wording of questions all distort answers.

Correlational studies. These examine relationships where manipulating the variable would be impractical or unethical, and can use existing data cheaply. They cannot establish cause.

Sampling

The target population is who you want to generalise to; the sample is who you actually study.

Random sampling gives every member of the population an equal chance of selection, so the researcher's preferences cannot influence who takes part. It reduces bias but cannot guarantee a representative sample — an unrepresentative one can still arise by chance.

Opportunity sampling uses whoever is available. It is quick and cheap, which is why it is so common, but those present in one place at one time may be quite unlike the wider population.

The size of a sample matters less than its composition. A study of a hundred students from one college tells you about students at that college. Findings can only be generalised to people like those who took part — and with few participants, one atypical individual can shift the whole result.

Reliability and validity

These are routinely confused and the distinction is straightforward.

Reliability is consistency: repeat the study and you get the same result. Standardised procedures produce it, because every participant is treated identically so the study can be repeated exactly.

Validity is accuracy: the study measures what it claims to.

A study can be reliable without being valid. A measure that consistently assesses the wrong thing, or instructions that consistently prompt the same demand characteristic, will produce beautifully consistent results that tell you nothing true. Reliability is necessary but not sufficient.

Replication matters for the same reason. A single result can come from chance, from a peculiarity of that sample, or from an undetected flaw. A finding becomes trustworthy only when others reproduce it independently — which is why full reporting of method is not bureaucracy but the mechanism by which claims get tested.

Ethics

The main principles are informed consent, the right to withdraw, protection from harm, confidentiality, and debriefing.

The genuine difficulty is that these can conflict with obtaining valid data. Telling participants the true aim can change the behaviour being studied, so the most valid data sometimes requires withholding information — yet consent given without knowing what the study involves is not informed. Researchers manage rather than eliminate this tension, using presumptive consent, a clearly stated right to withdraw, and full debriefing afterwards.

A debrief tells participants the real aim, offers them the chance to withdraw their data, and addresses any distress caused. It limits the harm of deception; it does not make deception acceptable on its own.

Descriptive statistics

Mean: add all values and divide by how many there are. It uses every score but is pulled off centre by one extreme value.

Median: the middle value when scores are ordered. With an even number of scores, take the mean of the middle two. It is unaffected by extremes.

Mode: the most frequent value. The only average usable with categories.

Range: highest minus lowest. Simple, but one unusual score distorts it.

Percentage: part ÷ whole × 100. Percentages let groups of different sizes be compared fairly.

For the scores 4, 7, 7, 9, 13: the mean is 40 ÷ 5 = 8, the median is 7, the mode is 7, and the range is 13 − 4 = 9. Note the mean and median differ here because 13 pulls the mean upwards — which is exactly when the median is the better summary.

Correlation is not causation

If revision hours and exam marks rise together, three explanations remain open. Revision may raise marks. Higher-achieving students may choose to revise more, reversing the direction. Or a third factor such as motivation may drive both. No variable was manipulated, so cause cannot be established. Only an experiment can do that.

A positive correlation means both variables rise together; a negative correlation means one rises as the other falls.

Worked examples

Example 1: Identifying variables

Researchers test whether background music affects test scores. One group works in silence, the other with music. Identify the IV and DV. (2 marks)

The independent variable is the presence or absence of background music (1). The dependent variable is the score achieved on the test (1).

Example 2: Suggesting a control

Identify one extraneous variable in the study above and explain how it could be controlled. (3 marks)

Differences in participants' existing ability could affect test scores (1). This could be controlled by randomly allocating participants to the two conditions (1), so that ability is spread evenly across both groups and any difference in scores can be attributed to the music rather than to ability (1).

The third mark is for explaining why the control works. Students routinely stop after naming it.

Example 3: Calculating and choosing an average

Six reaction times in milliseconds are recorded: 210, 220, 225, 230, 240, 615. Which average best summarises them, and why? (3 marks)

The median is the more appropriate average (1). It is found by taking the mean of the two middle values, 225 and 230, giving 227.5 ms (1). The score of 615 is an extreme value that would pull the mean well above every other score, whereas the median is unaffected by it (1).

Example 4: Evaluating a method

A psychologist uses an opportunity sample of college students. Explain one limitation. (2 marks)

Those available in that place at that time may not represent the wider target population (1), so the findings can only be generalised to people similar to that group rather than to people in general (1).

Common mistakes and how to avoid them

Swapping the IV and DV. The IV is what the researcher changed; the DV is what they measured. Read the method, not the title.

Writing an aim instead of a hypothesis. "To investigate whether…" is an aim. A hypothesis predicts: "participants who… will…".

Saying a correlation "proves" something causes something. It never does. Use the words "is associated with" or "is related to".

Confusing reliability and validity. Reliability is consistency; validity is accuracy. A study can be reliable and completely invalid.

Writing "it was unethical" with no detail. Name the principle breached, then say what could mitigate it.

Naming a control without explaining it. The marks are for the mechanism — why that control removes the problem.

Assuming a bigger sample fixes bias. A large sample of the wrong people is still the wrong people.

Treating qualitative data as inferior. It captures meaning and context that numbers cannot. Each type answers a different question, which is why many studies collect both.

Exam technique for Research Methods

Expect this topic everywhere. Research methods questions attach to any study in any topic. A question about memory or obedience may actually be testing sampling or ethics — answer the question asked.

Show your working in calculations. Method marks are available even where the final figure is wrong.

Check the units and the rounding instruction. Give the answer in the form asked for.

Use the two-part structure for evaluation. Name the issue, then state the consequence for the study. "The sample was all students" is half an answer; adding "so the findings may not generalise to older adults" completes it.

Learn one worked example of each average. Being able to produce the mean, median, mode and range from a small data set quickly is reliable marks.

When asked to improve a study, be concrete. Say what you would change and why it fixes the problem, not simply "use more participants".

For any method, know its trade-off. If you can state what a method gains and what it gives up, you can evaluate any study you are shown.

Quick revision summary

• IV is manipulated, DV is measured, extraneous variables are everything else that might interfere
• Laboratory: high control, low ecological validity. Field: the reverse
• Observation gives natural behaviour but no control over variables, so no cause
• Self-report reaches thoughts and reasons, but social desirability and question wording distort it
• Random sampling removes researcher bias; opportunity sampling is quick but possibly unrepresentative
• Generalise only to people like those who took part
• Reliability is consistency; validity is accuracy; a study can be reliable and invalid
• Replication is how findings become trustworthy, which is why method must be reported fully
• Ethics: informed consent, right to withdraw, protection from harm, confidentiality, debriefing
• Debriefing limits the harm of deception without justifying it
• Mean uses every score but is distorted by extremes; median is not; mode works with categories
• Percentage = part ÷ whole × 100, and allows comparison between different-sized groups
• Correlation never establishes cause — a third variable or a reversed direction is always possible
• Pilot studies find problems with instructions and materials while they are still cheap to fix