11.1.141PS0 · Section 11: Research methods — How do you carry out psychological research?

Correlation

Key learnings

•
Correlation: relationship, no manipulation
A correlation measures the relationship between two co-variables — whether, and how, they change together.
- The two measured things are co-variables, not an IV and a DV, because the researcher manipulates neither.
- Results are plotted on a scatter diagram, whose points show direction and strength.
- Because nothing is manipulated or controlled, a correlation shows co-variables are related, never that one causes the other.
•
Direction and strength from the scatter
Correlations are described by direction and strength, both read from the scatter diagram — strength from how tightly the points cluster, not as a number.

Direction, from the slope (left to right):
- Positive — both rise together (more revision, higher score); slopes up.
- Negative — one rises as the other falls (more screen time, less sleep); slopes down.
- None — no consistent pattern.
Strength: the closer to a line, the stronger.
•
Spots a relationship, cannot explain it
Strengths and weaknesses flow from one fact: a correlation measures its co-variables, it does not manipulate them.
- Strengths: shows whether a relationship exists and its direction/strength; can study links unethical or impractical to manipulate; is a useful starting point; and can use existing data.
- Weaknesses: cannot show cause (only co-variation); a third variable may explain the link; it detects only straight-line relationships.

Common mistakes

Drawn from real examiner reports.

Define it, don't just name the graph

Define it precisely, or lose the marks.

"Two things" is too vague — it does not say a relationship between two co-variables is measured.
Naming the tool ("scatter diagram") or a type ("positive/negative") is not a definition — the graph is only how a correlation is shown.

The definition: a correlation measures the relationship between two co-variables.

That "two things" is too vague and that mentioning scattergraphs or positive/negative correlations is not a definition are recorded in June 2023 Paper 2 Q4a.

Correlation is not causation

The biggest error is claiming a correlation shows cause and effect. It never does.

It shows only that two co-variables change together, not that one produces the change.
Nothing is controlled, so a third variable could cause both, or the link runs the other way.

Only an experiment that manipulates an IV can show one variable causes a change.

Misreading the scatter, unlabelled axes

Reading the scatter diagram is a documented weakness.

Direction misread. Answering "positive" where points slope down (answer: "negative"). Read left to right: up = positive, down = negative, flat = none.
Axes left unlabelled. Both axes need the co-variable labels, with every point plotted; an unlabelled axis loses marks even when the shape is right.

Naming "positive" where "negative" was correct is recorded in June 2023 Paper 2 Q4b; failing to plot points and label the y-axis on a scattergraph is recorded in June 2019 Paper 2 Q1d(ii).

Correlation vs experiment

Do not call any study with two measured variables an experiment.

Correlation: measures two co-variables, manipulating nothing; shows only a relationship.
Experiment: manipulates an IV and measures a DV; can test cause.

The decider is manipulation: no IV manipulated means a correlation, which is why it cannot show cause.

Co-variables, not IV and DV

In a correlation the two measured variables are co-variables, not an IV and a DV.

The label matters: an IV and a DV belong to an experiment, where the IV is manipulated. In a correlation nothing is manipulated — both are simply measured — so calling them IV and DV wrongly implies manipulation and control that are not there.

A strong correlation proves cause

A strong correlation looks convincing, so students think it proves one co-variable causes the other. But however strong, it only shows the co-variables change together. Only an experiment that manipulates an IV and controls variables shows cause — a correlation controls nothing, so a third variable could drive both, and it can only suggest a link to test.

Exam tips

Name the scenario's own co-variables
Application (AO2) means using the scenario's specific co-variables — e.g. "hours slept and reaction-time score", not "the two variables" or the researcher's name. Tie each evaluation point to those co-variables; losing AO2 also forfeits AO3, since AO3 needs AO2 present.
The 12-mark Evaluate essay
The 12-mark Evaluate essay has three AOs. AO1 (weak spot): two co-variables, neither manipulated; give direction and strength. AO2: use the study's two co-variables. AO3: link each point to the method (no cause; third variable), balance, conclude.
Reading and drawing a scatter diagram
On a scatter diagram, plot one point per pair and label both axes with the co-variables. Read direction by tracing left to right: up = positive, down = negative, no slope = none. To read a value off the graph, go up from the x-axis to the point, then across to the y-axis.

Revision notes

1. Correlation

A correlation is a research method that measures the relationship between two co-variables — whether, and how, they change together. Neither co-variable is manipulated; both are simply measured as they already are.

Key terms

Correlation — a research method that measures the relationship between two co-variables.

Co-variables — the two things measured in a correlation. They are not called an IV and a DV, because neither is manipulated by the researcher.

← All revision topics